2 T Fl NCHS =< v RR S LASTS Quality Control and Measurement of Nonsampling Error in the CEL OR TCT TTR TTS U.S. DEPARTMENT OF HEALTH, EDUCATION, AND WELFARE Public Health Service Health Services and Mental Health Administration Vital and Health Statistics-Series 2-No. 54 For sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, D.C. 20402 Price 85 cents domestic postpaid or 60 cents GPO Bookstore DATA EVALUATION AND METHODS RESEARCH Series 2 Number 54 Quality Control and Measurement of Nonsampling Error in the Health Interview Survey A report describing the selection and training of interviewers, interviewer observation program, a reinterview program, measures of interviewer perform- ance, the editing and coding of questionnaires, a response error study, and an interviewer variability study. The report is based on the Bureau of the Census experience with the Health Interview Survey. DHEW Publication No. (HSM) 73-1328 U.S. DEPARTMENT OF HEALTH, EDUCATION, AND WELFARE Public Health Service Health Services and Mental Health Administration National Center for Health Statistics Rockville, Md. March 1973 NATIONAL CENTER FOR HEALTH STATISTICS THEODORE D. WOOLSEY, Director EDWARD B. PERRIN, Ph.D., Deputy Director PHILIP S. LAWRENCE, Sc.D., Associate Director OSWALD K. SAGEN, Ph.D., Assistant Director for Health Statistics Development WALT R. SIMMONS, M.A., Assistant Director for Research and Scientific Development JOHN J. HANLON, M.D., Medical Advisor JAMES E. KELLY, D.D.S., Dental Advisor EDWARD E. MINTY, Executive Officer ALICE HAYWOOD, Information Officer OFFICE OF STATISTICAL METHODS MONROE G. SIRKEN, Ph.D., Director E. EARL BRYANT, M.A., Deputy Director DIVISION OF HEALTH INTERVIEW STATISTICS ELIJAH L. WHITE, Director ROBERT R. FUCHSBERG, Deputy Director RONALD W. WILSON, Chief, Analysis and Reports Branch KENNETH W. HAASE, Chief, Survey Methods Branch COOPERATION OF THE BUREAU OF THE CENSUS Under the legislation establishing the National Health Survey, the Public Health Service is authorized to use, insofar as possible, the services or facilities of other Federal, State, or private agencies. In accordance with specifications established by the National Center for Health Statistics, the Bureau of the Census, under a contractual arrange- ment, participated in planning the survey and collecting the data. Vital and Health Statistics-Series 2-No. 54 DHEW Publication No. (HSM) 73-1328 Library of Congress Catalog Card Number 72-600133 FOREWORD This report is one in a series designed to document the methodology of the Health Interview Survey (HIS) and to investigate the quality of HIS statistics. In previously published reports, the emphasis was on questionnaire development (Series 1, Number 2) and on sample design (Series A, Number 2). Other reports (e.g., Series 2, Numbers 6, 7, 18, 28, and others) present findings of methodological studies that investigated the accuracy of health data collected in household surveys. Specifically, this report deals with the quality-control proce- dures for the data-collection operations of the Survey. It describes procedures for selecting, training, supervising, and observing interviews and measuring interviewer performance, and for editing and coding questionnaires. It also describes the reinterview program. In this program, a staff of field supervisors and senior interviewers reinterview subsamples of households in the Survey. This report presents estimates of nonsampling error based on the reinterview program and estimates of the interviewer contribution to nonsampling variance based on results of a special study designed for this purpose. These statistics have a dual utility. They are useful in evaluating the quality of HIS data and in improving the design of the Survey. Through contractual arrangements with the National Center for Health Statistics, the Bureau of the Census prepares the sample and conducts the field collection process of the Health Interview Survey, and, until 1968, also carried out the data-coding and initial editing procedures. The particular quality-control proce- dures described in this report are essentially applications of methods that are used by the Census Bureau to monitor the field operations of national household surveys. However, the findings presented in this report relate only to the Health Interview Survey. The work for this report was done under a special contract with the Statistical Research Division, Bureau of the Census, in close collaboration with the Office of Statistical Methods and the Division of Health Interview Statistics. Elijah L. White Director Division of Health Interview Statistics Monroe G. Sirken Director Office of Statistical Methods SYMBOLS Data not available------s--semmmemmmmmremceeeeceeeaen Category not applicable-----------seeeeememmnoonneees Quantity zero Quantity more than 0 but less than 0.05----- 0.0 Figure does not meet standards of reliability or precision------------------eeeseeeenee CONTENTS Page Foreword . . . . . . eee eee iii Introduction « + « cv » v0 0 2 +5 “BAF RIF BP BR EHR A ELE ELE ES 1 Purpose of the Health Interview Survey . . . . . ............. 1 Brief Description of the Survey . . . . . .... 00000. 1 Control of the Survey Process . . . . . .. ................. 2 Measurement of NonsamiplIng ELor . . « + + ss «+ 2 vs 5 0 2 ¢ #23 % + 3 Part I. Control of Data Collection and Data Processing . . . . . ....... 3 Introduction . . . . . . LL LL ee eee ee 3 Selection and Training of Interviewers . . . . . . ............. 3 SEleCtion . v « = 2 2s wv ss uw BBG HAE EH EAE ERE REE ES 3 IGAlTIANING = + « + 5 2 2 2 ws 2 2 ot 8 0 ¢ 2 9 68 8 588 8% 4 » 4 4 Continuing Training. . ©... . o.oo eee. 5 Observation of Interviewers . . . . . . . . .. LoL... 6 Introduction .: «snes nine ro pra a ren rE HET EBL us 6 Types of Observations « + : « sc t+ a ww ss wt sw sx ews #2 9 5 3 5 3 6 Supervisory Reinterview Program . . . . . ................. 8 Introduction . . . LLL LL eee 8 SPIE DERISN + » vv 2 2 +8 Hr 2 ® EL BREF LB RE KT ERT EE 8 Content of REIMIEIVIEW « « + x ¢ 2 o # v.40 5 08 5 5 5 8 » w+ 2 5 8 & & 8 FieldProcedur€s « : . s + + s+ a #5 + 2 vs 5 wo % 9» v + % 8 45 o 8 Quality Control of Interviewers’ Work . . . . . ............. 10 Editing and Coding of Completed Questionnaires . . . . . . . . ...... 10 Introduction... LLL LLL ee eee 10 Regional Office Edit. . . . . . . .................... 11 Quality Control of Clerical Coding Operations . . . . . . ........ 12 Central Office Edit . . . . . .. FE MBAR EEN NEEM NEE EE WS 13 Measures of Interviewer Performance . . . . . . .. ............ 13 Part II. Measurement of Nonsampling Error . . . . . . ... ......... 16 Introduction . . . . LL LLL LLL eee ee 16 Response Errors as Determined by a Reinterview Survey . . . . .. .. .. 7 IntroQUEHION. . « « wor vo 0 « % vt v2 9 8 £0 w Bs EE PE EBLE 17 HIS Reinterview Survey Results... . . . 0... ....... 17 Interviewer Variability Study . . . . .................... 24 Introduction. : & « wu o 5 « ® 5% fk 34 ua ha wns Ee ne 24 Design of the Interviewer Variance Study . . . . . . . . .. ...... 24 Method of Analysis . . . . . ...... 0... 26 Result8 « « ¢ + 0 2 ts wo se 8 ws 0 0 sm v0 m 8 69 # +3 8 85.5 8 % 5 28 Summary . . sss a ask hss asa va ss saw aE ne Fe ‘vw. 32 References . . . . . Le ee ee ee ee ee eee 33 Vi CONTENTS—Con. Page Appendix I. Formula for Computing Error Rate . . . . . . ........ 35 Appendix II. Time and Cost Model for HIS Interviewing . . . . . . . . .. 36 General , ou os ww worm mre mm MELEE HE rE EEE REE 36 Models for HIS Assignments . = . ."« « «+ 5 5 + 5 % 65 « 5.2 2 2 ¢ + 5 = + 36 Appendix III. Some Theory of Measurement Errors . . . . . . .. ..... 38 Some Definitions . . . . . LLL Lo 38 The Desited Measure or Trae Value . « = + oc « + « 2 x2 2 5+ 2 2 53 2 5 8 38 The General Conditions That May Affect the Results of a Survey . . . . . . 38 An Estimate from a Survey (or Trial) Taken Under a Set of General Conditions : « + + + ws 2 wo + 5 5 ¢ 6 % sw ow but 8 B 5 4 5 54 39 The Mean Square Error of an Estimate from a Survey (or Trial) . . . . .. 39 Gross and Net Differences . . . . . . o.oo 0000 40 Gross and Net Differences as Evidence of Response Variance and Bias . . oo. LLL LLL ee eee 41 Index of INCONSISLENCY » o + = wo + 0 2 ws ® € + 8 8 $0 8 6 + 5 8 4 3 & 8 2 42 Appendix IV. Least-Squares Solution . . . . ................ 43 Appendix ~~ V. HIS Observation Report (NHS-HIS-406) . . ......... 45 Appendix VI. HIS Reconciliation Form (NHS-HIS-R-IX-T) . . ....... 48 Appendix VII. Appendix VIII. Summary Report of NHS-HIS Reinterview (NHS-HIS-R-401) . 50 Production Guide for NHS (11-102C) . . « vc» 0 ¢ 5 « » » 52 QUALITY CONTROL AND MEASUREMENT OF NONSAMPLING ERROR IN THE HEALTH INTERVIEW SURVEY David A. Koons, Statistical Research Division, U.S. Bureau of the Census INTRODUCTION This report presents a summary of procedures used in the Health Interview Survey (HIS) to control the quality of the data collection and data processing operations. It also provides some results of measurements related to the quality of HIS statistics. PURPOSE OF THE HEALTH INTERVIEW SURVEY The Health Interview Survey is an integral part of the program of the National Center for Health Statistics (NCHS). This program is de- signed to provide continuing statistical measure- ments of the extent of disease, disability, and other health characteristics of the population. The legislation authorizing the HIS, The National Health Survey Act,! contains the fol- lowing provisions: (b) It is, therefore, the purpose of this Act to provide (1) for a continuing survey and special studies to secure on a non-compulsory basis accurate and current statistical infor- mation on the amount, distribution, and effects of illness and disability in the United States and the services received for or because of such conditions: and (2) for studying methods and survey techniques for securing such statistical informa- tion, with a view toward their continuing improvement. This provision for methodological research has strongly influenced the NCHS program. Since the HIS began, emphasis has been placed on improving statistical output rather than on continuity and comparability of estimates. Changes to improve the methods and procedures used in the survey have been made since it began in 1957. BRIEF DESCRIPTION OF THE SURVEY A description of the HIS is necessary to understand the quality control procedures used during the collection and processing of data. The Health Interview Survey uses a question- naire to obtain information on injuries, acute illnesses, chronic conditions, impairments, utili- zation of medical services, and other health topics, in addition to information about per- sonal and demographic characteristics. The findings from the survey are tabulated for the Nation as a whole and published by NCHS. Separate reports are issued which cover one or more of the specific topics. The population covered by the sample for the Health Interview Survey is the civilian, noninsti- tutional population of the United States living at the time of the interview. Persons in long-stay hospitals, nursing and convalescent homes, and so forth are excluded from the universe to be sampled. The sampling plan of the survey follows a multistage probability design which permits a continuous sampling of the civilian population of the United States. The first stage of this design consists of drawing a sample of 357 from about 1,900 U.S. geographic divisions called primary sampling units (PSU). A PSU is a county, a group of contiguous counties, or a standard metropolitan statistical area (SMSA). Within PSU’s, ultimate stage units called segments, selected from clusters of 18 neigh- boring households or addresses, are defined so that each one contains an average of six house- holds. (In July 1968, the average segment size changed from nine to six households.) Two general types of segments are used: (1) area segments, which are defined geographically, and (2) other segments, which are defined from a list of addresses from the 1960 Decennial Census and a current Survey of Construction. Prior to interviewing in area segments, inter- viewers make a list of the addresses of all households or dwelling units in the selected segments. Wherever possible, the visit of the interviewer is preceded by a letter from the Director of the U.S. Bureau of the Census announcing that an interviewer may be expected to visit and setting forth the general purposes of the survey. The confidential treatment that will be accorded any information given is emphasized. As a general rule any adult member, 19 years of age and over, of a family may be interviewed concerning the characteristics of all the members of the family. Persons in the household who are not related to the head of the household are expected to answer all questions about them- selves. Exceptions are made for persons who are not competent to answer for themselves. Persons aged 17-18 may respond for themselves, while persons under 17 must be responded for by an adult. The sample is evenly distributed throughout the year, so that interviews are conducted in approximately 800 households each week. Since household members interviewed each week are a random sample of the population, samples for successive weeks can be combined into larger samples. Thus the design permits both con- tinuous measurement of characteristics of high incidence or prevalence in the population and, through the larger consolidated samples, more detailed analysis of less common characteristics and smaller categories. This continuous collec- tion of information has administrative, opera- tional, and technical advantages since it permits field work to be handled by an experienced, stable staff. In addition, this design eliminates biases due to the seasonal nature of certain conditions or the occurrences of short-run epidemics. Approximately 100 interviewers, about half of whom work each week, are used in the HIS. Each interviewer is assigned an average of three segments (about 18 households) as a week’s work. (As of July 1968, an average interviewer assignment changed from two nine-household segments to three six-household segments.) The interviewers, as well as the entire field staff for the HIS, are employees of the U.S. Bureau of the Census. Specifications for the survey are established by the NCHS. In accord- ance with these specifications, the U.S. Bureau of the Census selects the sample, conducts the field interviewing as an agent of the Center, and checks questionnaire entries. Data preparation, consisting of the preliminary editing and the coding of questionnaires, is carried out by the NCHS. Further editing and preparation of tabu- lations is done by NCHS using electronic computers. The Bureau of the Census has 12 regional offices located in 12 major cities where super- visors of the HIS are stationed. Each supervisor spends a great deal of time visiting the approxi- mately 30 PSU’s in his region in which the interviewing is carried out. Since there are three to four PSU’s per interviewer, many of the interviewers are also required to do a consider- able amount of travel. CONTROL OF THE SURVEY PROCESS The quality control program for the HIS has two purposes: to minimize errors in the survey results and to provide data to evaluate the extent of bias caused by interviewers and re- spondents. Nonsampling errors can occur at any stage of a survey. They may result from the improper statement of the objectives, from faulty con- cepts, or from improper definition of the popu- lation to be studied. They may arise during the sample selection, during the conduct of an interview, or during the processing of the data, e.g., coding, editing, or tabulating. Both the field and office quality control programs of the HIS strive to minimize these errors and to maintain the quality of the interviewing and of the editing, coding, and other data-processing operations. The quality control activities in the field are process controls rather than product controls. That is, very little work in the field is done over again because it does not meet quality control standards. To control errors contributed by interviewers, the program seeks to identify interviewers whose work is defective in partic- ular areas so that remedial action can be taken to improve future work. Such remedial action generally takes the form of retraining the inter- viewers on those aspects of the survey in which their performance was poor. Sometimes, how- ever, interviewers cannot be helped by retraining and must be replaced. Some interviewers resign when they discover through the field quality control program that their performance is not up to par. The turnover, including interviewers whose performance is satisfactory but who resign for personal and other reasons, is about 15 to 25 persons per year among the approxi- mately 100 interviewers assigned to the HIS. However, about 40 percent of the interviewers have been with the survey for at least 5 years. MEASUREMENT OF NONSAMPLING ERROR Nonsampling errors that arise during the interview may have as their source the respond- ent, the interviewer, or the questionnaire. They may result from such causes as respondent memory lapse, the misunderstanding of a question, improperly omitted questions, or incomplete answers. PART I INTRODUCTION Quality control is commonly achieved by measuring performance and setting standards. Illustrations of performance measures are nonin- terview rates, item-response rates, editing-failure rates, and error rates in clerical and card- punching work. Such measures are frequently the basis for applying formal quality controls in the conduct of surveys. These quality controls are specified in terms of minimum performance standards that maintain the quality of work in various operations and thus contribute to the accuracy of survey results. This part of the report describes the controls imposed on the collection and processing of HIS data. Quality control measures are applied at five different stages in the HIS: (1) interviewer selection, (2) training of interviewers, (3) obser- Two of the many attempts that have been made to measure the nonsampling error in the HIS are discussed in this report. One attempt has been the systematic super- visory reinterview which consists of reinterviews conducted by the field supervisory staff and senior interviewers at a subsample of households included in the survey. The results of the reinterview survey are compared on a case-by-- case basis with the results of the original survey. Data from these comparisons are presented as net and gross differences. Net differences are differences between the statistics produced from regular HIS interviews and the statistics pro- duced from the reinterviews. Gross differences are disagreements in individual classifications made by the interviewers and the reinterviewers. A more detailed discussion of these measures is included in this report in the section part II Response Errors as Determined by a Reinterview Survey. Another approach has been an interviewer variability study in which interviewer assign- ments were randomized to obtain estimates of between-interviewer variance. (See Interviewer Variability Study in this report for a more detailed discussion.) CONTROL OF DATA COLLECTION AND DATA PROCESSING vation of interviewers, (4) supervisory reinter- view, and (5) editing of the completed questionnaires. In the HIS, about 30 percent of the total field budget goes into quality control. Table 1 pro- vides a distribution of the costs incurred by activity for calendar year 1968. SELECTION AND TRAINING OF INTERVIEWERS Selection HIS interviewers are selected with great care. Because of the potentially delicate nature of an HIS interview, candidates must have not only the necessary qualifications for handling the interview questionnaire but also unusual tact and sensitivity. Table 1. Percent distribution of field costs by detail expense item: Health Interview Survey, 1968 Detail expense item Total Salaries | Travel | Other! TOMB vv vv vv vt ww me ys me ws kn he Rk md WEEE AE 100.0 47.6 27.7 24.7 INIRIVIEWINID! : © ¢ ¢ © 5 % 5 i SIE B 5 5 6 5 B &§ $8 $8 1 0 55 em 3 me ww mv wm wk 63.3 29.5 20.8 13.3 Initial training Reinterview? Office work? All other Groupand home training . « « = +5 +5 5 4 x 2 » 8 & sw 5 5 0» Observation . . . . . ou i i ee ee ee eee ee ee ees EEE EE mE EEE Es 4.1 2.0 0.8 1.2 YE BW ME EME WW kN 5.3 2.3 1.7 1.3 PRs A BENE NEE EWEN 8.3 3.3 27 23 IEA REBE Ba Ee W EH 4.4 1.8 1.2 1.4 var mma mw BREWS 12.1 7.5 (*) 4.6 Ce eee ee eee 2.2 1.2 0.3 0.6 ' Includes overhead. 2 Includes a check for completeness of coverage. 3 Includes preparation of reinterview assignments. #4 Less than 0.05 percent. Employment as HIS interviewers is limited to women. The typical respondent, a housewife, is generally thought to be more willing to reveal complete health information to another woman than to a man. In addition, employment of women with formal nursing or medical training as HIS interviewers is discouraged because inter- viewers with such training may tend to diagnose or interpret rather than merely record the information obtained from respondents during the interview. Interviewers must pass a test that measures reading comprehension, arithmetic ability, and map-reading ability. An elaborate program of training and observation early in an interviewer’s career also appears to aid in the selection process. The very low level of refusal, about 1 percent of households contacted, and of complaints received seems to show that the interviewers who are finally selected appear to be doing a good job of gaining public cooperation. Initial Training The initial training consists of five separate stages: preclassroom training, classroom training, postclassroom training, on-the-job training, and editing of questionnaires by the supervisor. Preclassroom training. —Preclassroom training is designed to familiarize the new interviewers with both the purpose, scope, and general uses of the HIS and the interviewing materials and the interviewing job. The usual method of presenting such training is through self-study materials. The interviewer is given a self-study package to complete before reporting for classroom training. The contents include administrative materials, a copy of the HIS questionnaire, an interviewer’s manual with instructions to read certain sections, and copies of the letters that the respondents receive. Occasionally, in order to become better acquainted with the survey, the new interviewer also spends 1 day observing an experienced interviewer. Classroom training and practice interview- ing.—Classroom training consists of 5 days of instruction, which covers the interviewer’s manual, the questionnaire and related forms, and interviewing techniques. This training is usually conducted in one of the 12 Census Regional Offices, permitting the trainee to become acquainted with some of the regional office staff members and with general office procedures. The classroom portion of the initial training combines formal classroom training with mock interviewing. Mock or hypothetical interviews are created from situations the interviewer may face. The formal classroom training is primarily for teaching survey concepts. Several training techniques besides mock interviewing are employed in the classroom. They include lectures by the trainer, reading portions of the interviewer’s manual, answering questions, participating in group discussions, and completing written exercises. Practice field interviewing gives the trainee a chance to apply her knowledge of the survey materials to actual interviewing situations. This serves to familiarize her as early as possible with the work she will be doing as well as to stimulate her learning of the concepts and techniques by using them under actual conditions. An observer accompanies the interviewer during practice interviewing assignments. He coaches her on how to handle difficult situations and explains interviewing techniques. During the interview, the observer does not interrupt the trainee unless she becomes very confused. It may occasionally be necessary for the observer to conduct the first interview so the interviewer can get an idea of how the interview should be conducted. The observer uses an observation report to record all pertinent details of each interview, including any errors the interviewer makes. After they leave the household, the observer discusses with the interviewer the points that he has marked on his report and gives her some hints on how to improve her interviewing techniques or to solve problems that arose during the interview. He encourages her to look up the solutions to problems in the interviewer’s manual. In addition to evaluating the interviewer’s technical understanding of the rules and defini- tions that apply to subject matter, the observer checks the interviewer’s performance in the following specified areas: 1. Introduction at the doorstep 2. Use of identification card 3. Explanation of survey 4. Getting settled in the household 5. Interviewer’s ability to maintain a busi- nesslike but friendly attitude with the household members 6. Ability to handle unusual or difficult situations 7. Adeptness with forms (i.e., following skip patterns, probing where answer is incom- plete, asking questions as worded, record- ing answers as instructed) 8. Dress and posture Postclassroom training. —Postclassroom train- ing is designed to familiarize the interviewer with rules and procedures that, although impor- tant, are not as frequently used as those covered during the classroom training. The first postclassroom assignment is com- pleted after classroom training and before the first interviewing assignment. It consists of reviewing classroom topics, completing a lesson that describes the persons to be included in the survey along with a description of housing and sample units, and reading a discussion of admin- istrative forms the interviewer will use. The second postclassroom training assignment is completed at home by the interviewer between her first and second interviewing assign- ments. This assignment is designed to help the interviewer understand sample unit coverage in area segments, the use and background of the address lists taken from the 1960 Decennial Census, and the procedures to be followed at special dwelling places, e.g., motels or convents. The third postclassroom training assignment is completed at home by the interviewer imme- diately before her first listing assignment. A list of all housing units in each area segment must be compiled before any addresses can be selected for interviewing in these segments. The inter- viewer travels around the segment and records the addresses or other description of all places where people live or might live within the segment. On-the-job training.—On-the-job training is conducted by the supervisor-trainer during the interviewer’s first two interviewing assignments and her first listing assignment. This type of training is usually referred to as initial observa- tion and is discussed more fully in the section “Observation of Interviewers.” Edit of questionnaires.—All work of new interviewers is edited by the field supervisor. This includes questionnaires for about 70 house- holds from about four assignments conducted over an 8-week period. A complete check is made of the questionnaire, errors are identified and tallied, and the number and description of the errors is given to the interviewer. Continuing Training Several different kinds of continuing training are used. Group training. —Group training is used for experienced interviewers and normally takes place twice a year. Interviewers are brought together in the 12 Regional Office cities. This provides an opportunity for the widely scattered interviewers to meet one another, exchange views, and receive formal training. There is one session in December or January of each year to learn the new questionnaire for the next cal- endar year, and a midyear refresher session usually held in June or July. Home study and exercises.—There is a formal home-study program for which the interviewers are paid. Home-study assignments, which gener- ally take an average of 3 hours to complete, are made four times a year. These assignments cover various aspects of the interviewer’s job, such as the order in which the questions should be asked, when a question is necessary, etc. They also emphasize the importance of the survey and of the interviewer’s vital role in its operation. Feedback of errors.—Supervisors in the regional offices edit 1 week’s assignment per interviewer per quarter. Additional editing is done as needed, i.e., when previous editing results, observation, or reinterview indicate any consistent type or pattern of omissions or inconsistencies. On the average, about one-sixth of the completed questionnaires are edited each quarter in the regional offices. Certain categories of errors are identified and tallied, and the errors are recorded on forms that are forwarded to the interviewer immediately upon completion of the editing. For example, the interviewer may be required to give an explanation of each error or to make a written reference to the part of the interviewer’s manual that describes the correct procedure. A second edit is carried out at an early stage of the data-processing operation in Wash- ington, D.C. Here the data from all question- naires are examined. This edit takes place, however, weeks or even months after the ques- tionnaires have been filled out by the inter- viewers. Although the immediate feedback that is provided by the field edit is lost, the degree of uniformity in the detection of errors made by interviewers becomes much greater in the central office edit. This procedure therefore provides a better basis for the numerical error scores that are an important part of each interviewer’s performance record. Other training.—Informal training takes place every time supervisors and interviewers get together in connection with the quality control programs discussed in the next two sections. In addition to formal training and the infor- mal meetings with supervisors, interviewers are encouraged to use referral sheets for describing problems. They can mail these sheets directly to the regional offices where solutions for their problems can be quickly determined and mailed ‘back to them. OBSERVATION OF INTERVIEWERS Introduction An important part of the quality control program for the HIS is the observation of interviewers. Either the HIS Program Supervisor, Alternate Supervisor, or Senior Interviewer observes in each regional office. An interviewer is observed in a group of households in her assignment. The observation program contrib- utes to on-the-job training as well as evaluation of the interviewer's performance. The main focus of the observation is to see how the interviewers conduct themselves in obtaining information in their assigned households. The type of controls possible through the observa- tion program depends a great deal on the ability of the supervisor to detect inadequacies and correct them. An HIS observation report (appen- dix V) is used as a guide while observing interviews. On it are recorded the observer’s impression of the interviewer’s performance. Some of the items are entered on a person-by- person basis, and some relate to the day’s work. A copy is placed in the interviewer’s perform- ance file in the regional office, and the original is sent to the U.S. Bureau of the Census. Types of Observations Observations are classed as initial, systematic, and special needs. Although the same procedure is followed for all kinds of observations, they have different purposes. Initial observations.—The purpose of the ini- tial observations is to give new interviewers on-the-job training to correct weaknesses at the beginning of their interviewing career. Initial observations are conducted for each new inter- viewer for 2 full days on her first interviewing assignment, for 1 full day on her second interviewing assignment, and for part of a day on her first listing assignment. The new interviewer begins her day by locating the segment in which she will work. The observer allows the interviewer to find the segment without guidance, unless she becomes completely confused in reading the maps. If this occurs, he assists her in finding the first house- hold. On interviewing assignments, the interviewer introduces herself, and the observer merely observes unless the interviewer asks for help or makes errors. If the interviewer is having serious difficulties, the observer then assists by con- ducting as much of the interview as needed to show the interviewer the proper procedures. At the end of the interview, the observer reviews with the interviewer any general prob- lems that have arisen, such as misapplication of definitions or poor interviewing techniques. He discusses his notes with her immediately after leaving the interviewed household and before going to the next. He also reviews (edits) the questionnaire for completeness. At the end of the observation, he may also review the inter- viewer’s time and mileage records and discuss any general points he noted for special attention. Systematic observations.—Most observations are regularly scheduled visits by the supervisor. These are called systematic observations and are designed to serve three broad purposes: to give on-the-job training in areas where specific weak- nesses are observed; to allow each interviewer a regular opportunity to discuss her work with her supervisor, to make suggestions, and to com- municate in general with the regional office; and to provide information for evaluation of the overall quality of the interviewing in the HIS. Systematic observation assignments are made by the regional offices. One-half of the expe- rienced interviewers are observed each quarter. A systematic observation is made of newly trained interviewers in the quarter following their initial training. No observation is con- ducted when assignments are also scheduled for reinterview. Before conducting a systematic observation, the observer reviews records of the interviewer’s past performance. In addition, he edits recent examples of her listing of households, reviews the office copy of the report of her last observation, and, in general, tries to determine which points should be observed most closely. Special-needs observations.—Some inter- viewers need more contact with their supervisor than is provided by the systematic observations, and the regional offices need the flexibility of being able to give additional training to inter- viewers when it is needed. Special-needs observations are used for this additional training. A special-needs observation is usually made for an interviewer whose work is rejected in reinterview. Rejection in reinterview is based on the number of differences between the original interviewer’s results and the reinterviewer’s results. An interviewer’s work is rejected if the number of differences is in excess of specified limits given in a table of acceptability. (See the next section for further discussion of the reinter- view program.) The special-needs observation is scheduled for the interviewer’s first assignment following the reinterview. A special-needs observation may also be made for an interviewer whose work falls below certain minimum performance standards such as the following: poor production, e.g., too few completed interviews per day or too much travel time; a high noninterview rate; an excessive number of recording errors and omissions on the questionnaire filled out by the interviewer; and poor performance on recent observation. The purpose of the special-needs observation is retraining. If, in the judgment of the super- visor, an interviewer needs retraining after falling below the minimum standard for some aspect of her work, an observation is conducted. Some- times a seemingly poor performance can be explained, for example, a high noninterview rate in the summer, and no retraining is necessary. Except in the case of rejection in reinterview, the decision to conduct a special-needs observa- tion is made entirely in the regional offices. The procedure for the special-needs obser- vation is the same as that for the systematic observation, but with special attention given to the aspect of the interviewer’s work that needs improvement. SUPERVISORY REINTERVIEW PROGRAM Introduction The major purpose of the program of super- visory reinterviews is to control quality. The program provides a process control on the work of individual interviewers so that interviewers with high levels of error can be identified and remedial action taken. The remedial action is intended to improve the quality of the individ- ual interviewer’s work through retraining, obser- vation, and discussion of errors with the inter- viewer, as necessary. The reinterview also serves as a periodic check on interviewers to see that assignments are carried out as instructed. This process also provides assessments of the reliability and accuracy of the HIS because the quality control technique employed in this program requires that the supervisors fill out complete questionnaires that can then be matched with the questionnaires filled out by the interviewers. Since these questionnaires con- stitute, by design at least, a probability sample of all HIS interviews, estimates bearing on the reliability and accuracy of HIS statistics can be made. Sample Design Reinterview assignments are made on the basis of interviewer workload, that is, inter- viewers with larger workloads have more reinter- views in a given year. A reinterview assignment consists of one weekly work assignment, and there is an average of three reinterview assign- ments per year per interviewer. Originally, one reinterview assignment was carried out for each interviewer per quarter. In January 1963, the program was reduced to three assignments per year per interviewer. On July 1, 1965, the selection method was changed to reflect the variability of interviewer workload. An additional reinterview assignment is made for interviewers rejected in the previous quarter. The number of interviewers for which this is done is limited to not more than 10 percent of the total number of interviewers. Reinterview assignments are evenly divided among the weeks of the quarter, and there is only one reinterview assignment in a regional office area in any given week. A reinterview of 12 of the 18 households in a typical interview assignment is conducted with one person reinterviewed in each of the 12 households selected. The reinterview sample is divided into two parts. The sample of households selected for reinterview is subdivided into an 80-percent subsample and a 20-percent subsample. In the 80-percent subsample of households, the super- visor carries out a reconciliation of reinterview results with the results of the original interview. No reconciliation is carried out for persons in households designated for the 20-percent sub- sample. The division of the reinterview sample into an 80-percent subsample and a 20-percent subsample began in January 1959. Before that, reconciliation was carried out for the entire reinterview sample. Content of Reinterview The first part of the reinterview is a coverage check to see if all household members have been properly included in the survey. The second part of the reinterview deals with the reporting of personal and health characteristics. In general, the reinterview covers all questions relating to the reporting of health conditions and their characteristics originally included in the first interview. Supplements, such as hospital insurance, eyeglasses, or hearing aids, are not usually included in the reinterview. Field Procedures At the beginning of each quarter, regional supervisors are told which weeks will have a reinterview assignment. One week before inter- view week, they are told which assignments are to be reinterviewed and given instructions for selecting the subsample of households desig- nated for reinterview and the sample persons within reinterview households. One person per household is randomly selected for the part of the reinterview that covers personal and health characteristics. The reinterviewer is instructed not to look at the original interview results before reinterview. The HIS reconciliation questionnaire (appen- dix VI) containing the transcribed information from the original interview is given to him in a sealed envelope. He does not open this envelope until he completes the reinterview. For the 20-percent subsample of households for which reconciliation is not carried out, the original questionnaires are not transcribed. For these households, the note “Omit Content Rec- onciliation” is placed on the reconciliation questionnaire inside a sealed envelope. Thus the supervisor is not supposed to know in advance the households where he will not do reconcili- ation. The data from the 20-percent group are used to test the extent to which accessibility of original responses to the reinterviewer has apparently affected reinterview results. The reinterview fieldwork includes verifying the original interviewer’s work in the listing of addresses in area segments, checking household composition in the sample households, and reinterviewing one person in each reinterview sample household. The reinterview is scheduled for the week following the original interview and must be completed no later than 2 weeks after the date of the original interview. Since the questions on the HIS schedule refer to specific time periods, such as “last week or the week before,” “a year ago,” and “past 12 months,” the reinterviewer, in asking these questions, must be certain to get information for the same time period used by the original interviewer. In order to do this, the reinterviewer must specify the exact dates of the reference period used in the original interview. The reinterviewer makes a personal visit to each household selected for reinterview. The questions relating to coverage of persons within the household may be asked of any eligible respondent. For the health information for adults, the most acceptable respondent is the person who provided the data in the original interview. If he is not available, however, the sample person may be interviewed. (Before July 1, 1965, the only acceptable respondent was the sample person himself.) Information for children is obtained from parents or an adult responsible for the child’s care. Responses are entered on the reinterview questionnaire and changes are not made after this part of the reinterview is completed. (Before January 1967, supervisors used a special ques- tionnaire containing only the subjects covered in the reinterview. Since then supervisors use a blank HIS questionnaire for recording the rein- terview results.) In 80 percent of the reinterviews, differences in responses from the two interviews are recon- ciled immediately after completion of the re- interview. For personal characteristics, the re- interviewer transcribes the information he has obtained to the reconciliation questionnaire. He then compares these answers with the original responses and reconciles any that are different. Next, he compares the responses to the health questions on the reinterview questionnaire with the reconciliation questionniare, which contains the information from the original interview. If the reinterviewer finds that differences exist, he attempts to determine from the respondent the proper response and any possible reasons for differences. The reconciliation questionnaire provides space for recording reasons given by the respondent for differences between the original interview and the reinterview on the reporting of illnesses and other health conditions and hospitalizations. Table 2 shows the number of persons reinter- viewed in fiscal years 1959 through 1967. Table 2. Number and percent of persons interviewed and rein- terviewed: Health Interview Survey, fiscal years 1959-67 Number of Number of Fiscal year complsted compiecsd Percent interviews | reinterviews (persons) (persons)! 1969 ............ 126,841 3,478 2.7 1960 ..enscarnens 118,068 3,061 26 1B cy sanscmmrnm 112,086 3,206 29 1962 i :ivnvimms in 118,432 2,839 24 1963 ............ 139,055 2,995 22 1964 ............ 129,801 2,391 18 1965 sivas snwers 139,152 2,081 1.5 1968 .iruwssmmses 139,486 2,053 15 1967 uv :nms sums tn 133,916 1,933 1.4 'The decline in the number of completed reinterviews is the result of a cutback in the reinterview sample size. Quality Control of Interviewers’ Work One purpose of the supervisory reinterview program is quality control. Specifically, the program is designed to check on coverage and content errors. Errors in coverage of the popula- tion can occur because of incorrect listing of addresses in sample segments, failure to conduct interviews at the correct addresses, and incorrect application of definitions of housing unit and household member. Content errors are errors in the data obtained by the interviewer concerning personal and health characteristics of members of the sample household. Through the recon- ciliation of original interview and reinterview results, the supervisor tries to obtain the best answers to the HIS questions. A second purpose of the reinterview program, that of obtaining measures of nonsampling errors and biases, is discussed in part II of this report. After a reinterview assignment has been com- pleted, the reinterviewer completes a summary report of the HIS reinterview (appendix VII) showing the number of differences for five categories of the interviewer’s work: listing; household composition; personal characteristics; characteristics of conditions and hospitaliza- tions; and number of conditions, hospitalization, and injuries. Tolerance limits are established for each category. The interviewer's work is required to meet the standards for each category separately. The tolerance limits are listed in a table of acceptability, which shows for each category separately the number of differences that are acceptable for a particular sample size. Only cases where the respondent was the same on both interviews are used in the table of acceptability. The acceptance numbers are set so that a difference rate at a 5-percent level will be accepted 95 percent of the time. An interviewer's work is rejected if the number of differences in any classification is in excess of the numbers given in the table of acceptability. An analysis of reinterview assign- ments was carried out for the period July 1, 1962, through June 30, 1967. During this time, 1,554 original interview assignments were re- interviewed. Rejections were noted in 115 assignments in one or more categories. These 115 assignments were rejected on the following grounds: Category of rejection Number All categories of rejection . ............. 139 LASHING was sms sums ames mens abe swmE sus ¥ 18 Houshold'COMPOSILION «os mm sis mms dwn sms 5 7 Personal characteristics . ................... 31 Characteristics of conditions and hospitalizations . . . 30 Number of conditions and hospitalizations ....... 53 Of the 115 assignments that were rejected, 95 were rejected on one category, 16 were rejected on two categories, and 4 were rejected on three categories. No assignments were rejected on more than three categories. Supervisors in the regional offices initiate retraining, observations, etc. of interviewers whose work is rejected. A report of actions taken is made to headquarters in Washing- ton, D.C. Table 3 presents the actions taken for those interviewers whose assignments were rejected in reinterview by reason for rejection for the period July 1, 1962, through June 30, 1967. EDITING AND CODING OF COMPLETED QUESTIONNAIRES Introduction When completed assignments are received in the regional offices from interviewers, the ques- tionnaires are edited for consistency and com- pleteness. A systematic edit is carried out for a specified sample of assignments. The results of this edit are sent to the interviewers with identification of errors and specific references to sections of the interviewer’s manual to review. Further editing is carried out during coding and processing operations at headquarters. The NCHS assumed responsibility for the coding and data preparation in 1968. A new questionnaire format and new coding procedures were adopted at that time. The coding and quality control procedures described in this report are those used by the Bureau of the Census before 1968. About 800 household questionnaires are re- ceived each week for processing. After the Table 3. Percent distribution of actions taken in cases of interviewer rejection by nature of rejection: Health Interview Survey, July 1, 1962-June 30, 1967 Category of rejection Action taken after rejection oo Household Personal Hearth Listing ot red Tables I. composition | characteristics conditions Percent distribution Total rejections . . . . . . . i.e eee. . 100.0 100.0 100.0 100.0 100.0 Retrained , « : «co vs vo #5 v5 % + 5 5 5 8 5 + 5 s 8 5 5s 835 ® 0 278 14.3 3.2 3.3 3.8 Observed : : : 5s 5: ws 5am sis $i Bins ms Mam smsma 33.3 14.3 64.5 40.0 415 Retrained and observed . . . . . . . ................ 5.6 14.3 129 30.0 15.1 Errorsdiscussed . . . . . . LLL. eee ee 22.2 14.3 9.7 3.3 9.4 Resigned , os ws vs 0s ws vis 69 BF Mi meme ws 53 #4 0.0 0.0 0.0 0.0 1.9 Dismissed ; u 3 5 5s 5 3% 8% $F HE FEW Em A BREE EE a 0.0 0.0 0.0 33 5.7 No action necessary’ . . . . . . ieee eee 0.0 28.6 6.5 10.0 15.1 Action notreported . . . . . LL... ee eee 1.1 14.3 3.2 10.0 7.5 'The supervisor may decide that no actions are necessary if, in his judgment, the differences are due to factors beyond the control of the interviewer. For example, one confused respondent may contribute all the differences because he misunderstood questions, questionnaires are checked in, they are grouped into work units of approximately 25 question- naires each. The questionnaires are assigned in work units to clerks who check the question- naires for completeness, assign codes to the information on the questionnaires, and tran- scribe all of the information on the question- naires to punch card transcription sheets. Ques- tionnaires go- through nonmedical coding and medical coding operations. Nonmedical coding assigns codes to the demographic items and items related to health conditions. Medical coding, which is a more complicated operation, assigns detailed diagnostic codes to the illnesses, injuries, and hospitalizations reported on the questionnaires. Diagnostic codes are assigned, with some modification, according to Inter- national Classification of Diseases (ICD). To control the level of errors in these coding operations, specific quality control procedures are followed. These procedures are discussed in the section on quality control of clerical coding operations. Regional Office Edit Specifications for carrying out a regional office edit are sent to each office at the beginning of a year. These specifications define the minimum editing that must be done. Addi- tional editing is carried out on the basis of need, i.e., if previous edit results, observation results, or interview results show errors such as omis- sions and inconsistencies. The specifications for editing the work of experienced interviewers generally provide for more editing at the beginning of the year when new items are added to the questionnaire. As interviewers become more experienced with new items, the amount of editing is reduced. How- ever, the first four assignments of new inter- viewers are always edited. There are two types of edits performed in the regional offices, diagnostic and nondiagnostic. The diagnostic edit must be done by the HIS supervisor. Errors are assigned for missing or inadequate entries for illnesses, injuries, or hos- pitalizations. The nondiagnostic edit can be done by a qualified clerk and consists of identifying omissions and incorrect entries in identification and control items on the ques- tionnaire. The results of the diagnostic and nondiag- nostic editing are sent to the interviewers and provide some immediate feedback on errors. A copy of the results is retained in the regional 1" office for comparison with the more intensive edit and identification of errors made during the central office coding and processing operations. Quality Control of Clerical Coding Operations There have been a number of changes over the years in the processing of HIS question- naires. From the beginning of the survey in 1957 until November 1965, processing consisted of transcribing and coding information from ques- tionnaires to document-sensing cards from which IBM punchcards were mechanically pre- pared. The assignment of diagnostic codes to illnesses, injuries, and hospitalizations was inde- pendently verified on a 100-percent basis. Two coders independently assigned diagnostic codes on the information in the questionnaire. These codes were compared, and differences were resolved by a supervisor. The coding and tran- scription of nonmedical entries was completely verified by a second coder’s examining the entries on the document-sensing card to see if they had been correctly transcribed by the first coder. In November 1965, a new schedule format was introduced into the survey. Entries on this schedule could be read directly by machine, thus bypassing a large amount of clerical transcrip- tion. In addition some of the codes for non- medical items, such as age, were entered on the schedule by interviewers and required no further coding. However, a substantial amount of editing and coding was still required, particularly for diagnostic entries that had to be medically coded. At the time this new schedule was introduced, it was decided that sample verifica- tion to control the quality of coding would provide a better use of resources than 100- percent verification, particularly for medical coding. Consequently, starting in November 1965, a sample verification plan was introduced into the medical coding operations of the survey. The plan provides for two stages of control in the medical coding operation: a training and quali- fication period during which the coders’ work is independently verified 100 percent, and a post- training period during which the coder’s work is independently verified on a 10-percent sample 12 basis. All errors detected during verification are corrected. During the qualification period the new coder codes to a work sheet. Then the coding is done over again by a qualified coder independently on an HIS schedule. A comparison clerk matches the medical codes entered on the worksheet with the medical codes on the corresponding schedules. Differences in medical codes are reviewed by an adjudicator. The adjudicator assigns an error if, in his judgment, the original coder assigned the wrong code. If, however, the differences in codes are a matter of coder judgment, a decision concerning the proper code is made, but an error is not assigned. In order to qualify for sample verification, a coder must code four consecutive work units out of a maximum of eight with an error rate of 4 percent or less for each work unit. If a coder fails to qualify within the first sequence of eight work units coded, a second sequence of eight for qualification is started. A coder has a maximum of three sequences in which to qualify. If a coder fails to qualify in the third sequence of eight work units, he is not considered for sample verification. Once a coder has qualified for sample verification, his work continues to be verified on a 10-percent sample basis. A record of verification is maintained for each medical coder. When the cumulative number of verified codes reaches 45, a decision is made to deter- mine if the coder’s work is still acceptable. If a coder’s work is rejected three or more times in 10 decisions, he must requalify for sample verification. During the requalification period his work is verified on a 100-percent basis. If the coder fails to requalify, he is no longer considered for sample verification. Additional changes have been made in the format of the questionnaire since the sample verification plan was introduced. However, essentially the same verification procedures con- tinue to be used, i.e., independent verification of medical coding on a sample basis and 100- percent dependent verification of nonmedical coding. Records for the period April 1, 1967, to March 31, 1968, show an average error rate of about 2 percent in assignment of medical codes for experienced coders on sample verification. For the same period, the nonmedical coding error rate was also about 2 percent. Central Office Edit As part of the coding operations described above, a comprehensive review of each question- naire is carried out to identify omissions, inade- quate entries, and inconsistencies. Error codes are entered on the processing record so that the number and identification of errors can be tabulated and summarized for each interviewer. In addition, specific descriptions of errors are provided on a separate document. These descrip- tions identify the number and type of errors for different sections of the questionnaire, for example, person page, condition page, hospitali- zation page, etc. These errors are divided into two main categories: diagnostic errors on condi- tions and hospitalizations, and nondiagnostic errors. The forms used to record the number and types of errors are the same used in the regional office edit. Weekly summaries of diagnostic errors are sent to regional offices, and quarterly summaries of both diagnostic and nondiagnostic errors are sent to the regional offices which in turn notify individual interviewers. The quar- terly summaries also form the basis for com- puting an interviewer error rate, which is one of the measures used to evaluate interviewer per- formance as described in the next section. Additional editing is done on the computer, which performs a series of adequacy and con- sistency edits. Individual records with errors are identified, the original questionnaires are located, and corrections made, as necessary, to the records. MEASURES OF INTERVIEWER PERFORMANCE In the preceding sections of this report the activities for controlling the quality of survey results have been presented. Results from these quality control activities are combined with other data to provide an overall evaluation of interviewer performance. The measurement of interviewer performance in the HIS is a combination of subjective ratings by supervisors and quantitative measures based on an examination of an interviewer’s completed work. Minimum standards of performance on the quantitative measures are set up. A cumulative record of performance for each interviewer is maintained in the regional office. If, at any time, this record indicates that an interviewer’s work has fallen below the minimum standard, correc- tive action is taken. This corrective action may consist of retraining, observation, or, in some cases, replacement of the interviewer. In practice the evaluation of interviewer performance is based on the pattern of performance over time and on different aspects of the interviewing job rather than performance on any single aspect. In the HIS, three quantitative measures of performance are computed on a continuing basis. They are the error score, the noninterview rate, and the production ratio. The error score is computed as follows: (number of errors)/(total conditions + total acci- dents + total hospitalizations). (See appendix I.) The numerator is the number of errors identified during processing. Errors include omitted entries, missed conditions, missed hospitaliza- tions, and diagnostic errors. Missed conditions and missed hospitalizations are those identified in the early or probing section of the interview but not followed up for additional information in the latter section of the interview. Diagnostic errors occur when the interviewer fails to record sufficient information to allow a medical coder to assign diagnostic codes. The noninterview rate is computed as follows: (number of noninterview households)/(number of interviewed households + number of non- interview households). The noninterview house- holds are households eligible to be included in the HIS, but for which no interview was conducted. Included as noninterviews are those the interviewer has reported as “refusals,” “no one at home,” “temporarily absent,” etc. The production ratio is measured as follows: (estimated time based on production stand- ards)/(actual payroll time charged by the inter- viewer). The numerator is estimated from a mathematical equation (appendix II) that takes into account such things as the average time per household, the number of assigned households, and the distance to area of assignment from interviewer’s home. 13 In addition to quantitative measures used to evaluate interviewer performance, the results of supervisory reinterviews and observations are used as much as possible. Also, if specific individual interviewer errors are discovered during the processing operation at the central office, they are noted and forwarded to the regional supervisor. He in turn informs the interviewer of these errors and suggests means of eliminating them. After a probationary period of 6 months, each interviewer receives a report on her per- formance over the past quarter. The report contains both a descriptive rating and a numeri- cal score. The descriptive ratings are “Excel- lent,” “Satisfactory,” “Needs Improvement,” or “Unsatisfactory.” If an interviewer receives a rating of “Needs Improvement” or “Unsatis- factory,” she will receive a warning notice unless, in the judgment of her supervisor, there are extenuating circumstances. In addition to the quantitative measures used in evaluating individual interviewer performance, other measures are used to provide an overall summary of performance. These include number of conditions per person, number of missed conditions and hospitalizations, number of diag- nostic errors, and proportion of reinterview assignments accepted. (See ‘Regional Office Edit” for a discussion of diagnostic errors.) Tables 4 and 5 show the average rates for HIS interviewers over a 4-year period for five of these rates and over a 7%-year period for three of these rates. The variation from quarter to quarter is small; the only apparent trend being in the number of conditions per person, which has steadily increased throughout the survey. Table 6 shows average rates for interviewers Table 4. Average interviewer performance on various measures by survey quarters: Health Interview Survey, January 1962-June 1965 Average Average number of Average number Average Average proportion Number error rate missed conditions pitalizations production of reinterview Survey quarter of mer (percent) per 100 persons per 1,000 persons ratio assignments accepted Rate | N! Rate N Rate N Ratio N Proportion N 1962: Jan.-Mar. ........ 107 46 | 105 10 105 .64 105 1.09 | 100 95 94 Apr.-June ....... 128 58 | 128 25 127 .83 127 91 114 05 05 July-Sept. .. wuss 120 | 10.0 | 119 .30 119 .45 118 89 | 110 95 92 Oct Dec, oiivivesns 117 8.3 | 116 .38 116 .68 115 90 | 103 91 94 1963: Jan Mar, sieves 112 58 | 112 32 112 49 112 99 91 .94 62 Apr.-June ....... 104 59 | 103 23 103 .64 103 97 87 95 66 July-Sept. ....... 99 5.0 98 .09 98 24 98 1.01 91 .85 61 Oct.-Dec. . ....... 104 6.9 | 104 31 104 45 104 97 98 .80 70 1964: Jan-Mar. ........ 98 7.3 98 40 98 42 98 1.06 92 .89 61 Apr.-June ....... 98 6.9 98 33 98 .39 98 1.03 94 95 64 July-Sept. ....... 102 6.8 | 101 .28 101 24.46 101 1.00 | 100 92 48 Oct-Dec, oo: aus +n 101 6.9 | 101 42 100 42 100 1.02 | 106 .89 65 1965: Jan-Mar. ........ 100 7.8 | 100 .16 100 .66 100 1.05 96 .90 68 Apr.-dune ....... 97 6.4 97 A2 97 .58 97 1.03 96 92 61 !N = number of interviewers included in the computations. 2This high rate is due to one interviewer who interviewed 10 persons but had 4 missed hospitalizations. Excluding the inter- viewers work, the measure is .54. Because of the differences from one assignment to another in the population covered, the number of conditions per person is not used as a performance measure for individual interviewers. 14 Table 5. Average interviewer performance on various measures by survey quarters: Health Interview Survey, January 1958- June 1965 Average Average number | Average number of Number | noninterview of conditions diagnostic errors Survey quarter of inter- | rate (percent) per person per 100 conditions viewers Rate N! Rate N Rate N 1958: Jan-Mar. . . . LL. 102 4.8 102 94 102 4.8 102 Apr.-dune LLL eee 103 4.3 102 92 102 4.6 102 July-Sept. . .. LL 104 5.6 104 84 104 5.1 104 Oct-Dec. . . .. ii ee 108 4.6 106 .90 106 4.2 106 1959: Jan-Mar. LLL LLL 87 - - - - - - Apr.-dune LLL LLL ee eee 92 4.4 84 91 84 4.4 84 July-Sept. . . LL. eee 94 7.4 93 .86 92 4.2 91 OcteDee, = . ss 6s vs vm i ws 8 3s Mrs SR eH sb % 0 a 88 4.7 85 .90 84 3.7 84 1960: Jan-Mar. ©... LL eee 85 4.0 82 97 82 2.4 82 APEAIIRE | 4; ws ws 0m mE Be WE HE Ms WERE ES 93 34 92 93 93 2.1 93 LUE EE TE EE EE EE RE EE YY 93 6.2 92 .95 92 2.1 92 OCYrDEG: 5 2 5 2% 3 5 5 5 5.68 i» mem mv mmm mm 98 3.8 98 94 98 1.4 98 1961: JAN AMOE, oc os maw sm rm rs HEE HE EE MEE EE 103 4.1 103 1.00 99 23 99 ADraJung , ..w:sr ws aE Rs Rid R ET Ew 108 4.3 108 97 105 2.7 105 JUlyBept. |... ci iia se ee mem mn 109 5.7 109 .98 109 33 109 Oct-Dec. . . .. . i iii eee 108 4.5 108 1.00 108 3.8 108 1962: JA AMVBE. oc ws ws 5 8 Hs E EE ESE REE EE GREW 107 4.7 105 1.07 105 3.4 105 APY JUNE 4s 55 5 «5 2% 3 5 88 04 hot 3 5 Bk hE Ea 128 4.5 128 1.02 128 5.0 128 July-Sept. . . . Lee eee 120 7.0 119 1.01 119 6.0 119 Oct-Dec. . . . . . i ieee 117 4.9 116 1.04 116 5.9 116 1963: JEN MBE. «5 coms ws wr sw EEE es EEN RE RE we 112 3.6 112 1.09 112 4.2 112 BRrpdUng oc vu wm we Bi Bs BEBE HEE EWN YE Re 104 4.3 103 1.04 103 4.1 103 JulySept. . i: sc scsi sa sms Ts EEE IE BE Ba 99 5.3 99 1.03 99 3.0 99 Oct-Dec. . » : 5 5.5 v5 5.7 5 5 3s Bs so wr mor osm 104 3.5 104 1.06 104 3.6 104 1964: Jan-Mar. LLL 98 3.7 98 1.08 98 3.9 98 Apr-dune LLL LL eee 98 3.9 98 1.06 98 3.6 98 JUYSEPE. « 14 swims IHL HEP IH EH EW gE 102 4.8 101 1.06 101 5.3 101 Oct-Dec. ; . : ws vs vss 3 sams vw sv Ems nina 101 3.6 101 1.12 100 4.0 100 1965: Jan-Mar. . . LLL ee 100 4.0 100 12 100 4.1 100 Apr.-dune LLL LLL eee 97 4.2 97 14 97 37 97 'N = number of interviewers included in the computations. 2 No information available for this quarter. 15 Table 6. Average interviewer performance on various measures by number of quarters of experience Average Average Average number Aerate id Averge number Average number Average Average propor- Number of Number | error! rate | noninterview of conditions of diagnostic © seed of missed Bospi- production | tion of reinterview . errors per 100 conditions per talizations per + + quarters of of inter- (percent) rate (percent) per person COGTIONs 100 persons 1,000 persons ratio assignments accepted experience viewers ! Rate | N* | Rate N Rate N Rate N Rate N Rate N Rate | N Rate N Vivnnvensoomes 97 | 123 | 48 5.2 97 1.01 95 6.9 95 40 46 .76 48 J1 | 4 79 19 Zoiviiinsaunme 94 | 129 | 46 4.1 91 .96 89 5.9 89 26 46 1.61 46 78 | 42 83 35 Rn 90 94 | 46 4.4 89 95 89 4.7 89 21 48 .30 47 90 | 42 .89 35 a 82 75 | M1 4.3 80 1.00 81 4.1 81 .26 44 .78 42 95 | 37 83 29 - JO 77 6.0 | 38 5.0 52 1.06 52 38 52 20 40 1.51 40 | 1.04 | 33 91 23 B.uvvsnsnnnamnn 75 53 | 38 4.0 73 1.02 73 28 73 14 40 1.00 63 | 1.05 | 34 97 29 Tov vv vv bo wnis swims 74 6.7 | 38 4.4 74 1.01 74 3.1 74 10 40 .49 40 99 | 42 94 31 Br vv n vn n wn pinrnne 67 41 | 33 35 67 99 67 24 67 N2 39 .28 40 | 1.00 | 37 1.00 29 DE REE Eh ere 63 44 | 33 38 63 1.01 63 2.3 63 10 36 15 36 | 1.05 | 33 .88 24 VW cuevvnnnsing 63 46 | 34 3.3 63 1.02 63 24 63 2 40 .29 38 | 1.07 | 37 89 36 ¥ ncesvsasine 60 49 | 32 3.7 60 1.02 60 2.2 60 .15 39 12 36 | 1.05 | 36 93 29 12 ciieenies am 58 44 | 34 3.1 58 1.01 58 24 58 16 39 52 36 | 1.06 | 35 1.00 23 Wisusnsnansamy 58 44 | 39 4.2 58 1.06 58 2.6 58 14 42 .28 40 | 1.10 | #1 1.00 31 Bovicinnsns um 57 49 | M1 29 57 97 57 2.7 67 19 a4 .49 43 | 1.05 | 42 91 32 LE 56 3.7 | 56 4.2 56 1.01 56 25 56 16 48 11 48 | 1.10 | 46 94 36 18... cvuvun nine 55 42 | 55 38 65 1.06 55 29 55 12 48 .20 48 | 1.05 | 46 1.00 34 ¥en nme 48 44 | 47 4.1 a7 1.08 47 2.6 47 .16 47 .40 47 | 1.10 | 47 88 a 18... 48 39 | 48 3.2 48 1.06 48 2.6 48 .09 48 1.13 48 | 1.10 | 48 95 38 W.sisessasime 44 45 | 44 4.1 44 1.06 44 2.6 44 14 44 .00 44 | 1.03 | 44 95 37 200: ci vui naam 41 4.7 | M1 3.6 41 1.03 41 2.8 41 a7 aM 37 41 | 1.01 | 40 93 30 Rp 38 35 | 38 4.0 38 1.06 38 22 38 .28 38 .28 38 | 1.03 | 38 87 30 22 uv ivivinss we 37 34 | 37 3.9 37 1.06 37 20 37 15 37 13 37 | 1.07 | 37 96 23 2B rv vw we wn 36 40 | 36 3.7 36 1.10 36 1.9 36 24 36 .26 36 | 1.10 | 36 96 25 24 ............ 34 43 | 34 35 34 1.06 34 2.2 34 24 34 .19 34] 1.06 | 34 86 22 Bir ions. nin 30 44 | 30 3.0 30 1.06 30 2.2 30 24 30 .23 30 | 1.02 | 30 95 20 Bic rriina an 29 44 | 28 28 28 1.12 28 2.2 28 .36 28 .07 28 | 1.05 | 28 1.00 19 2 wiping mn 28 43 | 28 56.3 28 1.06 28 23 28 .20 28 69 28 | 1.06 | 28 1.00 18 wv osrrisssnw 24 4.2 | 24 23 24 1.14 24 21 24 22 24 42 24 | 1.05 | 23 .88 17 29 sisssnen 17 38 | 17 3.7 17 1.21 17 23 17 .03 17 .00 17 | 1.07 | 17 .90 10 BOivvivw ww vesssas 15 35 | 156 32 15 1.27 15 2.2 15 .03 15 .00 15 1.12 | 15 1.00 10 ! Before survey quarter July-Sept. 1961, the definition of error rate was actually an omission rate. Therefore rates for quarters of experience 1-14 exclude quarters before July- Sept. 1961. 2N = number of interviewers included in computations. NOTE: Restricted to HIS interviewers employed during the period Apr.-June 1965. by the number of quarters of experience in the HIS. As expected, the more experienced inter- viewers have a better performance, on the average, than new interviewers. From this table it appears that about eight quarters, or 2 years, of experience are necessary for interviewers to PART Il. MEASUREMENT INTRODUCTION A sample survey must take into account nonsampling errors and methods of control as well as sampling errors. The allocation of resources between control of nonsampling errors and increase in sample size is a complicated question, an answer to which is not attempted here. 16 achieve maximum performance in terms of error rates, noninterview rates, and production. How- ever, the data in table 6 do not represent a pure learning curve since the same interviewers are not included in all quarters. OF NONSAMPLING ERROR Estimates of total sampling variance for im- portant statistics can be made more or less routinely. The estimation of response variances and variances contributed by other aspects of the survey process, e.g., editing and coding, is more difficult. Particularly difficult is the esti- mation of biases in the measurement process. This part of the report describes two pro- grams carried out in connection with the HIS to obtain estimates of nonsampling error, including bias. One is the supervisory reinterview program described in part I, which, in addition to serving as a field quality control device, provides overall estimates of response variance and bias. The second is a study designed to measure the interviewer contribution to the variance of estimates from the survey. RESPONSE ERRORS AS DETERMINED BY A REINTERVIEW SURVEY Introduction In the HIS, information is obtained by per- sonal inquiry or a self-administered questionnaire on age, number of chronic conditions, number of hospital episodes, diability, and other charac- teristics. The set of measurements or observa- tions recorded in the collection operation ordinarily is examined for internal consistency and acceptability, certain ‘‘corrections” are made, and some of the entries coded to identify them in a classification system. Results are then summarized into totals, averages, correlations, or other statistical measures. Taken together, the collection and processing operations constitute the measurement process and are the source of any measurement errors. The interpretation of reinterview survey results or comparisons of results from a survey with case-by-case matched responses or measure- ments from some other source has been the subject of much research and study.?"® Some theory of measurement errors that may help in the interpretations of the results of two sets of measurements is presented in appendix III. The first set of measurements is obtained by the regular survey procedures. The second set is obtained from reinterviews or through matching of survey results, unit by unit, with records providing information similar to that obtained in the survey. HIS Reinterview Survey Results The use of the reinterview program as a device for evaluating the reliability and accuracy of statistics of the HIS is a byproduct use. There are two important respects in which the super- visory reinterviews do not meet the standards that are imposed for the original interviews. First, the supervisory reinterviews are by and large conducted by men, whereas it is a require- ment that HIS interviewers be women. Second, there is a longer time interval between the reporting and occurrence of health-related events in the reinterviews than in the original interviews. The reinterviews occur at least a week later than the original interviews. There are, however, some offsetting factors. For 80 percent of the households selected for reinter- view, the reinterviewer has the benefit of the results from the original interview. Where differ- ences exist, the reinterviewer is to determine the proper answer and also possible reasons for the difference. It seems reasonable that in general better responses would be obtained from recon- ciliation of two interviews than from a single interview. The original survey data can be compared with reinterview data under three procedures. Procedure 1.—The reinterview in 20 percent of the households in the reinterview sample is conducted without the results of the original interview being available to the reinterviewer. No reconciliation of results is carried out. Procedure II. —The results of the original interview are available to the reinterviewer for 80 percent of the households in the reinterview sample. However, the reinterviewer is not to examine the results of the original interview until after a reinterview has been completed. Procedure II is a comparison of the results of the original interview with the reinterview before any reconciliation of responses in the two interviews is carried out. If the reinterviewer follows instructions, this comparison is the same one as procedure I. Procedure III.—After conducting the reinter- view in 80 percent of the households, the reinterviewer compares the responses obtained in the two interviews. Where differences exist, the reinterviewer tries, with the help of the respondent, to decide upon the proper response. Results of this reconciliation are compared with original results under procedure III. Summary measures.—To analyze the data obtained from a case-by-case comparison of an 17 original and a reinterview survey, certain sum- mary measures should be defined. Table 7 compares the results of an original survey with a reinterview survey. The total number of differ- ences affecting the tabulated figure for a given class is equal to the number of cases included in that class in the original survey but not in the reinterview survey plus the number of cases included in the reinterview survey but not in the original survey. This sum is called the gross difference for the class in question. In terms of table 7, b+c¢ is the gross difference, and (b + c)/n is the gross difference rate. The net difference of the tabulated figure for a given class is the difference between the total for the class obtained in the reinterview and the original surveys. The gross difference usually includes differences in both directions that partly or substantially offset each other. The net difference is the nonoffsetting part of the gross difference. In table 7, the net difference is b - ¢, and (b - ¢)/n is the net difference rate. Net differences.—Table 8 summarizes the net difference rates for procedure III for 7%-year averages. Since procedure III provided an oppor- tunity for reconciliation of differences, the estimated net differences obtained from it are regarded as the best estimates of bias that the supervisory reinterview program can provide. Except for persons with one or more chronic conditions, the net difference rates would be regarded as small by almost any standard. However, as indicated in the table, all of the net difference rates are statistically significant; i.e., significantly greater than zero. Many of the results from the HIS are pub- lished as rates per person. Table 9 presents rates from the reinterview survey for the original interview and for the reinterview after recon- ciliation. The percent net differences shown in the table can be considered as an estimate of the relative bias of the original survey results. According to the reinterview, chronic conditions tend to be underreported by about 24 percent. Disability days are underreported by about 13 to 18 percent. Table 9 shows that hospital episodes and hospital days are better reported. Estimates made from 1959-61 reinterview survey results show percent net differences of about 8 percent for hospital episodes and about 5 percent for hospital days. Gross differences and the index of inconsist- ency.—Gross differences are differences in indi- vidual classifications between the original inter- view and the reinterview. As discussed in appendix III the gross difference rate can be used to estimate the simple response variance of the original survey estimates, that is, the basic trial-to-trial variability in survey responses. (Appendix III also shows the derivation of an index of inconsistency based on the gross difference rate. This index provides a measure of the unreliability or inconsistency of classifi- cation and is defined as the ratio of the simple response variance to the total variance.) Reinterview without reconciliation (proce- dure I) provides the best estimate of simple response variance. However, the data from pro- cedure I were tabulated only for fiscal years 1959-61. The gross difference rates and indexes of inconsistency shown in table 10 are based on data from procedure III after reconciliation. Table 11 shows consistent declines in moving from procedure I to procedure II to procedure Table 7. General representation of results of original and reinterview surveys for identical persons Results of original survey . i Number Number Results of reinterview survey ; : having not having Total the charac- | the charac- teristic teristic Number havingthecharacteristic . . . . . . . «+ + 4 ct tv vt vv tv eer een a b atb Number not having the characteristic . . . . . . . . . . . . . «vv uu.. c d c+d LC LP atc b+d n=at+tb+c+d 18 Table 8. Estimated proportions, net difference rates, and standard error of net difference rate for procedure 111 after reconciliation for a 7%-year quarterly average 3 Estimated standard Percent in . Net Percent in . error of average . : class on difference y Survey item, persons with: Lo. class on net difference rate original 3 rate interview reinterview (percent) Underestimate | Overestimate One or more chronic conditions . . . . .......... 42.3 49.2 -7.0 0.2 0.4 One or more hospital episodes in past 12months . . . . . . eee. 9.3 10.0 -0.6 0.1 0.1 One or more restricted activity days in past 2WeEKS , 4 iw iw Ee BE REE IE SERB EE RE 10.6 12. -1.5 0.1 0.2 One or more bed days in past2weeks . . . ........ 5.6 6.4 -0.8 0.1 0.2 One or more time-lost days in past 2 weeks . . . ..... 3.4 4.2 -0.8 0.1 0.2 NOTE: Includes fiscal years 1959, 1960, 1961, 1963, 1964, 1965, one-half of fiscal year 1966 and all of fiscal year 1967. A fiscal year runs from July 1 to June 30. Table 9. Estimated annual rates per 100 persons, original interview and reinterview, and percent net difference for procedure Ill after reconciliation LL Reinterview Percent i Original Characteristic . . after recon- net interview ren . 1 ciliation difference Rate per 100 persons Chronic conditions? . . . i.e ee ee ee ee 82.0 107.4 -23.6 Restricted activity days? . . . . . oo. i te ee ee ee eee 1,383.6 1,596.7 -13.3 Bed days? LL. Le ee ee 466.3 544.5 -14.4 Time-lost days? LL LLL ee ee eee ee 287.1 351.2 -18.3 Hospital opisodes® . . wc ves ss ER i HE TIBI BE NAH UE HEHEHE NETL HF wx 9.9 10.7 -7.5 Hospital days® ws ns is ss Hi Re Fs MER EA MW MF BERN SR AGEN EF 8 0 3 94.6 99.7 -5.1 re ; ; Original — reinterview X 100. reinterview %7%-year averages. 33-year averages. Table 10. Estimated proportions, gross difference rates, and indexes of inconsistency for procedure 111 after reconciliation for a 7%-year quarterly average pa/eatin Percent in Gross Index of i class on . . t Survey item, persons with: original class on difference [inconsistency . 9 : reinterview | rate X 100 X 100 interview Oneormore chronic conditions . « « « 5 w 5 oo 5:8 20 3 8 5 § #1 5 & & 42.3 49.2 8.5 17.2 One or more hospital episodes in past T2months ©... LL ee ee ee eee ee eee 9.3 10.0 0.9 5.5 One or more restricted activity days in past 2WBBKE . . . yi vs mses wt EE EEE EY BE NE EEE EE 10.6 12.1 3.4 17.2 One or more bed days in past2 weeks . . . . .. ............. 5.6 6.4 2.3 17.5 One or more time-lost days in past 2weeks . . . . . . ... ........ 3.4 4.2 1.6 23.4 NOTE: Includes fiscal years 1959, 1960, 1961, 1963, 1964, 1965, one-half of fiscal year 1966, and all of fiscal year 1967. A fiscal year runs from July 1 to June 30. 19 Table 11. Estimated indexes of inconsistency by three procedures, 3-year averages, fiscal years 1959-61 Procedure | Procedure Il | Procedure 111 Survey item, persons with one or more: (no recon- (before rec- | (after recon- ciliation) onciliation) ciliation) Chronic conditions inpast 12months . . . . . . . . . «ov vii 30.9 22.2 17.4 Hospital episodes in past 122 months. . . . . . . . . . .. .. 7.6 7.0 6.0 Restricted activity days inpast 2weeks . . . . . . . . . ..... eee 44.5 28.6 18.3 Bed daysinpast2weeks . . . . .. LL... eee 41.1 26.6 15.8 TIMRAOSE dayB IN PASE2 WEBKE + o vs ws 0 5 © 6 Bi iW Wi Si & # BE ® 8 8 20 4® 3 37.6 32.9 21.4 Hospital days inpast 2weeks . . . . . ou uv i tie ee ee ee 12.8 19.5 19.1 III in the estimated indexes of inconsistency. If procedures I and II were carried out as specified, then the expected difference between the indexes of inconsistency would be zero. The estimated indexes for procedure II, however, are about 20 percent smaller than the indexes for procedure I. The differences between procedures II and III are in the direction that would be expected: reconciliation reduces the gross difference rate. Thus, the estimates in table 10 are an under- statement of the gross differences that would occur if repetitions of the HIS were carried out without reconciliation. Some values of indexes of inconsistency for demographic items computed from other studies are provided in table 12 by size classes and compared with indexes for health items com- puted from the reinterview program of the HIS. The index of inconsistency for hospital epi- sodes in the past 12 months is in the same size class as the simpler demographic items such as sex, color, and age. Time-lost days compare with the more diffi- cult items to measure such as income and educational attainment. Comparison of self-respondents with proxy respondents.—In the original interview for adults, the health questions are asked of the person himself if he is home at the time of the interview. If he is not at home, a related adult may provide the information. The person who is not present at the time of the interview is referred to as a proxy respondent since the information on such a person is obtained by proxy. However, in reinterviews all adults are self-respondents (in all reinterviews conducted during fiscal years 1959-67). Table 13 shows 20 3-year averages of proportions, net difference rates and indexes of inconsistency for six survey items as reported by self-respondents and proxy respondents in the original survey. The data are based on results after reconciliation (proce- dure III). Comparison of self-respondents and proxy respondents are limited by the fact that the selection of the respondent on the original interview is not a random selection. Thus part of the differences in level may be attributed to inherent differences between respondents who are available to report for themselves and re- spondents who are not available at the time of interview and whose health conditions are reported by another member of the household. For four of the six items, the reconciliation tends to bring the proportion in the class for proxy respondents closer to that of self- respondents, that is, the net difference rates are greater for proxy respondents. The estimated prevalence rates for chronic conditions per person from the reinterview survey are presented in table 14 for self- respondents and proxy respondents. In attempt- ing to estimate what effect the respondent has, certain assumptions were made about the differ- ences. Specifically an assumption was made that for self-respondents the net difference between the original rate and the reinterview rate can be considered as the difference due to the second interview. For proxy respondents, the net differ- ence consists of second-interview differences and differences due to the use of a proxy respondent and that these differences are addi- tive. Furthermore, an assumption was made that the differences due to the second interview are the same for self-respondents and proxy Table 12. Comparison of estimated indexes of inconsistency for Health Interview Survey items with demographic items Current 1960 Decennial Population HIS Size of index (X 100) Census Evaluation Survey . i 1 3 J Reinterviews Program Reinterviews 1961-662 Lg | ERR SY RE Tt E SE EE A Et AE EE EY Sex Employed Hospital episodes Color In labor force Age 1120 «5 26s ssn sd md hd TERI BIBI RIB IE SE GS Labor force Unemployed Hospital days Mobility Chronic conditions Bed days Restricted activity days jo 1 EE EE SE EE EE EE EE AE ET Educational attainment Time-lost days Income 'U.S. Bureau of the Census: Office, 1964. Evaluation and Research Program of the U.S. Censuses of Population and Housing, 1960: Accuracy of Data on Population Characteristics as Measured by Reinterviews. Series ER 60-No. 4. Washington. U.S. Government Printing 2U.S. Bureau of the Census: The Current Population Survey Reinterview Program January 1961 through December 1966. Tech. Paper No. 19. Washington. U.S. Government Printing Office, 1968. respondents. The last column of table 14 shows the net difference expressed as a proportion of the reinterview estimate, and the difference between these relative net differences is a measure of the effect of proxy respondents, that is, that about 16 percent of the relative net difference for persons reported for by proxy respondents can be attributed to the use of a proxy respondent. Effect of nonreporting on estimates of magni- tude. —Differences in reporting of the number of conditions, days, episodes, and so forth between the original interview and the reinterview are classified as follows: differences due to a change in the number of conditions, days, episodes, etc., reported on the two interviews or as differences due to a report of no conditions, days, episodes, etc., on one interview and a report of one or more conditions, days, episodes, etc., on the other interview (table 15). The change from a report of none on one interview to a report of one or more conditions, days, episodes, etc., on the other interview has a relatively small effect on estimates of propor- tions, but with the exception of hospital days in the past 2 weeks it has a major impact on estimates of magnitude. This section presents estimates of the part of the net and gross differences for estimates of magnitude that can be accounted for by a change from a report of none on one interview to a report of one or more on the other interview. Table 16 shows estimates of the components of the total net difference and the ratio of each component to the total net difference. Table 16 shows that for hospital episodes, restricted activity days, bed days, and time-lost days, the net increase on reinterview for esti- mates of magnitude is principally due to the change from a report of none on the original interview to a report one or more on reinter- view. For hospital days the number of days accounted for by changes from a report of none on the original interview to a report of one or more on reinterview is about the same as the number of days involved in changes of a report of one or more on the original interview to a report of none on reinterview and thus cancel for estimates of net difference. For chronic conditions, the changes in the number of condi- tions reported for cases which are one or more on both interviews are close to the number of conditions accounted for by changes of a report of none on the original to a report of one or more on reinterview and increase the volume of conditions reported on reinterview. Changes 21 Table 13. Estimated proportions, net and gross difference rates, and indexes of inconsistency by subject and respondent on original interview compared with proxy respondents, procedure 111, 3-year averages, fiscal years 1959-61 Persons . Average Average . . in class Index Survey item, subject and respondent net gross of on original interview, persons with: . difference difference |. LC . Reinter- inconsistency Original y rate rate view One or more chronic conditions: Adult, self-respondent . . . . . ........ 0... 59.4 64.8 -5.4 6.1 13.1 Adult, proxy respondent . . . LL... ee eee eee. 47.4 58.5 -11.1 14.9 30.3 (self-respondent — proxy respondent) . . .. ........ 12.0 6.3 5.7 One or more hospital episodes in past 12 months: Adult, self-respondent . . . . ..... 13.2 13.5 -0.3 0.6 2.4 Adult, proxy respondent . . . LL... eee ee ee 9.0 10.0 -1.0 1.3 73 (self-respondent — proxy respondent) . . .......... 4.2 35 0.7 One or more restricted activity days in past 2 weeks: Adult, self-respondent . . . . . ................ 135 14.3 -0.8 35 14.8 Adult, proxy respondent . . . LL... LLL... 8.2 11.0 -2.8 5.0 28.9 (self-respondent — proxy respondent) . . .......... 5.3 33 2.0 One or more bed days in past 2 weeks: Adult, self-respondent . . . vx fie ns ws EEE su se ss 6.7 7:7 -0.4 1.7 13.4 Adult,proxyrespondent . . . . ....... 4.00420... 4.7 55 -0.8 2.0 21.1 (self-respondent — proxy respondent) . . .. ........ 2.0 1.6 0.4 One or more time-lost days in past 2 weeks: Adult, self-respondent . . . . ... Le 3.1 3.5 0.4 0.8 11.8 Adult, proxy respondent . . . . LL... ee eee. 4.3 4.7 -0.4 2.4 28.2 (self-respondent — proxy respondent) . . .......... -1.2 -1.2 - One or more hospital days in past 2 weeks: Adult, self-respondent . . . . . . . . LL... 0.7 0.7 - 0.3 21.6 Adult, proxy respondent . . . . LL... ue ee eee 0.4 0.4 - 0.2 23.9 (self-respondent — proxy respondent) . . . .. ....... 0.3 0.3 - Table 14. Estimated prevalence rate of chronic conditions per person by subject and respondent on original interview, original and reinterview estimates for a sample of identical persons, reconciled reinterviews, fiscal years 1959-61 Chronic conditions per person (12 months) Subject and respondent on original interview Original Reinterview Net (a) - (b) (a) (reconciled) | difference ET (b) (a) — (b) Adult, self-respondent . . . LL LL LL LLL ee ee ee ee 1.26 1.52 -.26 -17 Adult, proxy respondent . . . . LL uh hh ee ee ee ee ee eee .84 1.26 -.41 -.33 Self-respondent — proxy respondent . . . LL... LL ee eee .16 Estimated sampling error of difference . . . . . . .. .. ............. .02 NOTE: A similar comparison for a different time period and including all conditions, acute and chronic, is presented in table 15. The magnitude of effect that the type of respondent has is about the same. 22 Table 15. Estimated rate of conditions per person by subject and respondent on original interview, original and reinterview estimates for a sample of identical persons, reconciled reinterviews, fiscal years 1963-67 Conditions per person Subject and respondent vi Reinterview Net at original interview Original . . (a) — (b) (a) (reconciled) | difference TT (b) (a) — (b) Adult, self-respondent . . . . LL LLL LL ee eee eee ee eee 1.61 1.74 -.13 -.07 AdUIL, roxy 7eSpONGENt + . « ww x + 5 5 5% Hs 58 EE EEE ELE Eee 1.00 1.31 -.31 -.24 Self-respondent — proxy respondent . . . . . LLL. 0 0a ee eee. 7 Estimated sampling error of difference . . . . . . ......... 0... .01 Table 16. Estimated total net difference between original interview results and reinterview results and components of net difference, reconciled reinterviews, fiscal years 1959-61 Ratio of components Components of total net difference to total net difference Total net Survey item ditraroncs Due to change Due to Change Due to change 01 Magnitude (x —y) in reporting in reporting from in reporting of . I Lo. . difference [difference from presence on absence on original | magnitude when 2) 13) 4) original to absence to presence on present on —m mM on reinterview reinterview both interviews? (1) (2) (3) (4) (5) (6) Chronic conditions in past 12 months . . . .... -26,700 964 -12,541 -15,123 43 57 Hospital episodes in past 12 months . ...... -881 220 -969 -132 85 A585 Hospital days in past 2WeEKS i cnn amansn 185 584 -552 153 16 .83 Restricted activity days inpast 2weeks ....... -13,908 7,244 -17,003 -4,149 .70 .30 Bed days in past 2weeks . LL... -3,692 1,833 -4,921 -204 94 .06 Time-lost days in past 2vweeks ;.. hvu ss -3,361 1.327 —-4,684 -4 1.00 (3) ! x is the estimate of magnitude from the original interview, and y is the estimate of magnitude from the reinterview. 2 Algebraically this component is [column (1)] — [column (2) + column (3)]. A minus sign indicates a net increase on reinterview. 3 Less than 0.005. NOTE: The tabulations for hospital days, restricted activity days, bed days and time-lost days are in terms of 2-day intervals. The components of the net difference were estimated by using midpoints of the 2-day intervals. Changes in response that did not result in a change of class interval have no effect on the estimates. from a report of one or more on the original to a report of none on reinterview do not have much effect on the estimates of chronic conditions. A large proportion of the gross differences in estimates of magnitude is accounted for by changes from a report of none to a report of one or more in both directions. The following index is an estimate of the proportion of the total response variance that is accounted for by cases which either change from a report of none on 23 the original interview to a report of one or more on reinterview or from a report of one or more on the original interview to a report of none on reinterview: 2 feolx = 0)2 +3 f4,(0- y)? x=] y=1 3 3 file 3) y=0 x=0 where x is the value on the original interview, y is the value on reinterview, f, o is the number of persons with x value on original and 0 on reinterview, f,, is the number of persons with 0 on original and y value on reinterview, and Ley is the number of persons with x value on original interview and y value on reinterview. Table 17 shows that except for chronic conditions most of the response variance for estimates of magnitude can be accounted for by changes in the reporting of the presence or absence of the characteristic. Tables 18-22 present estimates by year of the summary measures discussed in previous sec- tions. All of these results are for procedure III, that is, a comparison of the original interview with reinterview after reconciliation. The yearly estimates are subject to large sampling errors since the reinterview sample in any one year is relatively small. One additional summary meas- ure, the index of net shift, is presented. The Table 17. Proportion of total response variance due to changes in reporting of presence or absence of characteristic pro- cedure 111, reconciled results Characteristic G Chronic conditions in past 12 months . . . . .... .. 41 Hospital episodes in past 12 months. . . . . ...... 92 Hospital days inpast 2weeks . . . . ........... .96 Restricted activity days in past 2 weeks . . . ...... .88 Beddaysinpast2weeks . . . . .. .. . 4... .80 NOTE: Hospital days, restricted activity days, bed days, and time-lost days for the tabulations are in terms of 2-day intervals. Midpoints of the 2-day intervals were used for the x and y values. Changes in response which did not result in a change of class interval have no effect on the estimates. 24 index of net shift is simply the ratio of the net difference rate to the percent in class on reinterview. INTERVIEWER VARIABILITY STUDY Introduction The joint effects of sampling and nonsampling errors determine the accuracy of survey results. The mathematical model of response errors in surveys presented in appendix III shows how the mean square error of a statistic is divided into its various components: sampling variance, response variance, interaction, and square of bias. The response variance can be further divided into simple response variance and correlated response variance. In the preceding section, estimates were presented of the simple response variance and of response bias as measured by reinterviews. The usual estimates of sampling variance include the simple response variance and possibly a small part of the correlated response variance due to field interviewers. However, the major part of the interviewer contribution to response variability is not in- cluded in the estimates of sampling variance. This section describes an interviewer variance study designed to measure the contributions of interviewers to the variability of health statistics. Significant between-interviewer variance in the reporting of health data has been observed in a number of studies. Data from other studies also indicate that interviewer effects may oper- ate differently for different statistics. Design of the Interviewer Variance Study The interviewer variance study was conducted over the 4-year period 1960-63. For the first 2 years of the study, randomization of interviewer assignments was carried out in eight large SMSA’s where there were two or more HIS interviewers. The study included 10 SMSA’s during the second 2-year period. The assign- ments (within each pair or triplet) of inter- viewers within an SMSA were randomized in an interpenetrated design so that each interviewer of a pair would have produced results with the same expected value if there were no between- interviewer variability. Table 18. Estimated proportions, net difference rates, indexes of net shift, gross difference rates, and indexes of inconsistency for procedure Ill after reconciliation for persons with one or more chronic conditions in the past 12 months, by year Peroansle Percent in Net Index of Gross Index of Fiscal year original class on difference | net shift! difference [inconsistency . . reinterview | rate X 100 X 100 rate X 100 X 100 interview ODD) 51 5 50 151 br 4 5 nt or 8m we mR ee on 38.5 45.2 -6.8 -15.0 8.6 17.9 1960 oi eee eee 42.5 49.3 -6.7 -13.8 7.9 16.2 1B vss vv mss smn sma rma 41.7 49.2 -7.5 -15.3 8.9 18.1 1962 iii ae NA? NA NA NA NA NA OBB vias mai BA me cuins RRSP mA som mmm 41.5 48.2 -6.7 -13.8 8.3 16.8 1964 ea 42.8 50.5 -7.7 -15.2 9.5 19.3 18BB uns vmmsmmmiwmo imme s wus smbsme 43.7 51.3 -7.6 -14.8 8.7 17.6 198% ov inuimnmimms snus snus pA ITE 48.5 56.2 -7.7 -13.8 8.9 17.9 TOB7 wn iaintiiims ims idimd samp smm nm 46.3 51.9 -5.5 -10.7 6.7 13.4 3 Net difference rate Percent in class on reinterview 2 Not available. 32 quarters only (July 1, 1965-Dec. 1965). Table 19. Estimated proportions, net difference rates, indexes of net shift, gross difference rates, and indexes of inconsistency for procedure 11 after reconciliation for persons with one or more hospital episodes in the past 12 months, by year Percent m Percent in Net Index of Gross Index of Fiscal year Sass class on difference | net shift! difference |inconsistency orginal ointerview | rateX 100 | X 100 | ratex 100 | xX 100 interview V959 . uve runrums rons senmrwms amas 8.4 8.6 -0.1 -1.2 0.9 55 VOB0 , cuvvrmvwemms pumas immPs mes sma 9.4 10.1 -0.7 -6.9 0.9 5.2 VIBE : envi vmormms immws sRPE IRE mE 8.0 9.1 -1.1 -12.1 1:2 7.4 TOB2 4 coats swim smus sams emus ARBI smA ES NA NA NA NA NA NA TOBE . cvniims sommes smms in ha mn 9.6 10.9 -1.4 -12.8 1.5 8.3 1964 eee 11.1 11.8 -0.7 -5.9 0.8 3.6 1985 , uv vivmeroms same sume smmss mess 11.9 12.1 -0.2 -1.7 0.4 1.6 19667 Lo. 7.3 7.5 -0.1 -1.3 0.1 1.0 VOBT7 cows sms sums suis AMEE ARES EHR 8.7 9.1 -0.4 -4.8 1.1 6.9 Net difference rate Percent in class on reinterview 22 quarters only (July 1, 1965-Dec. 1965). For the first 2 years of the study, calendar years 1960 and 1961, the data cover the work in 25 interviewer assignment areas. Six of the eight SMSA’s had a pair of interviewers, Los Angeles had a triplet, and the New York SMSA had five pairs of interviewers. For the second 2 years, calendar years 1962 and 1963, the data cover the work in 30 interviewer assignment areas for two quarters and the work in 28 interviewer assignment areas for the remaining six quarters. (The HIS sample was redesigned in 1962, and one interviewer pair was dropped in the New York SMSA.) Assign- ments in Chicago and Los Angeles were ran- domized among three interviewers during the second 2-year period. During the 4-year period, a total of 6,415 segments of six to nine households were in- 25 Table 20. Estimated proportions, net difference rates, indexes of net shift, gross difference rates, and indexes of inconsistency for procedure |] after reconciliation for persons with one or more restricted activity days in the past 2 weeks, by year Pariabtls Percent in Net Index of Gross Index of Fiscal year origina) class on difference | net shift’ difference [inconsistency IEEE reinterview | rate X 100 X 100 rate X 100 X 100 1980 overt mn rms mm eS 10.4 11.9 -1.4 -11.8 4.0 20.6 1960 .... i ie ee 10.5 12.8 -2.3 -17.9 4.4 22.0 1961 . i eee 12.4 14.1 -1.7 -12.1 3.3 14.6 TOBZ cnn cmm ptm mR AEP Sm HE SEBS BI NA NA NA NA NA NA 1983 .isrunms smu inns sms FRAG EHBI ER 9.5 10.0 -0.5 -5.0 2.5 14.2 TO84 vin oninass inntmus ERE s0E0@E 10.9 11.9 -1.1 -9.2 2.3 11.1 1985 .:vinimsivimsa smi ds samme sons EH 11.1 12.3 -1.3 -10.6 3.0 15.0 1966” Le 6.4 7.5 -1.1 -14.7 18 13.9 1967 «it eee 10.9 13.6 -2.7 -19.9 4.5 23.1 1 Net difference rate Percent in class on reinterview 22 quarters only (July 1, 1965-Dec. 1965). cluded in the assignments. Of these segments, 1,204 were excluded from the analysis for the following reasons: Percent Reason for exclusion of total segments Total assigned segments not used in ONBIYSIS . vos 3.5 36 BE Fw WE HENS 18.8 Segments completed by “other than assigned IBrVIeWRrS” . . u,v as sw ss a sw ow 12.4 Segments in non-self-representing PSU's . . . . . . 3. Segments where it could not be determined who completed the assignments . . . .. ...... 1.2 Segments containing ‘‘special dwelling places” . . . 2.0 The remaining 5,211 segments used in the analysis contained about 34,000 households with 110,000 persons. Method of Analysis For this analysis, ratios of two variables were used at all times for two reasons. First, most HIS data are presented as rates or proportions of the total population; and, second, this method minimizes the effect of the variability in size of interviewer assignment on the analysis. The starting point for the analysis was a computer tape containing quarterly totals for 26 each of 84 health and demographic character- istics for each of the interviewer assignment areas. The mathematical model used to compute the total variance is basically a comparison of results between a pair of interviewers, extended to the whole study area. The task was to estimate the expected value of the difference between two interviewers’ findings for a specified period of time. In the model, the total relvariance for the study area is AE (r, Ty )? rs Fra — (1) Er? where r, is the ratio computed from a random half of the interviewers and r, is the ratio computed from the other random half and the expected value is taken over all possible half- samples. An estimate of this expected value was made by assigning each interviewer in a pair to a or b at random and then averaging over all pairs of interviewers.® Different permutations of the pairs give other estimates. Twenty-five permuta- tions were used to give the estimator: bgtrictly speaking, it is interviewer assignment area instead of interviewer. When there was a change in interviewers due to resignation, illness, or other administrative reason other than a temporary substitution, the work of the replacement interviewer was treated as part of the same interviewer assignment area. Table 21. Estimated proportions, net difference rates, indexes of net shift, gross difference rates, and indexes of inconsistency for procedure 11 after reconciliation for persons with one or more bed days in the past 2 weeks, by year Percent n Percent in Net Index of Gross Index of Fiscal year, Co » class on difference | net shift! difference |inconsistency ‘ gina reinterview | rate X 100 X 100 rate X 100 X 100 interview 1969 .... iii i ii rt eee 5.6 6.1 -0.5 -8.2 3.8 15.8 T1960 ... eee 55 6.4 -0.9 -14.1 2.1 18.3 FOBT .vsems swmpanis swms Simms § ims me 5.4 6.3 -0.9 -14.3 1.7 15.8 1982 :isivnvinaniissnis ss immrswumins NA NA NA NA NA NA 1983 ........c¢'ititrit rrr 5.3 6.0 -0.6 -10.0 24 19.9 1964... eee 6.0 6.4 -0.5 -7.8 1.6 138 VOB5 wv swmp imme smms sms (HE EH DINE 5.1 6.0 -1.0 -16.7 1.2 12.7 1087 vsiivs snus sho amme i was 6m osm 4.5 5.0 -0.5 -10.0 1.7 22.7 y 1: 7 Sg 6.8 8.8 -1.9 -21.6 3.4 25.6 i Net difference rate Percent in class on reinterview 22 quarters only (July 1, 1965-Dec. 1965). Table 22. Estimated proportions, net difference rates, indexes of net shift, gross difference rates, and indexes of inconsistency for procedure Ill after reconciliation for persons with one or more time loss days in the past 2 weeks, by year Percent in Percent in Net Index of Gross Index of Fiscal year iainal class on difference | netshift' | difference [inconsistency Srigina reinterview | rate X 100 X 100 rate X 100 X 100 interview 1959 LL. Fa 4.2 4.9 -0.7 -14.3 1.4 19.3 1960 ... eee 3.7 4.9 -1.2 -24.5 1.7 241 VB ous smssamssmws rear smmmemes 3.3 3.8 -0.5 -13.2 2.0 29.5 1982 : iwns sume mms sms s Mes Rms smote 3 NA NA NA NA NA NA T8983 : cows enmimus sms s RAP EWES IDE 3.1 3.8 -0.7 -18.4 1.1 15.8 2 2.6 31 -0.5 -16.1 16 31.5 1965 oe 3.4 4.2 -0.8 -19.0 1.2 15.0 19667 3.0 3.1 -0.1 -3.2 0.6 10.7 VTIBT uur vmmemms rnms slums simeisio oss 3.6 5.2 -1.6 -30.8 2.6 37.2 ! Net difference rate Percent in class on reinterview ?2 quarters only (July 1, 1965-Dec. 1965). y 23 . 5 interviewed a separate random sample of house- 95 2 Aria = Tip) holds. Estimates of V2 were made for a single 2 — =) (2) quarter of data, two quarters combined, three Tr r2 quarters, up to an estimate based on eight quarters of data. where j denotes the permutation and r the In the analysis, the results were treated as if estimated ratio computed from the work of all interviewers. This estimator has two components that must be identified separately: the between-interviewer relvariance and a sampling relvariance arising from the fact that each interviewer of a pair they were two independent studies of 2 years each. One reason for treating the results as two observations was to minimize the effect of interviewer turnover. Another reason was strict- ly practical; 2 years of data could be handled more easily than 4 years. 27 If it is assumed that the sampling variance is a function of sample size but that the estimate of the between-interviewer variance is a function only of the number of interviewers, then N ’ N V2 2 = 2 E Ve = Vi+ on (3) n . . where 4 is the estimate of the between- interviewer relvariance and 2 /n is the estimate of the sampling relvariance for a single quarter divided by the number of quarters used in the estimate. The parameters of this function were estimated in terms of V2 by the method of least squares (appendix IV). This least-squares solu- tion was then used to determine #2 and 2. Figure 1 shows an example of the expeciad behavior of 1) as the sample size increases. As the number of quarters included in the estimate increases, the sampling component of the variance decreases and the estimate of total variance approaches the between-interviewer variance asymptotically. The assumptions of equation 3 are un- doubtedly not fully warranted. There is evidence from other studies conducted by the Bureau of the Census that the response variance cannot be viewed as a constant even if, as is not the case in the present study, the interviewers did not change over the 2-year period. To illustrate further that the assumptions are not completely true, it can be shown that the response variance is the sum of two terms. The first is a simple response variance, which expresses the variability in repeated measures (interviews) on the same persons. This quantity varies inversely with the sample size and thus NUMBER OF QUARTERS Figure 1. Expected reduction in vi as sample size increases. 28 depends on the number of quarters of data. In the estimation scheme used in this study, the simple response variance has been treated as part of the sampling variance and subtracted out. Thus, the estimator V2 is somewhat of an underestimate. The second term is the product of two factors: the simple response variance and a correlation expressing the extent to which each interviewer tends to introduce her own system- atic response errors in her assignment (see appendix III). An assumption of this study is that this correlation is a constant over the 2-year period, although there is evidence from other programs that the correlation may decline with an increase in experience because of training, increased proficiency, and the increasing hetero- geneity in the population included in the assignment. Results Tables 23 and 24 contain the results of this study. These results are presented in two tables, one for each 2-year period. Each table shows results for 67 items. The estimated relvariances shown in the tables relate to annual estimates prepared from the work of approximately one-fourth of the inter- viewer staff. For some items the estimates of between-interviewer relvariances are negative. These are items which presumably have a very low interviewer variability. The sample size is too small to place much reliance on the specific estimates of the relvari- ances; however, the last column of the tables is probably sufficiently reliable to provide a general ranking of the characteristics. Fhis last column expresses the between-interviewer rel- variance as a proportion of the total relvariance. Discussion of the results shown in tables 23 and 24 is based on this last column. The results shown in the tables indicate that the reporting of chronic conditions and activity restriction associated with such conditions have the highest ratios of between-interviewer vari- ance to total variance. In addition, as might be expected, reporting of income also has a fairly high ratio. As has been observed in other studies of between-interviewer variance, reported in papers Table 23. Interviewer variance study of estimates of components of relvariance for annual estimates of selected characteristics based on 25 interviewer assignment areas in eight SMSA’s: United States, 1960-61 Between- g Denominator a Total rel- interviewer Sampling 2 Numerator a Ratio variance . relvariance | —L_ (total number of the following) 02 relvariance 02 02 T 0? E T Health characteristics (Magnitude items): CONSItONS 1 ¢ snivic sms sms 3 RSE S Ramps AME sos Households 2.563 .00395 .00369 .00026 934 Chronic conditions for females . . . . . ................. Females .769 .00483 .00442 .00041 915 Chronic conditions formales . . . .................... Males 612 .00434 .00370 .00064 854 Chronic conditions with 1 or more bed days in last T2MOMDS © vox mnie cma FEES Tamms smd mom rw Persons 115 .00476 .00404 .00072 .848 Restricted activity days for chronic conditions inlast2weeks . . . .. . Le eee Chronic conditions .633 .01328 .01045 .00283 .787 Restricted activity days inlast2 weeks . . . .. ............. Persons .602 .00411 .00296 00115 720 Acute conditions... Lo... ee Persons 115 .00285 .00202 .00083 .708 Chronic conditions with 1 or more bed days in last 2WeeKS LL. LL eee ee eee Persons .024 .00861 .00604 .00256 .702 Disability days in last 2 weeks from all accidents . . Lo. Lo... ee eee eee Accidents 1.182 .01289 .00842 .00447 663 Restricted activity days in last 2 weeks for acute CONDONE. ov «uv viv vw mis www ek mm ow iF Fe Ee EE Acute conditions 3.119 .00273 .00135 .00138 .494 Restricted activity days in last 2 weeks for acute conditions fOrmales . « « «vic sw ws sm ws ke s Ewa §E Acute conditions for males 2.996 .00524 .00244 .00281 .465 Bed days for chronic conditions in last 2 weeks . . . .......... Persons .136 .00823 .00361 .00462 439 Beddaysinlast2weeks . . . . ........ 4... Persons 227 .00268 .00115 00153 430 Bed days for chronic conditions in last 12 months. . . . ....... Chronic conditions 4.245 .00575 .00243 .00332 423 Days lost from school or work in last 2weeks . . . ........... Persons currently employed and persons aged 6-16 years 217 00359 .00135 .00224 .376 Days lost from work in last 2weeks . . . . .. ............. Persons currently employed .206 .00517 .00133 .00385 .259 Bed days for acute conditions inlast2 weeks . . . . ........... Acute conditions 1.261 .00203 -.00007 .00210 | -.036 Days lost from school in last 2weeks . . . . . . ............. Persons aged 6-16 years .239 .01041 -.00113 .01154 | -.109 Hospital discharges inlastTmonth . . . .... ............. Persons .103 .00049 -.00007 00056 | -.138 Hospitalizations in last 12months . . . . . . . . ............ Households .333 .00045 -.00011 .00056 | -.256 Hospital days in last 12months ... +5 5:5 sw wis vs mws sims sine Hospitalization in last 12 months 10.732 .00236 -.00133 .00369 | -.563 Hospital days for all discharged in last AZMOMNE ws cmv cas swois eR EEE FRG LBEE 2H E Hospital discharges in last 12 months 10.456 .00183 -.00170 .00353 | -.926 Health characteristics (Attribute items): Persons with 1 or more chronic conditions . . . . ............ Persons .389 .00196 .00174 .00022 886 Persons with 1 or more conditions. . . . ................ Persons .450 .00165 .00146 .00019 886 Males with 1 or more chronic conditions . . . . . ............ Males .369 .00230 .00193 .00037 .839 Males with 1 or more conditions. . . . .... ............. Males 428 .00198 .00165 .00033 832 Persons limited in kind or amount of activities . . ............ Persons .088 .00283 .00152 00132 535 Acute conditions medically attended . . . . . . ............. Acute conditions 731 .00031 .00009 .00022 .283 New cases acute respiratory conditions, 1 or more DRAdIYS « vv c vu sas PHBE EEE ER AEE FE HE FW Persons .026 .00424 .00108 .00317 .254 Hospitalizations for tonsillectomy or adenoidectomy formalesinlestI2ZMOonths . . uo « sow wns sw wie EBwE £0 Hospitalizations for operations, males 147 .01540 .00369 01171 .240 Hospitalizations for operations on the female genital sYStem inlast 12 months . . : wus v + wo) 5 3 € wis wit Be 5a Hospitalization for operations, females, exclusive of delivery .346 .00360 -.00107 .00467 | -.298 Persons unable to carry on major activity . . . . . .. .......... Persons .018 .00233 -.00081 00313 | -.346 Socioeconomic characteristics: Families with income > $5000 . . . .. .... ............. Families 521 .00115 .00077 .00039 665 Persons in families with income >$5000 . . . . . ............ Persons .606 .00099 .00066 .00033 .664 Families with income < $2000 . . . .. ................. Families AV? .00371 .00127 00244 .342 Persons who are not employed and not keeping house . ..... Persons aged 17 years and over .075 .00133 .00033 .00100 .246 Employed persons Persons aged 17 years and over .561 .00006 .00001 .00005 116 Persons in families with income > $2000 . .. ............. Persons .065 .00356 .00031 .00326 .086 Employedfemales ... ., .cus sass ssw mes amps sows sos Females aged 17 years and over .330 .00040 -.00003 .00042 | -.069 29 Table 23. Interviewer variance study of estimates of components of relvariance for annual estimates of selected characteristics based on 25 interviewer assignment areas in eight SMSA's: United States, 1960-61—Con. 7 Total rel- | BEWeEN | qornling 02 Denominator ; interviewer A 1 Numerstor (total number of the following) Ratio a relvariance ehvaripnce 02 T 0? E Tr Nonresponse Items: Persons with amount of education unknown . . . ......... , . . | Persons aged 17 years and over .021 .01961 .01535 .00426 .783 Heads of households with amount of education UNKNOWN vino vomim x mmis wre Wi% 5 OE wwe we we we Households .026 .02727 .02064 .00663 .757 Families with unknown family income . . ................ Families .065 .01275 .00798 00477 626 Conditions from accidents, unknown if motor vehicle . . ... ..... Accidents .003 .37186 .34642 02544 932 Conditions from accidents, unknown location . . . ........... Accidents .002 53790 58952 -.05162 | 1.096 Demographic characteristics (including health characteristics by age groups): Chronic conditions with 1 or more bed days, persons ged BAYES , . sus vm sR SE TERE AE EE Chronic conditions .018 .00883 .00545 .00337 618 Acute conditions, persons aged 25-44 years . . . . . . . ... ..... Acute conditions .239 .00133 .00036 .00098 267 Chronic conditions with 1 or more bed days, persons BGR BFE YBAS + « vk chk sk ke ss EE Ew. Chronic conditions .049 .00251 .00050 .00201 .200 Acute conditions, persons aged 15-24 years . . . . . .......... Acute conditions .102 .00520 .00068 100452 A31 Females, married, spouse present . . . . . . . . . . o.oo u i.e ea. Females, aged 17 years and over 643 .00013 .00001 .00011 .108 Persons of othervaces . « « + + wes sim as # vs mms vw ww www Persons .007 .03511 .00200 .03311 .057 Chronic conditions with 1 or more bed days, persons BRUNE BYBRIS . . up s wws sa 8 £6 EEE ERE EE Chronic conditions .007 .01386 .00076 01310 .055 Persons Of Negro race . . . . «wos sw as vs ss wwe swe 5 a6 » Persons .108 .01216 -.00011 .01227 | -.009 Acute conditions, persons aged 5-14 years . . . . .. .......... Acute conditions 241 .00167 —-.00008 .00175 | -.050 Chronic conditions with 1 or more bed days, persons OEY ANd OVEr . , ws vn ws sim win BE AME THEE REE Chronic conditions .028 .00376 -.00024 .00400 | -.064 Acute conditions, persons aged 65 yearsand over . . . . ........ Acute conditions .075 .00756 -.00066 .00822 | -.087 Males aged 17 yearsand OVer . . . . . . . . . vv vv vv i uv vo un Persons aged 17 years and over 465 .00005 -.00001 .00006 | —-.184 Persons aged under 1 year . . . . . .. .. .. . .' ou 'uuneneunn Persons 020 .00161 -.00033 00194 | -.203 Persons aged 1yearand over . . . . .. .. .. . ct vv vv uuu nnn Persons 980 .00000 -.00000 .00000 | -.214 Females aged 17 yearsandover . . . . . . . . . «vo v vv ven Persons aged 17 years and over 535 .00003 -.00001 .00004 | -.222 Acute conditions, persons aged 45-64 years . . . .. .......... Acute conditions 170 .00235 -.00074 .00309 | -.316 Acute conditions, persons aged 5 years andover . . . . ......... Acute conditions 173 .00282 -.00107 .00389 | -.380 Persons . . Lo... LL ee eee ee Households 3.167 .00008 -.00005 .00013 | -.677 Chronic conditions with 1 or more bed days, persons aged 16-24 years . . . . LL. ihe ee eee eee Chronic conditions 011 .00629 -.00430 .01058 | -.683 Chronic conditions with 1 or more bed days, persons B00 8BBA YARNS «. « «unix mimin kW RIE ER wee Chronic conditions .053 .00130 -.00092 .00223 | -.708 Persons aged 17 yearsand over . . . .. .... ..... outa. Persons 667 .00005 -.00004 .00009 | -.848 Persons aged unter 17¥0ar8. . . uu. 5 ow on + » wows vows waa Persons 333 .00020 -.00018 .00038 | -.896 by Eckler and Hurwitz,!0 and Hurley, Jabine, and Larson,!1 there are considerable between- interviewer variances in nonresponse rates. Note, for example, that the ratio of between- interviewer variance to the total variance is about .8 for the number of persons with education unknown. It is also of interest to note that the demo- graphic differentials in morbidity rates (distri- butions of acute and chronic conditions by age groups) are not subject to any significant between-interviewer variance. 30 The estimates of interviewer variability from the two periods (1960-61 and 1962-63) differ considerably for some items. The material for the first 2 years was investigated to determine the cause of the higher estimates of interviewer variability for these items. During the 1960-61 period, one interviewer of the 25 in the study contributed a disproportionate amount to the estimates of between-interviewer variability. However, the response variance study of the 1960 census! 2 demonstrated that the distribu- tion of individual interviewer contributions to Table 24. Interviewer variance study of estimates of components of relvariance for annual estimates of selected characteristics based on 28 interviewer assignment areas in 10 SMSA’s: United States, 1962-63 Between- i A Numerator Denominator Ratio You sh interviewer Sale 7 (total number of the following) (total number of the following) ~2 relvariance AD V2 Vr {2 Ve T / Health characteristics (Magnitude items): Conditions . . . . LL. ee Households 2.850 .00064 .00024 .00039 .384 Chronic conditions for females . . . . .................. Females .848 .00117 .00068 .00048 .586 Chronic conditionsformales . . . . ................... Males .667 .00093 .00029 .00064 312 Chronic conditions with 1 or more bed days in last 12ZMODIS « « vvis cwwis vwws = vw FF REBEL TE FETE Persons 132 .00142 .00064 .00077 455 Restricted activity days for chronic conditions in last DWEEKS wx sv mis rE mE rE HR RARE AES TEA RE Chronic conditions .563 .01089 .00839 .00250 770 Restricted activity days in last2weeks . . . . . . ............ Persons 610 .00314 .00204 .00110 649 ACB CONDIIONS ou 2 vm ws smi # FE Ws § SMa 8.408 Eonims » Persons .120 .00054 -.00012 .00066 | -.230 Chronic conditions with 1 or more bed days in last WOKS «i ih rm re Ee aa rae me Persons .028 .01029 .00734 .00295 3 Disability days in last 2 weeks from all accidents . . . .......... Accidents 1.022 .00680 .00353 .00327 .519 Restricted activity days in last 2 weeks for acute conditions . . . LLL. eee ee ee ee eee Acute conditions 2.812 .00118 .00015 .00103 126 Restricted activity days in last 2 weeks for acute CONtIONSFOrMABS + ov vn # smn o sworn swore sas 3 Acute conditions for males 2.716 .00232 .00042 .00190 .180 Bed days for chronic conditions in last 2weeks . . . .. ........ Persons A712 .01242 .00552 .00690 445 Bedidays inlest2wesks . . . .... pumas sos nes amas Seah § Persons .263 .00245 .00065 .00180 .265 Bed days for chronic conditions in last 12months . . . . . .. ..... Chronic conditions 4.677 .00273 —-.00006 .00279 | -.021 Days lost from school or work in last 2 weeks ~~. . . . . ........ Persons currently employed and persons aged 6-16 years .231 .00219 .00005 .00214 .021 Days lost from work in last 2weeks . . . . . . . ............ Persons currently employed 217 .00342 .00012 .00330 .036 Bed days for acute conditions in last2 weeks . . . . . .......... Acute conditions 1.286 .00172 .00004 .00168 .023 Days lost from school in last2weeks . . . . . . ............. Persons aged 6-16 years .240 .00712 -.00270 .00981 | -.379 Hospital discharges in last 12months . . . . . . . .. .......... Persons .071 .00102 .00034 .00068 332 Hospitalizations in last 12months . . . . . . . . ............ Households .382 .00061 .00017 .00044 .281 Hospital days inlast12months . . . . . . ... . ............ Hospitalizations in last 12 months 10.845 .00297 .00004 .00292 .014 Hospital days for all discharges in last 12 months . . . . .. .. ..... Hospital discharges in last 12 months 11.384 .00766 .00327 .00438 428 Health characteristics (Attribute items): Persons with 1 or more chronic conditions . . . . ............ Persons .408 .00034 .00016 .00018 475 Persons with 1 or more conditions. . . . . .. ............. Persons .468 .00026 .00011 .00014 .446 Males with 1 or more chronic conditions . . . . . ............ Males .387 .00046 .00017 .00029 .362 Males with 1 or more conditions. . . . . ... ............. Males 444 .00036 .00013 .00023 .354 Persons limited in kind or amount of activities . . . .......... Persons .096 .00211 .00131 .00080 621 Acute conditions medically attended . . . . . . ............. Acute conditions .766 .00015 .00006 .00010 .358 New cases acute respiratory conditions, 1 or more beddays . . . ......... Persons .030 .00272 .00016 .00256 .058 Hospitalizations for tonsillectomy or adenoidectomy, malesinlast12months . . . . . . . .. .. Lo . Hospitalizations for operations, males .149 .01129 -.00173 .01302 | -.153 - Hospitalizations for operations, female genital systeminlast12months . . . . . . .. .. 0... Hospitalizations for operations, females, exclusive of delivery .347 .00235 -.00020 .00255 | -.084 Persons unable to carry on major activity . . . . . . . Lo... Persons .018 .00253 .00003 .00250 .012 Socioeconomic characteristics: Families with income > $5000 . . . . . .. .. «ov vv viv vu wn Families .673 .00035 .00009 .00026 .260 Persons in families with income > $5000 . . . . . ............ Persons .657 .00027 .00001 .00026 .044 Families with income < $2000 . . . . . . . . . . «uv v nun vn Families 115 .00268 .00108 .00160 .403 Persons who are not employed and not keeping house . . . . . .. ... Persons aged 17 years and over .079 .00085 .00023 .00062 .269 EMPIOYEdpeISons. wis » vows o wom now wow we Pea EER Persons aged 17 years and over .560 .00004 .00000 .00004 N22 Persons in families with income < $2000 . . . . ............. Persons .064 .00351 .00078 .00272 .224 Employed females, . . vu + cvs s rnms emmms sms vows am Females aged 17 years and over .324 .00015 -.00011 .00025 | -.721 31 Table 24. Interviewer variance study of estimates of components of relvariance for annual estimates of selected characteristics based on 28 interviewer assignment areas in 10 SMSA's: United States, 1962-63—Con. . Total rel- teat Sampling 02 Numerator Denominator 3 = interviewer : / y . Ratio variance ; relvariance AT (total number of the following) (total numter of the following) AD relvariance £9 Vv: V. 02 Ve T Tr § Nonresponse Items Persons with amount of education unknown . . . . ........... Persons aged 17 years and over .016 .02338 .01950 .00388 .834 Heads of households with amount of education UNKNOWN: & co vis rm 93 SM HE BEDE § FREY § Rk sway Households .019 .02767 .02218 .00549 .802 Families with unknown family income . . . .. ............. Families .056 .01244 .00884 .00360 211 Conditions from accidents, unknown if motor vehicle . . ........ Accidents .008 .08156 -.00321 .08477 | -.039 Conditions from accidents, unknown location . . . . .......... Accidents .004 .18992 .08211 .10781 .432 Demographic characteristics (including health characteristics by age groups): Chronic conditions with 1 or more bed days, persons dgBd B14 YBIS \ : 25 5 supp Eas Bb EEE E YEE Ce Chronic conditions L017 .00401 .00040 .00361 .100 Acute conditions, persons aged 25-44 years . . . . . . . oo. 2... Acute conditions .246 .00109 .00023 .00086 .208 Chronic conditions with 1 or more bed days, persons BCA 2BAL YBOIS i» « ; + vv s ww HE CWE AE se ETE Chronic conditions .052 .00168 .00015 .00153 .091 Acute conditions, persons aged 15-24 years . . . . . .. . .. ..... Acute conditions 123 .00207 -.00053 .00260 | -.257 Females, married, spouse present . . . . . . . + . v0 hee eee as Females, aged 17 years and over .633 .00011 .00002 .00009 .208 Persons Of OthBrraces . . . « ws + vw wis vw sd & sis s was va Persons 011 .06787 .01118 .05669 .165 Chronic conditions with 1 or more bed days, persons BRAUN BYeaIs +, ; ins iwwis RIM EE BF is FT AMEE Chronic conditions .008 .00896 -.00147 .01043 | -.164 Persons OI NGQIrO TEC . 4 : + ua + iv 65 si Wis 8 s EWa ¥ CW WY # ws Persons .118 .01480 .00543 .00937 .367 Acute conditions, persons aged 5-14 years . . . . .. .......... Acute conditions .231 .00134 .00016 .00118 .118 Chronic conditions with 1 or more bed days, persons aged 65 yearsand OVer . . . . . . LL. uo u hee eee Chronic conditions .029 .00279 -.00112 .00391 -.404 Acute conditions, persons aged 65 yearsand over . . . . . ........ Acute conditions .064 .00630 .00076 .00555 120 Males aged 17 years and OVEr . . . . . . . . . ov vv vv vv nova Persons aged 17 years and over .466 .00002 .00000 .00002 .103 Persons aged under 1 year . . . . . . . . . . vv vie ena Persons .020 .00210 .00051 .00159 241 Persons aged 1 yearand Over . . . . . . . . . . « « tuo tee ae Persons .980 .00000 .00000 .00000 .275 Females aged 17 years and OVer . . . . . . . . «+ uv von vv oes Persons aged 17 years and over .634 .00002 .00000 .00002 .152 Acute conditions, persons aged 45-64 years . . . . . . .. ....... Acute conditions A .00252 .00053 .00199 .210 Acute conditions, persons aged under S5years . . . . . ......... Acute conditions .168 .00270 .00025 .00245 .093 PRISONS, wi o wns vow ain » 0 ie 3 4.8 W153 5 of Wi ® wo % ¥ Www ww ww Households 3.231 .00006 -.00001 .00007 | -.224 Chronic conditions with 1 or more bed days, persons B00) 15-2BYOBIS « own + + ws + SWE Fame wwe Nw. ww Chronic conditions .014 .00587 -.00009 .00596 | -.015 Chronic conditions with 1 or more bed days, persons BEd ABBA YEAS , , cvs rvs mE Ee Ewe we we Chronic conditions .052 .00119 -.00004 .00123 | -.037 Persons aged 17yearsandover . . . ...... . cure Persons 657 .00004 -.00000 .00004 | -.024 Persons aged under 17.years . ... «was ca mms cme www row Persons .343 .00015 -.00001 -.00015 | -.038 the between-interviewer variance is highly with records have been made in a number of skewed. For example, only about 5 percent of the Census enumerator pairs produced high estimates of response variance for four or more of six nonresponse items. SUMMARY In addition to the programs described in this report, three other approaches have been employed in efforts to assess the reliability and accuracy of the statistics produced by the HIS. The first is estimates of accuracy through record checks. Comparison of survey responses 32 special studies designed to assess the accuracy of reporting chronic conditions, frequency of hospitalizations, and frequency of visits to doctors.13-18 The samples have been limited to persons whose names appear on designated groups of records, e.g., patrons of the Health Insurance Plan of New York City and patients at designated hospitals in Detroit. The check starts from the records and goes back to a set of interviews. A second approach is cemparison of the statistics of the HIS with statistics from other sources and examination of the internal consis- tency and reasonableness of the HIS statistics. This is a continuing activity. The results, how- ever, have not been published. Third, experimental studies have been designed, in effect, to measure the difference in accuracy between the HIS interviews as con- ducted and other alternative data collection techniques. The criterion of the more, the better has been explicit in this type of study. That is, a procedure that gives higher estimates of morbid- ity, hospitalizations, etc., is regarded as having produced more accurate statistics than the one with which it is being compared. Thus the differences between the estimates obtained by the HIS procedure and the estimates obtained by alternative procedures that give higher esti- mates are regarded as lower bound estimates of the biases of the HIS procedure. None of the methods, singly or in combina- tion, has as yet produced an estimate of the total mean-square error of any HIS statistic. This is a task of formidable proportions that prob- ably has not been accomplished for any statis- tical program. Such an estimate requires not only the assessment of the accuracy of reporting by the respondent but also the assessment of the effect on statistics of such factors as errors of coverage, nonresponse, recording, coding, and other processing errors. The attainable goals of a program of measure- ment of the reliability and accuracy must, for the forseeable future, be regarded as rather modest ones. The chief benefit to be hoped for is that of providing a basis for detecting and correcting shortcomings in the data-collection and data-processing programs. The second goal is the rather vague one of increasing the awareness by the user of the limitations of the statistics. In this way, informed judgments, rather than esti- mates, of the orders of magnitudes of total mean-square errors can be made at least for some of the HIS statistics. It is the purpose of the research on measurement of error to improve the quality of these judgments. REFERENCES INational Health Survey Act, Public Law 652, Chapter 510, 84th Congress, 2d Session, S. 3076. Hansen, Morris H., Hurwitz, William N., and Bershad, Max A.: Measurement errors in censuses and surveys. BullInst. Internat.Statist. 38(2):359-74, 1961. 3Hansen, Morris H., Hurwitz, William N., and Pritzker, Leon: The estimation and interpretation of gross differences and the simple response variance. Contributions to Statistics Presented to Professor P. C. Mahalanobis on the Occasion of His 70th Birthday, pp. 111-36. Calcutta, India. Pergamon Press, 1965. U.S. Bureau of the Census: The current population survey reinterview program, some notes and discussion. Tech. Paper No. 6. Washington. U.S. Government Printing Office, 1963. 50.8. Bureau of the Census: Evaluation and research program of the U.S. censuses of population and housing, 1960: Accuracy of data on population characteristics as measured by reinter- views. Series ER60-No. 4. Washington. U.S. Government Printing Office, 1964. 61.5. Bureau of the Census: Evaluation and research program of the U.S. censuses of population and housing, 1960: Accuracy of data on population characteristics as measured by CPS-census match. Series ER60-No. 5. Washington. U.S. Government Print- ing Office, 1964. 7U.S. Bureau of the Census: Evaluation and research program of the U.S. censuses of population and housing, 1960: The employer record check. Series ER60-No. 6. Washington. U.S. Government Printing Office, 1965. 8U.S. Bureau of the Census: The current population survey reinterview program, January 1961 through December 1966. Tech. Paper No. 19. Washington. U.S. Government Printing Office, 1968. 9 Bailar, Barbara: Recent research in reinterview procedures. J.Am.Statist.A. 63(321):41-63, 1968. Eckler, A. Ross, and Hurwitz, William N.: Response variance and biases in censuses and surveys. Bull.Inst.Internat. Statist. 36(2):12-35, 1958. Hurley, R., Jabine, T., and Larson, D.: Evaluation studies of the 1959 census of agriculture. Proc.Soc.Statist.Sec. Am. Statist. A. Paper presented at 122nd Annual Meeting, 1962, pp. 91-103. U.S. Bureau of the Census: Evaluation and research program of the U.S. censuses of population and housing, 1960: Effects of interviewers and crew leaders. Series ER60-No. 7. Washington. U.S. Government Printing Office, 1968. 13National Center for Health Statistics: Reporting of hospi- talization in the Health Interview Survey. Vital and Health Statistics. PHS Pub. No. 1000-Series 2-No. 6. Public Health Service. Washington. U.S. Government Printing Office, July 1965. 14National Center for Health Statistics: Health interview responses compared with medical records. Vital and Health Statistics. PHS Pub. No. 1000-Series 2-No. 7. Public Health Service. Washington. U.S. Government Printing Office, July 1965. 33 15National Center for Health Statistics: Interview data on chronic conditions compared with information derived from medical records. Vital and Health Statistics. PHS Pub. No. 1000-Series 2-No. 23. Public Health Service. Washington. U.S. Government Printing Office, May 1967. National Center for Health Statistics: Reporting health events in household interviews: Effects of reinforcement, ques- tion length, and reinterviews. Vital and Health Statistics. Series 2-No. 45. DHEW Pub. No. (HSM) 72-1028. Washington. U.S. Government Printing Office, Mar. 1972. 34 17National Center for Health Statistics: Reporting health events in household interviews: Effects of an extensive question- naire and a diary procedure. Vital and Health Statistics. Series 2-No. 49. DHEW Pub. No. (HSM) 72-1049. Washington. U.S. Government Printing Office, Apr. 1972. 18National Center for Health Statistics: Effect of some experimental interviewing techniques on reporting in the Health Interview Survey. Vital and Health Statistics. PHS Pub. No. 1000-Series 2-No. 41. Public Health Service. Washington. U.S. Government Printing Office, May 1971. Error rate = where OS ao & =x APPENDIX | FORMULA FOR COMPUTING ERROR RATE (A+B+C+...+G) X 100 TC+ TA + TH is omissions from table I of question- naire, is omissions from table II of question- naire, is omissions from table A of question- naire, is missed conditions from questions 6-12 multiplied by 4, E F G Cc TA TH is missed conditions from table II multiplied by 4, is missed hospitalizations multiplied by 3, is diagnostic errors (inconsistencies or other omissions) multiplied by 2, is total conditions, is total accidents, and is total hospitalizations. The weights assigned to the types of error reflect the seriousness of the errors. 35 APPENDIX II TIME AND COST MODEL FOR HIS INTERVIEWING GENERAL Interviewer assignments in the HIS are classi- fied as either resident or nonresident assign- ments. Nonresident assignments require the interviewer’s staying away from home one or more nights; resident assignments do not. For convenience, separate models were developed for resident and nonresident assignments. MODELS FOR HIS ASSIGNMENTS The time T required for a resident HIS assignment, listing and/or interviewing, is expressed as T = nt; + (NS; +Sy - Ng) dry For nonresident assignments, + S,dgrsg + Soty + 2dyry where T is total time in minutes for an inter- view assignment; n 1s number of interviewed households; t, is time per completed interview, in- cluding interview waiting, homework, telephone, and time for non- interviews; to is time per segment for listing; A, is average number of visits per inter- view segment; A, is number of days on which travel is required; S; is number of interview segments; 36 So is number of list segments; d, is average distance between segments; dy is average distance from home to a segment; dg is average distance traveled within a segment (including all visits); ry is travel speed between segments (minutes per mile); ro is travel speed from home to seg- ments; and rg is travel speed within segments. Values of the parameters in both models are identical. Some of these values depend on the particular assignment: n, S; and Sy. Other values are estimated from accumulated data and are assumed to be constants in the model: A, dg, ry, 79, 73, tj, to. These estimates are prepared separately for five subuniverses that are defined by degree of urbanization and popula- tion density. The remaining values, except for \,, are functions of the home address of a particular interviewer and the location of the PSU’s where she works. The number of days on which travel is required A, applies to resident assignments only and is a function of other terms in the equation. To use the preceding equations for each interview assignment becomes very cumbersome and time consuming. Some simplification is needed so that the clerical computations can be handled routinely. When the expression for A, is substituted in the resident model and terms are collected, the equation reduces to T = (CA,)S; + (CAy)n + (CA3)Sy where CA is the computation allowance, so that C4, = \d +d + 1 1917 373 280 280 CAy = —— =~ 2 280 280 : 2dorot d,r,t CAgq = 272 2 _ 171 Ltdyr, +tg "280 280 The nonresident model reduces to T = (CA,)S, + (CAg)n + (CA3)Sy + CA where CA; = Ndr; +dgsrg CA, = t; 2dgredgrsg ~ dr dgrg CA, = 2dgy7y — 2dr, . If the subuniverse parameters A, dg, 7, 79, and rg are known (they are actually estimated from data), the computation allowances (CA) are functions of d; and dy only. Since an interviewer may treat a PSU assignment as resident one time and nonresident another, both sets of computation allowances are computed for each interviewer. The d; and d, values are flexible and can be changed when circumstances warrant, such as the interviewer moves or gets a new assignment, a replacement interviewer is hired, etc. Regional offices compute the computation allowances and keep a cumulative record (appen- dix VIII) of assignments for each interviewer. 37 APPENDIX III SOME THEORY OF MEASUREMENT ERRORS® SOME DEFINITIONS The term “survey” is used to refer either to complete censuses or sample surveys. In considering measurement errors we shall regard a survey as being conceptually repeatable, that is, repetitions relate to the same point in time so that carrying through the operation once does not influence results obtained through repeti- tions. The particular data obtained in a survey are the result of one trial. This concept provides the basis for defining variance and bias due to response, processing, or other sources of meas- urement errors. Such a postulate can reasonably approximate actual conditions for a single sur- vey regarded as a sample of one from such a set of surveys, even though in practice independent repetitions of a survey may be impracticable or impossible. THE DESIRED MEASURE OR TRUE VALUE We conceive of some desired measure or goal to be estimated from a survey. For simplicity, the assumption is made that the desired or true value to be measured is represented as a propor- tion of the population having a specified charac- teristic. Although ordinarily there will be many such values to be estimated from a survey, one will be considered. Thus, it is assumed that the population consists of N persons, each of whom can be regarded as having the value of 1 if the person has one or more chronic conditions (or has some specified characteristic) or as having the value of 0 if the person does not have one or more chronic conditions (or does not have the specified characteristic). The desired or true CThe discussion in this appendix is based on material in references 2 and 4. 38 proportion of persons having the characteristic is said to be estimated, even for a complete survey of the population under consideration, because only observations or responses, which are sub- ject to errors, can be recorded. THE GENERAL CONDITIONS THAT MAY AFFECT THE RESULTS OF A SURVEY Measurement errors have many different causes and depend on the general conditions under which a survey is taken. Some of these general conditions may be beyond the control or specification of the survey designer as, for example, the general political, economic, and social situation at the time of the survey. Uncontrolled conditions also include many temporary chance situations appearing at the time a response is obtained. Some conditions can be controlled to influence the quality of survey results in the sense that various aspects of the conduct of the survey are specified. These specifications are typically made in the effort to insure adequate quality and include question- naire design and survey procedures, personnel qualifications, pay system and rates, training, operating methods, inspection, and controls in the survey. Such conditions, which may be only partially subject to the sponsor’s control, are usually indicated in the form of fixed rules under which the survey is to be taken. Other controllable conditions that may be varied by design, or may be regarded as varying between the conceived repetitions of the survey, are the particular choice of interviewers and other per- sonnel chosen to do various aspects of the work, the specific assignments, and other similar variable factors. Actually, a survey sponsor is unable to specify all of the factors, controlled or uncontrolled, that may affect the survey results. He can introduce certain chance factors explicitly or implicitly, he can impose certain specifications or conditions, but he must accept the effects of other uncontrolled factors. AN ESTIMATE FROM A SURVEY (OR TRIAL) TAKEN UNDER A SET OF GENERAL CONDITIONS For simplicity, it is assumed that the survey is either a complete census or a sample in which all units have been given an equal probability of selection, but without, at this time, any other restriction on the sample design. In accordance with the definitions in the previous sections, this particular survey is regarded as one trial, i.e., one survey from among the possible repetitions of the survey under the same general conditions. An observation on a person or other unit in the survey has the value 1 if the unit is assigned to a particular class under consideration or the value 0 otherwise. A repetition of the survey on the same or different units would constitute a second trial. The general conditions, both con- trolled and uncontrolled, under which the sam- ple has been taken will have an effect on the observations made in a trial. In repetitions of the survey it is assumed that the response actually observed for any individual in the survey can be regarded as having been drawn by a random process from the possible answers he might have given under the same general conditions. (In practice a survey cannot be repeated independently under the same general conditions because respondents have been exposed to the original survey and because of other reasons. However, the initial survey can be properly regarded as a sample of one from a set of independent replications.) Thus, we are dealing with a random variable x;,, whose value is 1, if element j is classified as having some characteristic on trial ¢ of a survey (x may denote the class “one or more chronic condi- tions’’) or 0, if j is not so classified. The estimate obtained from a survey, i.e., a trial, is the proportion classified as having the specified characteristic in trial ¢ of a survey of n elements: ] n Pt =—2 Xt ] For example, in measuring the proportion of persons with one or more chronic conditions, x=1 in a particular survey (or trial) if the person is classified as having one or more chronic conditions; otherwise x = 0. Then is the estimated proportion of persons with one or more chronic conditions. THE MEAN SQUARE ERROR OF AN ESTIMATE FROM A SURVEY (OR TRIAL) Continuing the illustration, p, is the survey estimate of the proportion of the population with one or more chronic conditions and U is the true proportion in the population. While we generally cannot determine in practice the true value U; for any person, we can postulate that the goal of the survey is the true value of the proportion of the population having one or more chronic conditions. Thus, U, the desired true proportion, is estimated by the statistics actually obtained in the survey where SE b=3 2 4 0 For p,, the mean square error (MSE) is MSE(p,) = E(p, - D)? (2) where the expected value is taken over all trials. The mean square error can be divided into its two main components: MSE(p,) = Ep, - P2 + (P- 0)? (3) where P is the average of the estimates p, taken over all trials and over all possible samples. The first term in equation 3 is the total variance of p, and the second term is the square of the bias of p,. In practice, we are not able to measure the bias, P- U, but sometimes we can define and estimate useful approximations to it. For exam- ple, a superior procedure or measurement may 39 be identified as a standard. If s represents the value obtained from such a superior measure- ment, then (p, - s) may be used as an approxi- mate estimate of bias. The total variance of p, can be divided into the response variance and the sampling variance: MSE(p,) = Response variance + sampling variance + interaction + square of bias. Expressing the response variance of p, in terms of response deviations where dj; = (x;, = P;) is the deviation of the response recorded for person j on trial ¢ from the average value of the responses for person j over all trials, the response variance can be expressed as 2 _ 9 03 = “21+ pgpq,n- 1) where 02 4» the simple response variance, is the basic trial-to-trial variability in response averaged over all persons. The correlation term pg;,q4y, reflects the effect of correlated errors intro- duced into the survey process by interviewers, supervisors, coders, and by persons engaged in other operations. If the intraclass correlation among response deviations i i zero, the total response variance of 2 is 1/n 0%. On the other hand if the product (n-1)p is Ay the total response variance may be large even if the simple response variance is relatively small. Thus, this model of errors in surveys permits the partitioning of the MSE (p,) into a set of components. These various compo- nents may be estimated by means of special surveys and experiments. GROSS AND NET DIFFERENCES In comparing the case-by-case results of two sets of measurements, the total number of differences affecting the tabulated figure for any given class of a population is equal to the number of cases included in that class in the first trial but not in the second trial, plus the number of those included in that class in the second trial but not in the first trial. This sum may be termed the gross difference for the population in question. 40 The net difference of the tabulated figure for the given class is the difference between the total for the class obtained in the first and the second trials. Usually the gross difference will include differences in both directions, partly or substantially offsetting, and the net difference is the nonoffsetting part of the gross difference. For example, suppose that the survey identi- fies each person as having or not having one or more chronic conditions, and that a total of n persons have been sampled with equal probability and included in both a first and second trial. Table I shows that a of the individuals were classified as having one or more chronic condi- tions in both the first and second trials, a + ¢ were classified as having one or more chronic conditions in the first trial, and a +b in the second trial. The gross difference in the classifi- cation is b+c¢ and the net difference is (@atc)-(at+b)=c-b. Now let x;, represent the result recorded for a particular person in the first trial and Xj! the observation recorded for that same person in the second trial. Furthermore, x;, is assigned the value 1 if the person is recorded as having a particular characteristic, and 0 otherwise, and similarly for x;,+. Then the response difference for a particular Pann in the two surveys is represented as €; = Xj; = Xj. The sum of 1 the values of e; over the n= observations is the net difference between the two results. n oe Table |. General representation of results of two sets of measure- ments on identical persons Results of Results of first trial second trial xe = , xe ™ 0 Tom Xj = 2 b ath je = 0 ¢ d ct+d Total atc b+d n=g bh vo bd and n c-b n é = ¢ _ n is the net difference rate. Similarly, b + c¢ is the gross difference and is n equal to) eZ. This follows since 2 = 1 when- ever the response obtained in the first and second trials are different, that is, (0- 1)2 = 1. The gross difference rate is _ (b +c) n g GROSS AND NET DIFFERENCES AS EVIDENCE OF RESPONSE VARIANCE AND BIAS The estimated variance of the individual response difference is S(e-2? bc (cob) e n-1 Tn-1 an-1) (g - 2?) n n-1 where € = n de c-b n n Often 22 is small enough that s? will be very nearly equal to g, and it is then convenient to use the gross difference rate g as the measure of the variance of the response differences. In any event, g is the mean squared difference for the original and reinterview survey results and pro- vides a useful measure of the consistency or reliability of the measurement process. It can be an exceedingly useful measure of reliability of response with a well-designed evaluation study or reinterview survey. If the individual response differences were uncorrelated from one unit to another, the estimated standard error of the net difference rate would be [2 so Se sg = |—- n In practice, the individual response differ- ences will not be independent from one unit to another but will tend to be positively correlated. Under these circumstances, \/s2 /n gives a lower bound for s;. Given certain conditions, an overestimate of s; can also be obtained. These conditions, would be met if, for example, a survey is repeated over time or over different areas or population groups and if the reinterview survey is conducted on different units in each of these repetitions of the survey. This is the situation for the HIS, which is taken each week. The HIS reinterview survey is taken on a distinct set of HIS households each week and sum- marized quarterly. From the group of quarterly repetitions an overestimate can be obtained of the standard error of the average net difference. (It will be an overestimate of the standard error of the difference obtained by repetitions of the two surveys taken for different samples but with the same personnel.) Thus, from the HIS reinter- view survey, results of both net differences and net difference rates are obtained for each m quarters. If 2, is the net difference rate for quarter u, and n, is the number of persons in the reinterview sample in that quarter, then for the m quarters involved, the average net differ- ence rate can be expressed as m pA n e where n = Zn, is the total number of reinter- views over the m quarters. Then 5 n, (2, =2)2 n(m - 1) will be an overstatement of s;, the estimated standard error of the average difference rate. 41 Thus an overestimate and an underestimate of sz can be obtained. If these are not too different, they yield a measure of the standard error of the average net difference between the original and reinterview survey results. INDEX OF INCONSISTENCY An index of the reliability of measurement, called the index of inconsistency, can be con- structed using the gross difference rate. The index of inconsistency is the ratio of the simple response variance, estimated by g/2, to the maximum value it could take on, estimated by the binomial variance p(1 - p). In terms of the table, let p; = (a + ¢)/n; that is, p is the proportion, based on the original survey, of the population in the specified class and po = (a+ b)/n is the proportion based on the reinterview survey. Then, -_.& . p1(1-py) +pa(l-p9) aD The estimated maximum value for the gross difference rate between the survey and reinter- view is p; (1 - p) + po (1 - pg). This maximum value is obtained on the assumption that the survey and reinterview were conducted inde- pendently or that the results are positively 42 correlated to the extent that they were not conducted independently. A second assumption is that the reinterview is a repetition of the survey process and the expected value obtained in the survey. Under these assumptions, p1(1- py) +po(1- py) is very nearly equal to 2p(1 - p) where p is the average proportion in the original survey and reinterview having the specified characteristic. The index of inconsistency lies between 0 and 1 if the assumptions given above hold. However, the estimator can be greater than 1. A simple interpretation of I follows. Assume that a sample of n elements is drawn with equal probability and with replacement. Also, assume that the between-element covariance of response deviations is zero; that is, that the quality of response of one person is independent of the quality of response for any other person. Then the total variance defined in the first term of equation 3 of the statistic p, reduces to the sum of the simple response variance and the single random sampling variance. The simple response variance is equal to or less than p(1 - p). As the measurement of the specified charac- teristic becomes less reliable, but remains un- biased, the simple response variance increases and the sampling variance decreases; the total variance remains constant. A high index of inconsistency is associated with a high level of response error. APPENDIX IV LEAST-SQUARES SOLUTION The variance model used in the study is 2 _ po, VE Vs = Vi Bois (1) In this model, the interviewer variance V2 is assumed to be dependent only on the number of interviewers, which remained constant over the study period. On the other hand, the sampling variance term V2 /n is a function of sample size. In equation 1, n is the number of quarters of data used to compute the individual estimates of total variance. The data used are sets of estimates v2 for n=1,...,8. Each V2 is an average of the estimates based on n quarters. The V2 are estimates of V3 for n=1,...,8. The model given above represents the functional relation- ship existing between the variables V2, v3, and Va as n varies. The problem, then, is to estimate the parameters of this function so that the estimates of V2 and vi can be identified separately. There are several ways of estimating these parameters. The goal is to estimate Va so that the estimates of VZ | are as close as possible to the observed values PR. The method used was that of least squares. The criterion for the least-squares estimate is to make the sum of the squares of the difference between the estimate of 5 ”) and the observed values V2 as small as sti that is, to minimize the value of ¢, where 8 ¢ = > Vim)" V2)2. (2) n=1 These differences can be seen graphically as follows: wi 1 Q \ = A oc \ = 3 ae IN, — £1 IN o “ oS = — _ 5 ny Fa 2 I» “eel 5 iE ! Al HT 4 = v2 z wv Ww 1 | | l l | 1 [ 1 2 3 4 5 6 7 8 MEASURE OF SAMPLE SIZE (NUMBER OF QUARTERS) As required by the model, a hyperbolic func- tion, rather than a straight line, was fitted to the data. No boundaries were placed on the value of 2 ; it can become negative as n = oo, Using the method of least squares, the partial derivatives of equation 1 are taken V2 = _L 4 _E, 3 3 «fered (3) If 8 $= 2 1 TP Vi)? n=1 8 [2 12 2 = 3 (-L+ E72 (4) n=1 1 n ") then do 8 [02 p22 _ = 2 abt l= PA}(1) = 0 ar: 25 == V2) (1) (5) 43 and A 1 a SE VIZ —+ Vid 3 — = I — 3 I — Since and 44 . . ‘ No 2 the two partial derivatives Vi and V2 can be simplified to V2 2718 op td 5 pe 1%; Ao _ 2 8 z n Ve = (2.718)2 LETT = damned 8 = 1.655 vi 0.562 » V2 (9 = 1.655 3-0. 2 (9) and ny 2 VE 2.718 5, Vi = 8 - 8 Vi = 0.125 3. 72 - 0.340 V2. (10) By substituting in the value for P2 , D2 = 72 pe Vv? = 0.316 Y V2 - 0.562 yr (11) These estimates of if and Pz for one quarter were used to compute the values of V2 for one to eight quarters. The values shown in tables 23 and 24 are estimates for a sample covering four quarters. APPENDIX V HIS OBSERVATION REPORT (NHS-HIS-406) FORM HIS-406 U.S. DEPARTMENT OF COMMERCE 1. Regional Office 2. PSU HEALTH INTERVIEW SURVEY 3. Name of interviewer Code OBSERVATION REPORT 4. Date observed 5. Time observed From 6. Date of last observation 7. Type of observation [] Systematic [J Reinterview rejection [J] Initial — First assignment [] Initial — Second assignment [] Other — Specify — 8. Reference notes for special attention — e.g., reinterview results, notes from last observation, etc. Error rate Type A Production rate | | I | [ 1 9. Segment coverage Tally of Interviewed Type A Type B Segment househ olds or callback or number observed and type TA NTA Item Explain each **No” answer below Yes No N/A Yes No N/A|Yes | No N/A Did interviewer correctly — Use maps, locate segments, locate sample addresses? Check area segment boundaries? Canvass area segments and look for concealed units? Fill area segment listing sheets? Determine ‘“‘year built” when required? mM mon om > Fill Cols. 8, 9, and 13 of B segment address lists? G Fill extension sheets in B segment? 10. GENERAL PERFORMANCE Evaluate each point for the entire day’s observation 1 Asking probe questions when needed and only when needed lent Excel- Satis- factory Explain below Not Needs im-| Unsatis- applicable provement | factory . Neutral probing . Allowing respondent reasonable time for answering questions . Maintaining a business-like attitude and rapport with respondent . Listening carefully to respondent . Accurately recording respondent’s answers and completing all required entries on questionnaire (Evalute this item on the basis of respondent’s answers and your edit of the questionnaires after the interview.) . Accepting suggestions and criticisms . Applying housing unit definition 9 . Listing and interviewing within special dwelling places 10. Planning itinerary ASK INTERVIEWER DURING THE DAY: Is there any particular part of the procedure you feel unsure about or would like to have covered by further training? Remarks 45 Last name Person N Randied y If “No,"’ Name explain Age PROBE Person | Ques- PAGES EONDI- Person Condition Yes | No | Person Condition PAGE(S) HOSPI- P TAL erson | Page PAGE(S) 1 2 3 DOCTOR Yes | No | suPPLE- | Person VISITS MENT [None [J None PERSON PAGE(S) Ques- (Esp. y bul and coverage items) 5 6 COMMENTS (including edit) 46 RESULTS OF OBSERVATION Overall evaluation on all phases of work [1 Excellent [] Satisfactory [] Needs improvement [] Unsatisfactory Comments on general performance Recommendations for next observation OBSERVER: Note any area of the questionnaire or interviewer's instructions which in your view require modification or clarification. Observer’s signature Date To be completed if observer recommends termination, probation, or other administrative action Comments of the Regional Director Regional Director’s signature Date 47 APPENDIX VI HIS RECONCILIATION FORM (NHS-HIS-R-1X-T) A. RECONCILIATION SECTION FOR BED DAYS, B. RECONCILIATION SECTION FOR CONDITIONS AND HOSPITALIZATIONS REPORTED IN ONE INTERVIEW RESTRICTED ACTIVITY DAYS, AND TIME LOST [We are interested in finding out more about conditions (hospitalizations) which were reported at one time but not at DAYS IN PAST TWO WEEKS : Condition - . Date of hospitalizatios | Condition Date of hospitalization The original questionnaire showed . .. and | have . . . § which is the correct information? Original Reconciliation Explanation \|Expianation Sb. Bed days Bed days : [CJ None [J None Se. Cut down days Cut down days [J None [] None , Sf. Work days Work days [] None [] None 5g. School days School days [J None [7] Nong [CT] Sustained [Deleted i| [OJ sustained [Deleted C. RECONCILIATION SECTION FOR DIFFERENCES WITHIN MATCHED CONDITIONS — The original questibnnaire showed . . . and | have . . ., which is the correct i 1 2 + 3 2 ’ Match |Reinterview condition No. |1. Person No. [Match |Reinterview condition No. [1. Person No. [Match [Reinterview condition No. [1. Person No. [Match |Reinterview condition No. | 1. Person No. | Original Reconciliation Original Reconciliation Original Reconciliation Original Reconciliation 2. 2, 2. 2, [J Yes [CINo [C] Yes [No [Yes [JNo [CJ Yes CJNo [J Yes [CINo [] Yes Neo [J Yes [CINo [C] Yes CIN 3a. 30. 3a. . 3a. | 3b. 3b. 3b. : 3b. 1‘ 3c. Je. 3c. 3e. 3d. 3d. 3d. Y 3d. | Je. Je. Je. Je. Original Reconciliation Original Reconciliation Original R iliation Original Reconciliation 90. [JYes [JNo [J Yes [JNo 9a. [Yes [No [JYes [JNo 90. [Yes [No [Yes [No 90. [JYes [No [JYes [No 9b. [JYes [JNo [JYes [No 9b. [JYes [_JNo [CJYes [No 9b. [JYes [JNo [JYes [JNo 9b. [JYes [No [JYes [No 10. Days Days 10. Days Days 10. Days Das 10. Days Days 11. Days 1 Days | 11. Days 1 Days | 11. Days 1 Dats | 1. Days i Days | : [C) None | [) None ! ["] None | [J None | [1 None { [) None ! [C] None 1 [] None 12. Days [| Days | 12. Days | Days! 12. Days | Days | 12. Days | Days | ! [J None | ("None ! [J None { [J None ! [] None 1 [J None ! [J None 1 [None 13. Days Days | 13. Days 1 Days 13. Days Dears 1 13. Days Days | ! [J None ! (None ! [J None | [J None ! [None ! [J] None ! [None | [J None 14a. : ; Vda. : : T4a . 14a. : D f Di Bef D Bef D Bef ht Bef, Dori Bef £5 Quring () Before) [During [1] Before | C1 During [J Refers) [IDuring [1] fefore | ™* [J During C3 Beton] Clb enns C3 Before | £3 uring [3 Before) [Roving [] Before Mb, ub. 1p Bef P Before | 4% [] Past Bef, P Before "4b [J Past Bef P Bef, Olga, Oil Oa, Ofeien | ™ Ofer, Ofeler) Ofer, Ofer Oa, Oeie| Oat, Ofer | Olay, Ofeler) Opag, Opens Tae. Week Week | 14e Week Last Week |'4¢ [Last Week L Week |'€ [JLast Week L Week Obey Ofon | Clay Dfot |= Cle Ofek | Cloay Cfo |= Obey Clie | Clay OF [7° Diag Clot Clas Ove, V8 [3-12 [)Before| [3-12 [] Before | [J 3-12 [) Before] []3-12 [J Before |'> []3-12 [] Before] []3- 12 [] Before ['5* []3-12 [7] Before] [3-12 [] Before mos. 12mos.| mos. 12 mos. mos. 12mos4 mos. 12 mos. mos. 12mos. mos. 12 mos. mos. 12mos.| mos. 12 mos 20. (Yes [No COYes [No 20. [JYes [No [JYes [No 20. [JYes [JNo TJYes [No 20. [JYes [JNo [Yes [No 21. [Yes [Jiwo |[JYes [No 21. [JYes [JNo |[[JYes [JNo |21. [JYes [No ven I Ne 21. [JYes [JNo |[JYes [JNo 22. [Yes [No [CJYes []No 22. [JYes [No [JYes [No 22. [JYes [No C Yes J No 22. [JYes [JNo _JYes [No 23. Times | Times | 23. Times | Times | 23. Times | Tines | 23. Times | Times | i [J None { (J) None 1 [J None 1 (C] Noae 1 [CJ None { CJ None [J None { CJ None 24. Days | Days | 24. Days | Days | 24. Days | Days | 24. Days | Days | ! [CJ None : [C1 None 1 [J None y [] None 1 [C) None . : [_) None ! [J None : [J None 25a. [Great deal Great deal 25a. Great deal Great deal 25a. Great deal Great deal 250. Great deal Great deal Some Some Some Some me Some me me Very little Very little Very litle Very little Very little Very litle Very little Very little Not at all Not at all Not at all Not at all Not at all Not at ali Not at all Not at all Other - Specity— Other - Specity— Other Specity~ Other - Specify Other - Specity— Other - Specity= Other - Specity— Other - Specily— 256. [JYes [No [Yes [No 25b. [(JYes [No [JYes [No 25h. [Yes [JNo yes [No 25b. [JYes [No [JYes [No 25c. Cured Cured 25e. Cured Cured 2c. Cured Cured 25c. Cured Cured Under control Under control Under control Under control Under control Under control Under control Under control ! Other - Specity Other - Specity 3 Other - Specity gy Other - Specily Other - Specity Other - Specify Other - Specilyy Other - Specity 25d. Month(s); Year(s) Month(s) | Year(s) 25d. Month(s) | Year(s) Month(s) | Year(s) 25d. Month(s) | Year(s) Month(s) | Year(s) 25d. Month(s) | Year(s) Month(s) | Year(s) ' 1 ' ' ' ‘ ' 1 1 1 Ll 1 FOIM NHS-HIS-R-1X-T (8-8-87) BUT NOT IN BOTH INTERVIEWS another. Can you think of any explanation for . . . not having been reported in the (original interview/our interview today)? Condition Date of hospitalization | Condition Date of hospitalization | Condition Date of hospitalization Explanation Explanation Explanation [CT] Sustained [J Deleted [) Sustained (] Deleted [J Sustained [7] Deleted nformation? 6 7 8 Match |Reinterview condition No. [1. Person No. |Match |Reinterview condition No. [1. Person No. [Match |Reinterview condition No. [1. Person No. [Match [Reinterview condition No. [1. Person No. Original Reconciliation Original Reconciliation Original Reconciliation Original Reconciliation 2 2. 2, 2 ClYes [ON | CiYes [Ne | ClYes [No | [Yes [No | [iYes [ON | Yes [Ne | O¥es [IN | CIYes [Ne 3a. 3a. # 3a. 3a. 3b. 3b. 3b. 3b. 3c. 3c. 3c. 3e. 3d. 3d. 3d. 3d. Je. Je. Je. Je. Original Reconciliation Original Reconciliation Original Reconciliation Original Reconciliation 90. [JYes [No [[JYes [JNo 9a. [JYes [No [Yes [JNo 9a. [JYes [No [JYes [No 9a. [JYes [JNo [JYes [No 9b. [JYes [JNo | [Yes [No 9b. [JYes [JNo |[TJYes [No 9b. [JYes [JNo |[[JYes [No 95. [JYes [No |[JYes [No 10. Days Days 10. Days : Days 10. Days Days 10. Days Days 11. Days 1 Days | 11. Days 1 Days | 1. Days 1 Days | 11. Days | Days | | [None | [J None i CINoae | [J None | [CJ None ! [J None ! [] None | CJ None 12. Days ] Days | 12. Days) ! oo Days! 12. Days ! Days | 12. Days ! Days | None None None | None None None None \ one y CL} OJ | on 0 ‘0 102) v LJ 1 CIN 13. Days Days 13. Days Days | 13. Days 1 Days | 13. Days Days | | [CJNone { None ! [J None { CJ None ! [J None { CJ None ! [None { None 4a. ; 7 T4a. : 4a. Tdo. : : Durin Before| []During [] Before Durin Before| []Durin Bef o Duri Bel, Duri Bef D Bef D: Bel Ce DV ge] Clue pein | = Rung Cl eters Euro 0 tone | C2) Juste C0) gelond Curing I) Yelons | ™> £1 uring 0 Jotond] (Curing [0 Bele 4b. T4b. T4b. 14b. Past Before, Past Before Past Before Past Bef Past Bef: Past Bef Past Bef. Cf CI Peturet Cpa OO Jospre Ol Pome. Eltsrel OPane, Cl fetore CI fan, DO helore] DPag, OD fefere Cfo, DO gelere , Tae. Tde. Tde. Vdc. Last Week Last Week Last Week Last Week Last Week Last Week Last Week Li Week [as CVoon | Oley Oifed, 1™ Cit Ofst Oley Of ™ Diy Of, Oty Opp, I Ope, Opt | Obes Oss, 1s. 15. 15. 15. 3-12 Before] []3- 12 Before 3-12 Bef, 3-12 Bef 3-12 Be 3. 1 Bel, 3-12 Bel - 12 [J Bel, (2.72 Dp Dl Onl ™ D202 pl O02 Digsl™ O22 Dijged Dak Ogg [> 033-2 Digs 073-2 Cl fglme 20. [JYes [JNo- |[]Yes [JNo 20. [JYes [No [Yes [Neo 20. [JYes [No [JYes [No 20. [JYes [JNo [Yes []No 21. [JYes [No JYes [No 21. [CJYes [No [JYes [Ne 21. [JYes [JNo (COJYes [No 21. [JYes [JNo [JYes [No 22. [JYes [JNo [JYes []No 22. (JYes [No [JYes [J No 22. [JYes [No [CJYes [No 22. [JYes [No [JYes [J No 23. Times | Times | 23. Times | Times 1 23. Times | Times | 23, Timea | Times 1 1 [None { CJ None [J None ! J None , [None { CJ None 1 [J None | CC None 24. Days | Days | 24. Days | Days | 24. Days | Days | 24. Days | Days | + [CJ None ) [CJ None { [J None } [CJ None | [J None y [J None . i [J None y (J None 25a. Great deal Great deal 25a. [ | Great ded ] Great deal 25a. Great deal Great deal 25a. Great deal Great deal me Some Some Some Some Some Some Some Very little Very little Very liul: .__ Very little Very little Very little Very little Very little Not at all Not at all Not at all __ Not at all Not at all Not at all Not at all Not at all Other - Specity— Other - Specify [_] Other - Syesitymy = Other - Specity— Other = Sectly=p Other - Spectlyp Other = Specityy Other - Spesity— 25h. [_JYes []No [JYes [No 25b. Yes [JNo [JYes [No 25b. [JYes [No [JYes [JNo 25b. [JYes [JNo [JYes [Ne 2c. Cared Cured 25c. | Cured Cured 25¢. [Cured Cured 25¢. [J] Cured [J Cured Under control Under control =] Under cortrol Under control Under control Under control Under control Under contra] J Other - Specity Other - Specity TZ) Other - specity3 | [J Other - Specity 2 Other - Specify Other - Specify Other - Specity Other - wz] 25d. Month(s), Year(s) Month(s) | Year(s) 25d. Month(s) | Yeor(s) . Month(s) 1 Year(s) 25d. Month(s) | Year(s) Month(s) | Year(s) 25d. Month(s) | Yepr(s) Month(s) | Year(s) ' ! ' | } ' ! ' ! L . ! USCON MDC 49 APPENDIX VII SUMMARY REPORT OF NHS-HIS REINTERVIEW (NHS-HIS-R-401) Form NHS-HIS-R-401 U.S. DEPARTMENT OF COMMERCH 1: Interviewer’s name {Code | Telephone No. 2. Regional Office (8-85-66) BUREAU OF THE CENSUS] | | 1 l 3. Reinterviewer’s name | Code | Job title 4. PSU | | Program Supervisor . SUMMARY REPORT OF } Lf pass NHS-HIS REINTERVIEW 5. Reinterview date [] Senior Interviewer |6. Sample [] Other Section | - COVERAGE CHECK OF HOUSING AND OTHER UNITS Section II HOUSEHOLD COMPOSITION CHECK | Part A — Area Segments Part B Part C Number of persons Reinterview segment Number of units B Segment Check Number Segment Household | Before Number of i Extension sheet | ©f wrong No. serial No. |reinter- | Added | Deleted sample units| Listed ’ house- view before before eniriey holds Mm ©) ©) (4) (5) Number | Type renter reinter- Added Deleted inter= view vien 1 Correct | Incorrect | viewed (1) (2) 3) (4) (5) (6) (I) (8) 9 Total this printer revious cumulative total New cumulative total 1 Exclude units in special dwelling places in NTA segments and in ‘‘large’’ special dwelling places. If the most recent listing for an NTA segment was performed by another person, enter name in ““Explanation of Differences’’ and prepare separate form NHS-HIS-R-401 reporting columns (4), (5), and (6) data for the segment. Explanation of differences in sections | and || (Give reference to section and segment, numbers of added or deleted units, and to segment and serial numbers of added or deleted persons. Describe type of error if column (8) is checked. Explain changes in classification of ‘‘year built’’ here also.) Total this reinterview Previous cumulative total New cumulative total 50 Section Ill - CONTENT CHECK Part A = Personal Characteristics Part B — Characteristics Within Condi- Part C — Number of Conditions tions and Hospitalizations ond Hospitalizations “Different Same Different Same Different Same Segment Household respondent respondent! respondent respondent ! respondent respondent 1 Ne. saris). Ne. Number | Differ. | Number | pifer. [Number off Differ. [Number off pitfer. pe pipfer, Aoi Ditters o o ences persons ences persons ences | pecks * | NCS checks* | ©"°°S base? | "°° base 2 (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (n (12) (13) (14) Total this interview Previous cumulative total New cumulative total 1 Includes adults responding for children under 19. 2 (v) check conditions and hospitalizations plus sustained conditions and hospitalizations plus number of conditions and hospitalizations added and sustained to reconciliation form. 3 Number of original conditions and hospitalizations deleted from reconciliation form plus number of conditions and hospitalizations added and sustained to reconciliation form. * Definitions (v’) checks: Each check represents one condition or hospitalization reported on both interviews. Sustained conditions or hospitalization are those which are reported only on one interview and retained after reconciliation. Section IV — ACTION TAKEN TE EE Note: This section must be filled if the interviewer has been rejected in any of the preceding sections. Explain action taken | Decision lor planned to retrain interviewer if her work has been rejected (R). (If more space is required, use additional sheet.) Section A - fon (1) (2) I-A 1-C Il Hl - A I -B I -C USCOMM-DC 51 APPENDIX VIII PRODUCTION GUIDE FOR NHS (11-102C) AL OFFICE SAMPLE B- PRODUCTION GUIDE FOR NHS INTERVIEWER 4. INTERVIEWER CODE INTERVIEW HOUSEHOLDS SEGMENTS INTERVIEWED LIST SEGMENTS SUPPLEMENT THIS ASSIGNMENT CUMULATIVE PRODUC- TIO cA ca ca AL ALLOWED [PAYROLL | ALLOWED | PAYROLL NuMBER \ NuMBER 2 NUMBER 3 NUMBER | \y/NUTES | MINUTES | MINUTES | MINUTES | MINUTES (d) (e) f) (h) RATIO CHIEF, FIELD DIVISION COPY USCOMM-DC 21780 P-63 52 FORM 11-102Cc (10-7-63) INSTRUCTIONS FOR COMPLETING FORM 11-102C GENERAL Maintain a Form 11-102C for each interviewer ona three month (NHS Sample) basis. At the end of the three month period send the yellow copy to Chief, Field Division, retain the original in your files and start a new 11-102C for each interviewer. One line of the 11-102C should be completed for each inter- view and/or listing assignment. Columns (a)=(p) Enter the week in column (a) and the PSUin column (b). Post the computation allowances to columns (d), (f), (h), and (i), using Production Standards Memo- randum No. 5 (Formerly GAM No. 70) and your knowledge (from payroll records) of whether the assignment was overnight or non-overnight. Enter the workload associated with each allowance in columns (c), (e), and (g). Enter the number of current supplements (if any) completed in column (j), and the allowance per unit, as given in Operations Memorandums, in column (k). Multiply each allowance by its workload and add the results to obtain the total allowance. (Be sure to add in CA 4.) Enter this total in column (1). Enter the payroll minutes in column (m). Revise the cumulative production ratio in columns (n), (0), and (p). # U. S. GOVERNMENT PRINTING OFFICE : 1973 515-212/48 USCOMM-DC 21759 P-63 53 Series 1. Series 2. Series 3. Series 4. Series 10. Series 11. Series 12. Series 13. Series 14. Series 20. Series 21. Series 22, VITAL AND HEALTH STATISTICS PUBLICATION SERIES Originally Public Health Service Publication No. 1000 Programs and collection procedures.— Reports which describe the general programs of the National Center for Health Statistics and its offices and divisions, data collection methods used, definitions, and other material necessary for understanding the data, Data evaluation and methods veseavch.— Studies of new statistical methodology including: experi- mental tests of new survey methods, studies of vital statistics collection methods, new analytical techniques, objective evaluations of reliability of collected data, contributions to statistical theory. Analytical studies.—Reports presenting analytical or interpretive studies based on vital and health statistics, carrying the analysis further than the expository types of reports in the other series, Documents and committee veports,—Final reports of major committees concerned with vital and health statistics, and documents such as recommended model vital registration laws and revised birth and death certificates. Data from the Health Interview Suvvev.—Statistics on illness, accidental injuries, disability, use of hospital, medical, dental, and other services, and other health-related topics, based on data collected in a continuing national household interview survey. Data from the Health Examination Survey.—Data from direct exarnination, testing, and measure- ment of national samples of the civilian, noninstitutional population provide the basis for two types of reports: (1) estimates of .the medically defined prevalence of specific diseases in the United States and the distributions of the population with respect to physical, physiological, and psycho- logical characteristics; and (2) analysis of relationships among the various measurements without reference to an explicit finite universe of persons, Data from the Institutional Population Surveys — Statistics relating to the health characteristics of persons in institutions, and their medical, nursing, and personal care received, based on national samples of establishments providing these services and samples of the residents or patients. Data from the Hospital Discharge Survey.—Statistics relating to disciuirged patients in short-stay hospitals, based on a sample of patient records in a national sample of hospitals, Data on health resources: manpower and facilities. —Statistics on the numbers, geographic distri- bution, and characteristics of health resources including physicians, dentists, nurses, other health occupations, hospitals, nursing homes, and outpatient facilities, Data on mortality.—Various statistics on mortality other than as included in regular annual or monthly reports—special analyses by cause of death, age, and other demographic variables, also geographic and time series analyses. Data on natality, marriage, and divovce.—Various statistics on natality, marriage, and divorce other than as included in regular annual or monthly reports—special analyses by demographic variables, also geographic and time series analyses, studies of fertility. Data from the National Natality and Mortality Surveys.— Statistics on characteristics of births and deaths not available from the vital records, based on sample surveys stemming from these records, including such topics as mortality by socioeconomic class, hospital experience in the last year of life, medical care during pregnancy, health insurance coverage, etc. For a list of titles of reports published in these series, write to: Office of Information National Center for Health Statistics Public Health Service, HSMHA Rockville, Md, 20852 Series 2 - Number 55 VITAL and HEALTH STATIST DATA EVALUATION AND METHODS RESEARCH ANC; Tenis Q 2 Z f°) NCHS bo 1] 7 ~ CAIN The Prediction Approach to Finite Population Sampling Theory: LT OIRO (CSE) Discharge Survey U. S. DEPARTMENT OF HEALTH, EDUCATION, AND WELFARE Public Health Service Health Services and Mental Health Administration Vital and Health Statistics-Series 2-No. 55 For sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, D.C. 20402 ‘Price 65 cents Domestic Postpaid or 45 cents GPO Bookstore Series 2 DATA EVALUATION AND METHODS RESEARCH Number 55 The Prediction Approach to Finite Population Sampling Theory: Application to the Hospital Discharge Survey DHEW Publication No. (HSM) 73-1329 U.S. DEPARTMENT OF HEALTH, EDUCATION, AND WELFARE Public Health Service Health Services and Mental Health Administration National Center for Health Statistics Rockville, Md. April 1973 NATIONAL CENTER FOR HEALTH STATISTICS THEODORE D. WOOLSEY, Director EDWARD B. PERRIN, Ph.D., Deputy Director PHILIP S. LAWRENCE, Sc.D., Associate Director OSWALD K. SAGEN, Ph.D., Assistant Director for Health Statistics Development WALT R. SIMMONS, M.A., Assistant Director for Research and Scientific Development JOHN J. HANLON, M.D., Medical Advisor JAMES E. KELLY, D.D.S., Dental Advisor EDWARD E. MINTY, Executive Officer ALICE HAYWOOD, Information Officer OFFICE OF STATISTICAL METHODS MONROE G. SIRKEN, Ph.D., Director E. EARL BRYANT, M.A., Deputy Director Vital and Health Statistics-Series 2-No. 55 DHEW Publication No. (HSM) 73-1329 Library of Congress Catalog Card Number 72-600134 FOREWORD The Center contracted with the School of Public Health, Johns Hopkins University and Dr. Richard Royall to investigate the possible application to the Hospital Discharge Survey of the prediction approach to finite population sampling. This report presents the results of the research completed under these contracts. The prediction approach is based on “super-population” probability models. It is an alternative to the conventional theory of sampling from finite populations and does not apply the conventional concept of repeated random sampling from a fixed population. Rather, it applies classical prediction theory to solve sampling problems. Viewing finite population sampling problems as prediction problems is a relatively new development and hence is probably known to only a few statis- ticians. Furthermore, Dr. Royall’s style is throughout the report quite elegant. Therefore, we asked him to prepare a nonmathematical description of the predic- tion approach and indicate how it differs from the classical approach. This material is presented in the Introduction. We commissioned this research project in anticipation of redesigning the Hospital Discharge Survey. Overall, the findings presented in this report throw a favorable light on the existing design and estimator. The findings suggest some changes for improving the design and also identify some areas for further research. We believe this report will help us to develop an improved design for the Hospital Discharge Survey. Dr. Jay Herson worked with Dr. Royall and the Office of Information in pre- paring this manuscript for publication. MONROE G. SIRKEN iii CONTENTS Introduction Summary . Part I. Single-Stage Sampling . Description of Problem Optimality . Effects of Errors in ‘he Model Estimation of Variance Part II. Two-Stage Sampling Description of Problem Optimality Considerations The HDS Estimator References Appendix Derivations of Conditions on Optimal Stratification with Equal Allocation and Defensive Sampling Derivation of Expressions (35) and (36) for Vorfnce Page YO NII 19 20 21 27 29 29 29 THE PREDICTION APPROACH TO FINITE POPULATION SAMPLING THEORY: APPLICATION TO THE HOSPITAL DISCHARGE SURVEY Richard M. Royall, Ph.D., Associate Professor, Department of Biostatistics, School of Public Health, Johns Hopkins University INTRODUCTION The material presented is the result of an un- orthodox approach to finite population sampling problems. Specifically, it describes the elements and results of an application of this approach to the Hospital Discharge Survey (HDS), a continuing sample survey of the Nation’s short-stay hospitals conducted by the National Center for Health Statistics. It is not presented as a finished and polished analysis but as a basic sketch whose contents must be critically evaluated, adjusted, and refined if it is to be of real value in HDS. The mathematical model used in this work expresses plausible initial assumptions about certain variables of interest. With experience will come increasing knowledge concerning the HDS population and relationships among its variables. Such information must be used to alter and develop the basic model described in this report. In this section the approach guiding the investiga- tion will be contrasted with the conventional ap- proach to finite population sampling problems. For purposes of illustration, an imaginary population of 50 hospitals in some relatively homogeneous geo- graphical region will be considered. The number of beds in each hospital is known. A sample of 10 hospitals is selected, and the number of patients discharged from these 10 during some given time period is observed. The problem is to estimate the total number of discharges from all 50 hospitals (the population total). In its basic, simplest version, the conventional approach treats the 50 unknown numbers of dis- charges as unknown constants. The only random variation in the problem is injected by the sampler, who uses a random sampling plan to decide which 10 hospitals will comprise the sample. This sampling plan specifies the probability of selection of each potential sample. A sampling and estimation pro- cedure consists of a sampling plan together with an estimator or formula for calculating estimates from samples. The characteristic feature of orthodox sampling theory is that a procedure is evaluated in terms of the statistical properties of the estimator, principally its expected value and variance, under the random sampling plan chosen by the sampler. Of course other factors, e.g., costs, feasibility, and ease of estimation of variance from the sample influence the choice of a procedure. Nevertheless, the basic objective is to find, subject to limitations such as cost, a procedure whose estimator is un- biased (at least approximately) and has small variance. For present purposes only one sampling plan and two estimators are considered. The plan calls for simple random sampling— only samples which con- sist of exactly 10 different hospitals are allowed, and all such samples are equally likely to be selected. lett), ts, . . ., tsorepresent the respective numbers of discharges from the 50 hospitals, let by, bs, . . ., bso be their respective numbers of beds, and let s represent the set of 10 hospitals in the sample. A 50 simple estimator of the population total, T= >t, 1 is the product of {the average number of discharges per hospital in the sample} X {the number of hospitals in the population}, i.e., (1) ( 3/10) 50. This is called the simple expansion estimator. Under 1 the present (simple random) sampling plan it is unbiased. Another estimator (the ratio estimator) esti- mates T by the product of {the average number of ‘discharges per bed in the sample} X {the total number of beds in the population}: (Su/sb) Se @) Under the simple random sampling plan the ratio estimator is biased. Two observations concerning the variances of the expansion and ratio estimators are needed: (i) both variances are defined as average values of squared errors over all samples, and (ii) the two variances are unequal. Such biases and variances are certainly relevant in planning surveys and choosing procedures which can be expected to produce good estimates. How- ever, after the sample s is selected the situation is drastically changed. As indicators of uncertainty in the estimator when it is applied to a particular sample, the conventionally defined bias and vari- ance can be quite misleading. For example, if the sample contains mostly small (few beds) hospitals, we can be confident that the expansion estimator (1) will give an underestimate of T. In this situa- tion, to describe the estimator as “unbiased” is at best irrelevant and at worst misleading. Here it would seem accurate and informative to describe the estimator as having a negative bias, yet this is impossible —for a given sample s there is no proba- bility distribution with respect to which bias can be defined. Similar remarks apply to samples containing a disproportionate number of large hospitals —in these samples the expansion formula tends to produce overestimates of T. In this con- text, the statement that the estimator is “unbiased” in the conventional sense simply means that samples containing too many small units, which tend to give underestimates of the population total, will be balanced, in a ‘hypothetical infinite sequence of samples, by samples containing too few small units, which tend to give overestimates. It would appear that when s contains an excess of small hospitals, an upward adjustment is required if (1) is to deserve the description “unbiased.” The adjustment might be made by multiplying 50 (1) by the factor (3 bi50) / (3 bio), the ratio 1 § of {the average number of beds per hospital in the population} + {the average number of beds per hospital in the sample}. The effect of this factor will be to increase the estimate when the average sample hospital is small and to decrease the esti- mate when the average sample hospital is large. The resulting estimator, {($ 050) /( 5 00) } {503 ero}. 3 is the ratio estimator (2), which, according to the conventional definition, is biased. Thus in this problem a notion of bias useful for inference from a given sample s must be in direct conflict with the conventional theory; the unbiased estimator should be called biased and vice versa. The orthodox variance (or its square root, the standard error) is not a satisfactory measure of the uncertainty in the estimator after s is fixed, although it is usually interpreted as such a measure. The two estimators (1) and (2) have different variances, yet when the sample is such that the average size of sample hospitals is equal to the average size for the whole population, the results of using (1) and (2) are identical. That is, when such samples are selected, the ratio and expansion formulas are the same and therefore equally precise, equally uncertain, equally accurate, etc. Yet orthodox theory assigns different standard errors depending on whether formula (1) or formula (2) was used. .The prediction approach recognizes that, after the sample is observed, the population total can be written = tit ti (4) where s denotes the collection of hospitals not in the sample. Since the first of the two sums in (4) is now known, the problem is to estimate the second sum, the total number of discharges in hospitals not in the sample. Any estimator of T can be written in a form comparable to (4), i.e., f= ut (F-Su) (5) Using T to estimate T is, in effect, using T— ¢; 8 to estimate Y ti. Clearly the questions of whether 8s a particular estimator when applied to a particular sample s is good or bad, reasonable or foolish, unbiased or biased, etc. are answerable only in light of the relationship between hospitals in the selected sample and those not in the sample. An estimator T is precisely as good for estimating T as is the difference T— ti for predicting the un- 8 observed sum > ti. 8 The prediction approach expresses the relation- ship between sample and nonsample hospitals by a probability model (‘“‘super-population” model) in which the numbers of interest, ¢,, t., . . ., ts, are thought of as having been produced by some probabilistic process described by a mathematical model. This process serves as a vital link between the observed and unobserved totals. What these two totals have in common and what enables us to use the observed to make inferences concerning the unobserved is that they were all produced by one underlying probabilistic process. Inferences from the sample can be made concerning certain important characteristics of the process; this information can then be used to predict the values of the totals not observed. The simplest model describing the basic structure of the hospital problem treats t;, the number of discharges from hospital i, as an observation on a random variable whose expected value is propor- tional to b;, the number of beds. That is, the ex- pected number of discharges is Bbi, where B is some unknown positive constant which can be estimated from the sample. If (1) and (2) are written in forms comparable to (4), then the expansion estimator is 2 ti + 40 (= 4/10). 6) s and the ratio estimator is" Sut(s v/ Sb) Sn a) 8 Using the probability model, s can be held fixed and the statistical properties of the estimators for the given sample examined. Thus the second terms in (6) and (7) are actually predictors for the random total discharges from nonsample hospitals. The properties of the expansion estimator for this sample are precisely the properties of 40 (= 10) 8 when itis used to predict > ti. The expected value of 8 the predictor is 40 (= gb10), while the variable predicted has expected value > Bbi. Since the ex- 8 pected value of the predictor is less than that of the variable predicted when the average size of sample hospitals is less than the average size of nonsample hospitals, the prediction approach describes the expansion estimator as ‘biased’ in this context. The ratio estimator, on the other hand, is called “unbiased” for every s since the expected value of the predictor, (= Bbi / > b) Y bi, equals the expected value, 8 > bi, of the variable predicted. The variance used to measure uncertainty in an estimate under the prediction model is, like that used in the conventional approach, the variance of the difference T—T between the estimator and the quantity estimated. But whereas the conventional approach calculates the variance of this difference with respect to the random sampling plan (the proba- bility distribution over all possible samples), the prediction approach calculates the variance with respect to the probability model with the sample s held fixed. Thus the conventional approach states for the ratio estimator, say, the same standard error for all samples of size 10, while the prediction approach quotes one value when the sample con- tains mostly large hospitals and a larger value when most of the sample hospitals are small (See formula (3), page 15.) Both the conventional and prediction variances are unknown and must be estimated from the sample. There is theoretical and empirical evidence that the latter is the more useful measure of the uncertainty in an observed estimate [1].! This simplified example suggests the inadequacy of orthodox notions of bias and variance for pur- poses of inference and points to the prediction approach as being more relevant and informative at the data-analysis stage. However, some of the most interesting implications of the prediction approach appear when the problem of sample selection is considered. When this approach is adopted random sampling loses its status as the one and only fundamental and indispensable component of finite population sampling theory; it ' Figures in brackets indicate the literature references at the end of this paper. assumes instead the more humble role of a useful and important tool. To apply the prediction approach to a real prob- lem, we must first be able to produce an adequate model which is simple enough to analyze. The adequacy of a model is to some extent a matter of judgment, but mathematical investigations can help. Thus considerable attention is paid in this report to the effects of errors in the basic model and especially to the identification of samples for which the conclusions derived from the model are relatively insensitive to the most obvious sorts of departure from the model. The models in this report are used in two ways: to generate sampling and estimation procedures having certain desirable statistical properties and to provide increased appreciation of the properties of procedures currently in use. SUMMARY The author has recently been studying finite population sampling problems using an approach which is based on viewing such problems as straight- forward classical prediction problems rather than on applying the conventional concept of repeated random sampling from the fixed population. Previ- ous work by Royall [1, 2] has suggested that the prediction approach, which employs super-popula- tion probability models, is a useful alternative to the conventional theory and can be of value in illuminating the strengths and weaknesses of standard procedures as well as in suggesting and providing a theoretical basis for new procedures. Other recent studies viewing finite population sampling problems as prediction problems have been made by Ericson [3, 4], who adopts a Bayesian approach, and by Kalbfleisch and Sprott [5], whose approach is fiducial. There have also been other studies in which the classical (non-Bayesian, non- fiducial) approach is adopted, e.g., Brewer's paper [6] and parts of the paper by Scott and Smith [7], whose basic approach is Bayesian. HDS employs a two-stage sampling plan in which hospitals are the first-stage sampling units and pa- tient discharge records the second-stage units. Within each of four geographical regions, hospitals are stratified according to size, as measured by the number of beds (bed size) listed in the 1963 Master Facilities Inventory of Hospitals and Institutions (MFT). For the purposes of this study, the hospitals in the four geographical regions are treated as natural, distinct populations which represent four separate instances of the same basic problem. Thus the ‘“‘pop- ulation” referred to in this report corresponds to the HDS population within any of the four large geo- graphical regions, and stratification is on the bed size variable only, not on geographical region. In HDS a sample of hospitals is selected from each stratum, and a sample of discharges is drawn from each selected hospital. For each discharge in the sample a numerical characteristic of interest, or response, is observed. Sample discharges from a given hospital are used to estimate the total for all discharges from that hospital. These estimated totals for the sample hospitals are then used, along with the auxiliary variable, bed size, to construct a ratio-type estimator for the stratum total. This estimation procedure is applied independently within each stratum. In Part I of this report complications produced by the second stage of sampling are set aside, and only single-stage sampling problems are considered. The main purpose of this part of the study is to gain an increased understanding of the simple and valuable ratio estimator. Thus we consider a range of proba- bility models, but with more attention paid to study- ing the performance of the ratio estimator under such models than to describing optimal sampling and estimation strategies for each model. We see in Part I a new explanation for the success of the ratio estimator in practical applications: although real problems are not often depicted with great accuracy by the probability model under which the ratio estimator is optimal, frequently, for the particular sample drawn, the ratio estimator is approximately optimal under a wide range of models. Stratification on the size variable with separate ratio estimation in the strata is examined as a tech- nique for efficiently insuring unbiasedness. Finally, the effects of errors in the model on the performance of variance estimates are considered. In Part II the second stage of sampling is intro- duced. The problem is first studied in its simplest form; later the phenomena of out-of-scope and nonresponse discharges are represented in the model. Overall, the results throw a favorable light on the HDS design and estimator. This investigation suggests that the rule used to allocate the first- stage sample among the various strata might be improved, but that, given the rule actually used, the allocation of the second-stage sample is approx- imately optimal. Another suggestion is that the average bed size per hospital in each stratum’s sample should be approximately equal to the average bed size per hospital in the entire stratum. It is supposed that the present method of hospital selection produces samples which satisfy this con- dition, but this should be verified. Two areas in which further research with super- population models is expected to be fruitful are analysis of the HDS variance estimator and study of the sophisticated sampling technique known as “controlled selection.” The first of these is of more immediate importance since the current HDS variance estimator is an adaptation of the variance estimator conventionally used in single- stage ratio estimation problems. There are theo- retical results, supported by some empirical work, which imply that this conventional variance esti- mator should be replaced by one suggested by super- population theory [1]. Controlled selection procedures are used by the HDS to select the first-stage sample. Investigation along the lines leading to defensive samples in Part I would probably increase our appreciation of precisely what these procedures accomplish and how. Such an investigation should provide theo- retical support for these selection procedures. PART I. SINGLE-STAGE SAMPLING Description of Problem Terminology, notation.— The population of inter- est consists of M units labeled 1, 2, . . ., M. Asso- ciated with unit 4 are two numbers (By, t.) with B,. known and ¢, fixed but unknown. The units might be hospitals of a certain type with B, some measure, for instance, number of beds, of the size of hospital k, and t; some characteristic of interest such as number of days of care provided by hospital & during a particular month. A sample consisting of m units is to be selected from the population and the t-values associated with the sample units are to be observed. The objective is to estimate the total M T= wu 1) k=1 and give a measure of the precision of the estimate. The set of m labels identifying the sample units is denoted by s, and the set of M-m labels of units not in the sample is denoted by s. Probability models. —In this study the numbers ti, ta, . . ., tu, whose sum we must estimate, are considered to be realized values of independent random variables T,, Ts, . . ., Ty. The expected value and variance of T) depend on the size measure By and are denoted by h(B,) and a? v(B,), respec- tively. Thus we can write Ty=h(B:)+ ex Vv(By) k=1,....M where €;, . . ., ey are independent random vari- ables, each having mean zero and variance o2. In particular, attention ‘is focused on models in which h(B) is a polynomial, say, of order J (at most). That is, h(B)= roBo+ riB.\B+ r.8:B*+ co oP r,B.B’ where the r’s are zeroes and ones. If r; =1, it means simply that the term 8;B/ appears in the regression function; r; = 0 indicates the absence of this term. When the regression function h has the above form, we refer to the probability model as &(ro, ri, . . ., ry : v(B)). For example, £(0, 1 : B) refers to the model Tx = BiBx + € VB , in which both the expected value and the variance of Tx are proportional to the size Bx. As another example, £(1,1, 0, 1 : 1) refers to the model T= Bo+ BiBr + BsB; + €x. Here Var T= Var ex = 02, a constant. It should be emphasized that the fundamental problem is that of estimating the sum (1) of the actual ¢-values. If a particular model, say £(1,1: B), applies, an intermediate step in the process of estimating the sum is estimation of By and B,, but M the objective is to estimate tk, not the parameters K=1 in the super-population model. It will be especially important to keep this objective in mind when seeking optimal sampling plans since the plan which is best M for estimating > tx under a particular model is not k=1 generally the best plan for estimating parameters of the model. Under probability models of the sort considered M here, the problem of estimating the total > tx on the basis of a sample s is a version of the general problem of predicting future observations on random variables. This is evident when the total is expressed as the sum of two terms, > tx and > tx. The first - a 8 of these two is known after the sample has been observed, and estimating )' t is equivalent to pre- 8 dicting the sum of the unobserved random variables > Tk. For further discussion of this view of certain 8 finite population sampling problems as prediction problems, the reader is referred to Royall [2]. Optimality Best linear unbiased (BLUE) estimators.— For a given sample s and a given model £, an estimator T will be said to be unbiased if E(T—T)=0, where the expectation is taken with respect to the probability distribution specified by the model. For example, for all s the ratio estimator, (su /3m)3m is unbiased under the model £(0, 1:v(B)) for any variance function v: ills /3m)$n-3r] Under the model £(1, 1: v(B)) ef(zn/ga)Ea-3n] 1 mBo+ Bi > Bi. 7 = TSE Be 2 Br—MBo— Bi $ Bi m $ Bi = ul s 7 u) Thus under this model the ratio estimator is un- M biased only if > B. | M= > Bi/m. 1 8 Only estimators which are linear functions of the t's in the sample are considered here. The determination of a best linear unbiased estimator under a given model and for a given s is quite simple. We seek among all linear unbiased esti- mators T one which minimizes the mean square error (MSE), E(T—T)> The estimator T is un- biased if and only if the difference between T and the sample total > Ti is an unbiased estimate of the total for nonsample units, i.e., E (T= 1) =E (ST). Thus if T is unbiased, sr) ((r- rors) + Var (2 Ty) = Var (7- > T) + Var ( > T.) Note that linearity of T is equivalent to linearity of #3 T.. Therefore, under the model &(ro, ri, . .,rji:v(B)), T is a BLUE estimator for T if and only if T— STi is a BLUE estimator for the ex- J pected value of > Ty, > (> 5) riB;. J=0 “3 The generalized Gauss-Markov theorem (see Rao [8]) shows that the BLUE estimator for such a linear function of the regression coefficients is obtained by straightforward application of the familiar method of weighted least-squares estimation. Thus under the present model, T is the BLUE estimator for T' if T= > r+ ( 2 Bf Ss where the B's are the weighted least-squares estimates of the regression coefficients under the specified model. Two examples will perhaps clarify this point: Example: Under the model ¢(1,1 : 1) the weighted least-squares estimates of Bo and 3, are Bo(1,1:1) = (z83 T.- 3 BS BT) /D Bi(1.1:1) = (m NEUES T.)/D where D=m3B~(3 8) Ss The BLUE estimator for T is thus T(1,1:1) = Ti+ (M—m)Bo(1,1:1) +8:(1,1:1) 3 Bu. 8s Example: Under the model ¢0, 1: B), the weighted least-squares estimator for 83, is BiO.1:B)=3T, [SB Thus the BLUE estimator for T is 70,1: B)=3T,+p:(0,1:B) YB. This estimator can also be written in the more familiar form T(0, 1 :B)=( ST /SB)SE 2) So the BLUE estimator under the model £0, 1 : B) is the popular ratio estimator. Optimal samples.— The model £0, 1: B) is of particular interest since it is under this model that the standard ratio estimator is optimal. Here the expected squared error is E«T0.1:8)-1=c*( TB /S8)3 8. (3) From (3) it is apparent that in this context the optimal sample is one for which > B, attains its maximum value. This is simply the sample com- posed of the m units whose B values are largest. It is the sample which is optimal for use with the optimal estimator under the model £0, 1 : B) and will be denoted by s(0, 1 : B). (See Royall [2].) More generally, under the model E(roy ray vo, ry:v(B)), the sample for which E¢(T (ro, ry, . . ., ry: v(B))—T)? is minimized is optimal for use with the optimal estimator. This sample will be denoted by s(ro, r1, . . ., ry: v(B)). Effects of Errors in the Model We now assume that the population of interest is one for which £(0, 1 : B) is a plausible model but cannot ignore the possibility that this model is in- accurate. Thus we seek strategies which are nearly optimal under £(0, 1 : B) but will produce satisfac- tory results under various other models. Overall ratio estimator.—Under the model £(0, 1 : B) the optimal estimator is 7(0, 1 : B), the ratio estimator, and the optimal sample for use with this estimator is s(0, 1 : B), the sample consisting of the m units whose B-values are largest. That is, of all strategies (s, T) consisting of a sample s and a linear unbiased estimator T, the pair (s(0, 1 : B), T (0,1: B)) is optimal under the model £(0, 1 : B). Many questions arise at this point. How good is this strategy when £(0, 1 : B) is not the correct model? If we use T'(0, 1: B), is s(0, 1 : B) a good sample when the true model is £(0, 1 : v(B)) for some particular variance function v(B) # B? How can we find a procedure which is good under £(0, 1 : B) but performs adequately under the alternative model £(1, 1, 1 : B)? Answers to some questions of this sort are known. For example, it is well known that the unbiasedness property of BLUE estimators is not destroyed by alteration of the variance func- tion. Thus T(0, 1 : B) is unbiased under the model £(0,1:v(B)) for any variance function v. More generally, consider the estimator T(0, 1 : B) under the model (ro, ry, . . ., ry: v(B)). The bias is E«(T(0,1:B)—T) =F, (=z Tv SB) SB 7) , [SB SB “AP SETSE 25 J Note that the summand is zero when j=1; the bias is not affected by the regression coefficient 3,. It is clear from this expression that T(0, 1:B) is un- biased if and only if SB, XB SB 3B for all j such that the term B; B/ appears in the regression equation (i.e., for all j such that rj=1). It is easily shown that these conditions for unbiased- ness are equivalent to 1 oc 1% mB y2 Bl ToT @) m2 Be 2B for all j such that rj=1. Note that (4) is always satisfied for j=1. For example, T(0, 1:B) is un- biased under the model £(1, 1 : v(B)) if (4) is satisfied for j=0: 1 1 & = Be=j1 2 Br (5) 8 This estimator is unbiased under the still more gen- eral model £(1, 1, 1:v(B)) if, in addition to (5), s satisfies 1 a 1 Sh m2 B= 2B Suppose it is believed that £(0, 1: B) is an ade- quate model for a given problem, but the estimator must perform reasonably well when the model is in error and the actual regression function is not a straight line through the origin. The following theorem shows that by careful choice of the sample s we can insure the optimality of the ratio estimator under polynomial regression models. For any posi- tive integer J, let s(J) denote any sample satisfying 4) for j=0,1, .. ., J. Theorem: If s=s(J), then T(0, 1:B)=T(1:1), and this is the BLUE estimator under the models &(roy 1, rey 13, . . ,riB) and (1, rye. oy 121) for every sequence ro, Ii, 2, . . ., I; of zeroes and ones. Proof: Note that for any s, T(1 1) =M > Tylm, and for s=s(J) this statistic is also T(0, 1:B). 10 This estimator has already been shown to be unbiased when s=s(J) under J"-order polynomial regression models for any variance function. One can prove its optimality when s =s(J) under all models of the form &(ro, 1, rs, . . ., rs:B) by: (i) finding the weighted least-squares estimates Bj(ro, 1, rz, . . ., r;:B) for all j=0,1,2,...,J such that rj=1, under the model £(ro, 1,12, . . .,1s:B); (ii) forming the BLUE estimator for T, T (ro, 1, ni, «ey ri:B)=Y Tk J - +3 (SB) Bi, 1, r2, « « + r;:B) Tj; Jj=0 = and (iii) noting that when s=s(J) this esti- mator assumes the simple form M > Ti/m. 8 Alternatively, we note that since T (ro, 1, ri, w 1558) is unbiased under the model £(0, 1: B) and T(0, 1:B) is the BLUE estimator under this model, we have for all s E{(T(0,1:B) —T)*| £(0, 1:B)} E {(T(ro, 1, ro, . . ., 12:B) = T)2| £00, Y, 1s, « 72: BY) (7) Now E{(T(ro,1, rs, . .. rs:B)—T) | €(ro, | 2, « + + r:B)} = E{(T (ro, 1, ra, . .,r:B)—T)2| £00, 1:B)} for all s, and when s=s(J) this equality holds if T(ro, 1, ray . . ., rs:B) is replaced by T(0, 1:B). Thus (6) implies that when s=s(J), equality must hold in (7). Optimality under £(1, ry, ra, ., rs: 1) can be proved by entirely analogous arguments. Samples s(J) will be referred to as defensive samples. Selection of a defensive sample insures that the ratio estimator retains not only its unbiased- ness but also its optimality under the polynomial regression models. As noted above, one convenient feature of defensive samples is the simple form which the ratio estimator assumes. When Y B./ M 8 m= > Bi/M, the ratio estimator is simply the expan- 1 sion estimator, M y T./m. Example: Under the model £(1, 1: B), the BLUE estimator is 7(1,1:B)=Y Ti+ (M—m)Bo(1,1:B) +B:1(1,1:B) 3 By 8 where # T, B11: =(S FS Be—m 3 1), and Ty Bi(1,1:B) -(2 TS Sm 3 2). with = 1 2. “gr 3am M -— When > Bi/m=Y Bi/M=B, we have 8 1 T(1,1:B)=3 Ti+ mlzaiozn 5) g Li Ss s =M 3 Ti/m. When a defensive sample is used, the mean square error (MSE) of the ratio estimator under £(0, 1:B) is, from (3), E(T(0,1:B) —T)? = Zz ji ) *B. 8) When the estimator is unbiased, the MSE is simply the variance, and the variance does not depend on which terms appear in the regression model. Thus we see that when s is s(J), expression (8) applies under the model &(ro, ri, . . ., r;:B) for any combination ro, ri, . . ., r; of zeroes and ones. (Note that (8) does not apply under &(ro, ry, . . ., r;:1).) It follows that when T(0, 1:B) is the chosen estimator under the model ¢(0, 1:B), the ratio of the MSE when s is the optimal sample s(0,1 : B) to the MSE when s=s(J) forany J =1is min ( 2 B.] M—m )/( 2 Bi/m ) These results may be interpreted as follows: When £(0, 1:B) is the true model, the ratio esti- mator is optimal for any s. If the ratio estimator is used but the model is actually £(1, 1l:v(B)), a bias is incurred. We can guard against such a bias by choosing the defensive sample s(1) instead of the sample s(0, 1:B). Protection against a certain type of error in the model £(0, 1:B) is gained, and some efficiency under this model is lost. If we now decide to impose the additional conditions M SBm=3 BM, j=2, 3, ... J, 8 1 insuring the unbiasedness of our ratio estimator under any model of the form &é(ro, ri, . . .,ry:0(B)) (and insuring the optimality of our estimator under any model ¢(ro, 1, rey . oo, r:B) or £(1, ri, 1, ., r;:1)), we incur no additional loss in efh- ciency. Protection against many types of error in the model £(0, 1:B) has been gained at no cost in terms of additional loss of efficiency under £(0,1:B). Some rough idea of the cost of such protection can be gained by looking at a population in which thereby 11 the Bj are uniformly distributed over the interval (a, a(1+A)) for any a, A=0. In this case, for m BM = 1 8 Bi/m, then the term B'g(Cx) contributes no bias if the sample is “representative” in the sense that > 8 M g(Ci)/m =" g(C)IM, i.e., if the average value of 1 g(C.) in the sample is the same.as the average value in the population. The foregoing results provide some theoretical support for the procedure of selecting a sample at random and using either the simple expansion esti- mator or the ratio estimator. The average value of M oh 0 2 B//m over all (1) samples s is 2 Bi IM for j=1, 2, . . .. In precisely the same sense that the mean of a simple random sample can be expected to be approximately equal to the population mean, a sample selected at random can be expected to approximate s(1). This is true because s is s(1) M when Biim=Y Bi/M. The same reasoning 8 1 applied to higher powers of B implies that simple random sampling will frequently produce a sample 12 which is a fair approximation to s(J) for some J = 1. Whenever this occurs, the expansion and ratio estimators are approximately the same; both are approximately unbiased under the model £(1, 1, . . .,1:v(B)) and approximately optimal under this model when v(B)=B or v(B)=1. The same argument applies to the unobservable (or simply unobserved) regressor g(c). Unbiased estimation of T is possible only if the effect of g(c) is negligible or if the sample is “representative” in the sense defined above. An important role of random sampling is to pro- vide samples which are “representative” with respect to such regressors. Of course random sam- pling cannot guarantee successful choice of a repre- sentative sample, and the probability of a successful choice depends on the unknown distribution of g(c) in the population. Nevertheless, random sam- pling provides a basis for optimism, as shown by the Tchebycheff inequality. The use of simple random sampling as a means of obtaining a sample approximating s(J) produces samples which are not all good approximations to s(J) and are, on the average, less efficient than s(J). Under the model £(0, 1:B), for a given sample s the MSE of the ratio estimator is a? M 3 BY, Bi/Y Bk. The average value of this quantity ~ 1 8 over all (*) possible samples of the required size M M is 02 Bx 2 > Bi—1) where c is the average 1 1 M value of us Bim) over all () samples. Now a well-known inequality? shows that c is greater than M or equal to (> BM), with equality only in case 1 Bx=B, for all k,[=1, . . ., M. Thus, except when all B’s are equal, the average MSE over all possible .. . M2 m\ . = samples of size m is greater than = 1—%7 a? B, the MSE when the sample is s(J). . We have seen that the strategy (s(J), 7(0,1:B)) produces unbiased estimates under the model £(1, 1, . .. 1:0(B)). The strategy (s(1,1, .. ., 1:B), T(1,1, . . ., 1:B)), which is optimal under £1, 1, ..., 1:B), also produces unbiased esti- 1 1 “For any nonnegative random variable X, E —= —. Equality holds if and only if X X EX is a constant (with probability one). mates under this model. The MSE for this strategy is the minimum value over all s of E{(T(1,1, , 1:B)—T)*|&(1, 1, ,1:B)}, 9) which is less than the value of this expression when s=s(J). But when s=s(J) we know from the theorem that T(0, 1:B)=T(1, 1, , 1:B) and thus that E{(T(,1, ,1:B)—T)2|¢(1,1, . . .,1:B} =E{(T(0,1:B)-T)2|£¢(1,1, . . .,1:B)} =E{(T(0, 1:B)—T)2|£(0, 1:B)). Therefore, under ¢(0, 1: B) the MSE of the strategy (s(1, 1,..., 1:B), T(1, 1, . . ., 1:B)) is less than that of the strategy (s(J), T(0, 1:B)). Never- theless, because of the popularity and simplicity of the ratio estimator, as well as because the current HDS estimator is of the ratio type, the remainder of Part I is devoted to situations in which the ratio estimator (or a sum of ratio estimators) is to be used. Stratification on the size variable—We have seen that the unbiasedness of the ratio estimator can be preserved under J™-order polynomial regression models by the choice of a sample s(J) which is “like” the population in the sense that M YBiim=Y BM forj=1,2,...)J. 8 1 An alternative means of preserving the unbiasedness property employs stratification on the size (B) variable and use of a separate ratio estimate in each stratum. The double subscript hk denotes quantities associ- ated with unit k in stratum h. Thus Bj is the size and Tp, the response of unit 4 in stratum A. The number of units in stratum h is My, and T), and B) are the totals for stratum h. In this notation the grand total T is expressed as H H My T= > T,= > > Thi h=1 h=1 k=1 where H is the number of strata. The strata are defined as follows: the M, smallest units form stratum 1, the next M. smallest units form stratum 2, etc. Thus when h < hk’, By, < By for all 4= wMyand A'=1, .. ., M,. A sample s, consisting of my units is chosen from stratum h, and the total T) for that stratum is estimated by T,= Sn ¥ Bu. (10) Any sample s, for which 2 Bi | m= > Bi | My for j=1, 2, oy J will be referred to as sy (J). If sn=sn(J), then Th is an unbiased estimate of Ty under the model hk J Ty. = Sy riBiB) + en Vu(Bni) hk j=0 k=1,2,.. ..M, (11) where the €’s are independent, each with mean zero and variance o2, and ro, ry, ., I; is a sequence of zeroes and ones. The case in which (11) applies for h=1,2, ., H will, as previously, be denoted by E(ro,r, « . .,ry:v(B)). From the earlier results we see that the estimator T= y T) is unbiased with MSE h ” 2 Bi 3 Bg 5 Bk h=1 0 (12) under £(0, 1:B). The estimator is unbiased and has the MSE (12) under the more general model E(ro,r, . . .,rp:B) ifsn=sn(J) forh=1, . . ..H. Note that shen such a sample is chosen, dhe esti- mator T' becomes simply 3 (Mn > Twi | my), h=1 Sn 13 and the MSE is H M:? my _ (13) “ h a ot 3 (1 a, Br h=1 where By=Bn/Mhn. If a defensive sample is drawn within each stra- tum, ie., if sp=s,(J) for some J=1 and every h=1,2, . . .,H, then with proportional allocation (mn/My constant) the estimator T=" T» is simply h the overall expansion estimator M S$ > Tuilm. h=1 Sh In this case the MSE (13) becomes simply o* B M(M —m)/m, where B denotes the grand average H > BuM. h=1 Optimal allocation, subject to fixed total sample size m and with defensive sampling within strata, is easily seen to require that m, be proportional to M,NB,=VM,B) for h=1, . . ., H (cf. Cochran [9]. With optimal allocation the MSE (13) becomes o[(g vwm) $0) h=1 In order that proportional allocation be optimal, we must have M, proportional to M, VB, which means that B, must be constant. But with stratifica- tion on the size (B) variable, B; can be constant only in the degenerate case of a population whose units are all of the same size. Thus in nontrivial cases, proportional allocation cannot be optimal. The foregoing results establish the superiority, with respect to MSE, of the stratification procedure to the nonstratified defensive sampling procedure. We refer to these procedures as II and I, respec- tively, and summarize the argument establishing the superiority of the former. Procedure I. Choose any s(J) and use the esti- mator (2). Procedure II. Stratify on the size variable, choose any s,(J) from the h'" stratum, and use the H estimator T= > Th. h=1 14 (i) Both procedures (I and II) produce estimators which are linear and unbiased under &(ro, ri, . . ., r;:v(B)) for any - sequence rg, ry, . . .,r; of zeroes and ones. (ii) Optimal allocation for procedure II requires that m, be proportional to VM B. (iii) If proportional allocation is used in procedure II (mp=mMn/M), then proce- dures I and II have the same MSE. (iv) Proportional allocation cannot be optimal except in trivial cases. From (i)Hiv) we conclude that (v) If optimal allocation is used, then procedure II has smaller MSE than procedure I. Note that (v) is true regardless of the number and relative sizes of the strata. The only require- ment is that it be possible to use optimal allocation and defensive sampling. Now the same argument, (iv), which shows procedure II to be superior to procedure I can be applied within any stratum to which more than one observation is allocated. It pays to substratify. This implies that when J=1, the optimal number of strata is H=m (and optimal allocation is mp=1, h=1, . . .,m). Of course, if we want to guarantee unbiasedness under more general models (larger J), each mj, must be greater than or equal to J, because when my is less than J it is impossible to select spy=s,(J) except in highly special, degenerate cases. There are also the obvious problems encountered in selecting samples Mp sn(J); exact satisfaction of > Bi /mn=" Bi/Mh Sh 1 for j=1,2, . . ., J is ordinarily impossible. When the sampling fraction is small, however, and J is small, approximate satisfaction of these conditions is frequently easy to effect. Note that balanced sampling within strata provides an unbiased estimator under the more general model in which 8; varies from stratum to stratum. Such a model, even when J is small, say J=1, is frequently a good approximation to a model con- taining a quite general regression function. That is, when the intervals of B-values which define strata are narrow, a straight-line approximation within each stratum can provide a close fit to a general, smooth regression function. We see, then, that when optimal allocation (mx proportional to M,VB)) and defensive sampling within strata (sy=sx(J) for all. h) are employed, stratification produces smaller MSE’s than simple defensive sampling (s=s(J)). The next problem is that of finding good rules for stratifying a popula- tion on the size variable. For /J=1 the optimal number of strata is m with one-sample-unit-per- stratum allocation. How should the m — 1 boundaries which define the m strata be chosen? We consider a slightly more general problem: For a fixed number H of strata, given that equal allocation (m, = c, h=1, ..., H) and defensive sampling within strata are to be used, how should stratum boundaries be chosen? Under these conditions, the MSE under any model of the form &(ro, ri, . . ., r;:B), and in particular under the model of most interest, £0,1:B),is Cc EAT—T):=o* o Mi ) B & =o 2 c M, h - MB]. (14) Thus optimal stratification for equal allocation requires minimization of H S MiB». h=1 It should be noted that optimal stratification for equal allocation is not necessarily obtained by stratifying in such a way that equal allocation is optimal. Equal allocation is optimal when all My (Br)? are equal, that is, when all M, B, are equal. But it is easy to produce examples in which this way of stratifying does not minimize H S MB. h=1 It can be shown (a proof is contained in the appen- dix) that for a given stratification scheme to be optimal when equal allocation is used, it is necessary that Mi=M,= ...=M,. (15) This is established by demonstrating that for any h=1,..., H—1, whenever Mp < Mp,,, the MSE (14) is reduced if the strata are redefined so that the smallest unit in stratum A + 1 is shifted into stratum h. It is also true that, except for one special situation, for a given stratification scheme to be optimal it is necessary that B, Bj. but if we attempt to satisfy (16) by shifting a unit from the A" stratum to the h + 1%, the MSE (14) is increased. Note that in this case, shifting the unit introduces violation of the inequality M, = M,,,. The inequalities (15) and (16) indicate the essential features of a good stratification scheme for use with equal allocation, defensive sampling within strata, and the estimator (10). The strata should be so con- structed that there are more units in stratum A than in stratum h + 1, but there should not be so many more units in stratum Ah that the sum of the size measures in stratum h exceeds the corresponding sum in stratum h + 1. Three special cases in which both (15) and (16) are satisfied are: (1) M,=M,= .“. .=My, (ii) B,=8B,= oR” . = By, and (iii) MB, = M.B, = . .=MyBy. In case (i) equal allocation is proportional allocation, while in case (iii) equal allocation is optimal alloca- tion. In an obvious sense, (i) and (ii) represent two extremes among all stratification schemes consistent with (15) and (16), with all others, for example (iii), located between these extremes. It appears that the relative efficiency of (iii), with respect to the optimal scheme, is ordinarily quite nearly one. (Cf. Cochran (10].) Of course, if J = 1, yet fewer than m strata are to be created, optimal allocation requires that mj, be proportional to VM;,B,. With such allocation, optimal stratification is that which minimizes H y VM ,B,. h=1 The relation between stratified random sampling, using separate expansion or ratio estimators within strata, and the present results concerning defensive sampling within strata is quite analogous to that, discussed earlier, between simple random sampling, using either the simple expansion or the ratio esti- 15 mator, and defensive sampling. The average value of > Bj, | mn over all samples s; of size m, from stra- Sh tum h is S B), | My. Thus we should not be sur- prised to find that stratified random sampling frequently produces samples in which s, is approx- imately s, (J) for some J = 1. When this occurs the estimator is approximately (10), regardless of whether the ratio or the expansion formula is used, and is approximately unbiased under rather general models. The random sampling procedure chooses samples which are, on the average, less efficient than the nonrandom defensive strategy, as was shown to be the case when simple random sampling and defensive sampling were compared. Estimation of Variance A detailed study of variance estimation when using stratification has not been attempted here. In this section the stratum subscript & is dropped, and results are stated for an unstratified population. Of course, an example of such a population is an individual stratum in a stratified population. In this section, then, the bed size and the response associated with unit k are denoted by By and TY, respectively. Unbiased variance estimation.—Under a partic- ular model &(ro,ry, . . .,ry: v(B)), the MSE E¢(T—T)?is a measure of how much inaccuracy might be expected when T is used as an estimate of T. If T is unbiased under the model, then the MSE is simply the variance of the error T—T. For any linear estimator T= y ¢ Tk, this variance is easily calculated: Var (Sea-3 Ti)=Var (3 (xk—1)Ts +37) =o{3 (¢ x—1)" v(By) +3 v8). (17) Unbiased estimation of this variance requires simply that an unbiased estimate of o? be sub- 16 stituted for this unknown constant in (17) since for a given estimator, sample, and model, the rest of (17) is fixed and known. When using the BLUE estimator Tre, Fly « wong rs: v(B)), the usual estimator of o2, which is based on the weighted least-squares residuals, is unbiased: G2 (ro,ru, . rn(B) == 3 Ti ; (ro,ri, rvB) (Bi) (18) where Ti(ro,rvs . . .,riv(B))= 3 rrr, .« »r:0(B))Bj Z J and c=" rj, the number of regression coefficients j=0 estimated. Under the model £(0,1:B), this estimate of o? is given by ST a2(0,1:B) = — Bri |Bk. (19) a) T= SH and the MSE for a given s is estimated by ) By M a?(0,1:B) SB 2 Be (20) This statistic is not the estimate of the variance of the ratio estimate, which is usually used when s is selected by simple random sampling. As an indica- tion of the inaccuracy in an observed estimate T, (20) seems usually to be superior to the conven- tional variance estimate. For example, confidence intervals with width proportional to the square root of (20) are frequently more accurate indicators of the uncertainty in an observed estimate than are the same intervals with (20) replaced by the usual variance estimate. This point is discussed in Royall [1], where some theoretical results and empirical evidence are presented. Effects of errors in the model. —Under the more general model with regression function h(B)_and variance function v(B), i.e., Tx=h(Bx)+ €x Vu(Bk), the MSE of the ratio estimator is actually 2 Bry pii-n=r 3 v(Bk) 5 +3 Y a] fs $7 SB shin 21) The first term in (21) is the variance of the difference T—T, and the second is the square of the bias E(T —T). Under this general model the estimate (20) of the MSE has expected value (0:V+D)S BS Bel SB: (22) where v(Bx) YY v(Bk) V= 1 p——y Ge —— TT m— {= By > Bi and Lz 1 Lt “| 348 Y Bi Note that (22), like (21), is naturally represented as the sum of two nonnegative terms, the first depend- ing on the variance function and a2 but not on the regression function, and the second depending on the regression function but not on the variance. When v(B)=B the first terms of (21) and (22) are equal, and when A(B)=BB the second terms in both expressions vanish. In particular, under the model £(0, 1: B) the two expressions are equal and (20) is an unbiased estimate of E(T—T)2. If a defensive sample s(J) is chosen, the actual MSE under &(ro, ri, - :v(B)) is Ty or (1 - 9 (1-5) SHE +3 > v(Bx)IM— al which can be rewritten as oo (1-37) [Zan (23) aly 3 2B ER In this case the estimate (20) of this MSE has ex- pected value M = 1 or (1-57) B{y+ 0} (24) m Note that when v(B) is a J'"-order polynomial and s=s(J), the MSE is simply M2 m\ 1 XM or (1-31) wm 2 vB. The choice of s=s(J) protects the ratio estimator against bias in case a J'™-order polynomial regres- sion model applies. This protection does not extend to the estimate (20) of the variance of the ratio estimator, whose expected value depends, through the quantity D, on the regression coefficients. If the model ¢(0, 1 : B) is correct with respect to its specifi- cation of the variance function v(B) = B but errone- ous in its specification of the regression function, the variance estimate is biased by the amount ((24) minus (23) with v(B)= B): Mm? m (0 7 59. (25) The Cauchy-Schwarz inequality shows that D is nonnegative. Thus when we use the estimates (2) and (20) which are appropriate under ¢(0, 1: B), we choose a defensive sample s(J), and the true model is &(ro, ri, . . ., r;:B) with rj=1 for some j #1, we encounter a positive bias in our estimate of variance. When the model £(0, 1:B) is correct in its speci- fication of the regression function but erroneous in its variance function, how does the actual MSE (21) compare to the expected value (22) of our estimate? If in fact the true model is £(0, 1:v(B)), 17 then the expected value of the ratio of the estimated MSE to the actual MSE, the ratio of (22) to (21), can be written e)—ey I yr where B. & = m 2 * isdn. 1 "m=1{S vB) m& Bs and ' M RL Y Bi ey So § / SB . (26) 2 Bi; When v(B) = B?, the actual MSE is no less than = im (2 while the expected value of the estimate is no greater than 18 e-5) (3) Cr) The ratio of (22) to (21) is thus no greater than M (> BilM) / (= Bi/M — m). It is equal to this value only in fe degenerate case of all Bj equal. Thus when v(B)=B? with defensive sampling the variance estimator has a negative bias. When the sample is such that (2 BM) = Bum and v(B) =1, the ratio of (22) to (21) is no less than imag) which is no less than 1. Thus when s=s(J), and our model £(0, 1:B) is erroneous in that the actual variance function is not v(B)=B but in- stead v(B)=1, our estimator of the variance of the ratio estimator has a positive bias. PART Il. TWO-STAGE SAMPLING Description of Problem Terminology, notation.—In Part I a simple population of M units with associated size measures Bi, . . ., Bu was considered. For present purposes the units are hospitals and the size measures are their bed sizes as measured by MFI in 1963. The basic sampling unit in HDS is a patient discharge record. On this record the variable of interest, Z, is found. Thus for k=1, . . .,. M By is the bed size of hospital k; Ny is the number of discharges from hospital k during the period studied; Z,, is a number associated with discharge =1, 2, , Nk) from hospital k; and Tx is the sum, over all discharges from hospital k, of the Z-values: Vie Tr= > Z,,. =1 The sample is selected in two stages. First a sample s of m hospitals is chosen; then, if hospital k was selected, Ni is observed and a sample si of discharges is selected from hospital k&. The number of discharges in the second-stage sample is ny. The samples s and sx are represented as subsets of the sets {1, 2, , M} and {1, 2,..., Ni}, respectively. The expression ‘“k in s”’ means that hospital k is in the sample of hospitals, and “¢# in sk”’ means that discharge ¢ is in the sample of . discharges from hospital k. The objective is.to estimate the total, M Vi M -33a-gn k=1¢=1 which can also be expressed as 1-332 +334 337 27) $s 8 k Sk where 3 is the set of hospitals not in the sample, and Sk is the set of discharges from hospital £ which are not in the sample si. The first term in (27) is known from the sample; the second and third terms must be estimated. Most but not all discharges from the M hospitals are within the scope of HDS. Thus a discharge record which has been selected for the sample might be found to be either (i) out of scope or (ii) in scope but nonresponding (e.g., missing from its folder or lacking necessary information). These possibilities will be considered later, but for the moment attention is confined to the simplified case of all discharges in scope and 100 percent response. In this case the analogue of the HDS estimator [11] is > = ~> Il M y Bi (28) =v) = where Tv =Ni Y Z,, /m. 8. This estimator can also. be written as T=S 32 +3 Mi=m) (3 Zi [m) 5 & sn (3e/e) gn + (29) Sh sme Here the first term is that part of T which is known, the second term estimates the sum of the un- observed discharges from sample hospitals, and the third estimates the sum for nonsample hospitals. (Compare with expression (27).) Probability models.— The number of discharges N, is treated as a random variable whose expected value and variance are proportional to By : EN; =BB; and Var Ni =02Bk, with Ny and N; un- correlated for j # k. For a given value of Ni, the responses Zk, £ =1, 2, , Ni are treated as exchangeable random variables. That is, all of the permutations of Zi, Zi2, . . . , Ziv, have the same 19 joint probability distribution. Thus these random variables have a common mean 6, and variance o?; all the pairs (Z,, , Zi;) have the same covariance PLO Although pi and o? are treated here as constants (not depending on Nk), they might be more real- istically represented as functions of Nx and By. For Ny example, if the sum 2 Ze is fixed, then exchange- ability of the Z’s implies that, given Ni, cov (Zk, Zi) = — Var(Zke)[ (Nk — 1) for ¢ # ¢'. Thus if 02 is fixed, pr =— 1/(Nx — 1). What functions might represent the relation between 2, pk and Vi, Bx with useful accuracy and whether such representa- tions have a nonnegligible influence on the analysis are questions which call for further investigation, both theoretical and empirical. The expected values 0, 6, . . ., Ou associated with M hospitals are themselves treated as realized values of random variables ©,, ®,, . . .,0y, which are uncorrelated and have a common mean value 6 and variance 72. The random variables ®, and N; are uncorrelated for all £, j=1, 2, ., mb Optimality Considerations If all the Nx were observed and if 6, for k in s, and 0 were known, then the best unbiased estimator of T would clearly be 2 ZZ + 3 N= ne) + 83, Ne. s 8 k 5 If all Nx were observed but the 6; and 6 were un- known, then the best linear unbiased estimator of T would be 22+ (Ne— ni) 6+ 0S Ni (30) 8 Sk 3 where 72 A ken + bo2 (1+ (nk—1) pi) nk Or = % 2+ 02 (1+ (nk—1) pr) Ink > That the ©, have the same mean is an assumption whose plausibility is specific to a particular characteristic Z under consideration. For a characteristic such as length of stay, the expected value of © is probably dependent on Bx and is thus not the same for all hospitals. Sensitivity of subsequent results to deviation from the assumption of a common mean for the @, requires further investigation. At present it is uncertain as to how much deviation from this assumption can be safely disregarded. 20 and ile ST 1 This estimator was obtained in a Bayesian analysis for the case p=0 by Scott and Smith [7]. They showed that, given the observations, (30) is the expected value of T when all the distributions concerned are normal and 6 is itself given a uni- form distribution over the entire real line. They also showed that under the present model (6 fixed), (30) is the best among linear estimators whose mean square errors are bounded functions of 6. It might seem objectionable to estimate the parameter 6; for a sample hospital, not by the mean NZ nk of the sample from that hospital, sk but instead by a weighted average of this statistic and g, where g depends on the samples drawn from other hospitals. However, considering the case of 0 known and nj small will make it clear that such an estimator is quite reasonable under the present model. In (29) the expression (= Ne |S Be) 3 By estimates > Ni. Thus in the case of Nx known for all k=1, 2, .. ., estimator (29) is M, the analogue of the HDS Si, SoM > Zuty (Vem) (z Zum) $8 $s 3 Ni (3 Zul) +E SN. (3D > Nx 5 If the three conditions (i) pxr=0 for all £ in s, (ii) 72=0 (no variability among the expected values 6, . . ., By), and (iii) nx/Nyx=constant for all £ in s (proportional allocation) are met, then (30) and (31) are the same —in this simplified problem the analogue of the HDS estimator is the BLUE esti- mator. If 02=0 the two estimators differ only in their third terms. Even if the two formulas differ, they produce the same estimate when the sample is such that the sample means > Zin, k in s, Sk are all equal; they produce approximately the same estimate when the means are approximately equal. The analogue (31) of the HDS estimator is thus approximately optimal when the hospital sample means show little variability, as well as when (1)—(ii1) are satisfied. When, as is the case in practice, the Nyx, for k not in s, are unknown, the estimator obtained by replacing » Ni in (30) by its BLUE estimator Ss SBN [huis SV Ss Using the same estimate for > Ny in (31) gives (29), the analogue of the HDS estimator for this case. The conditions for equivalence of (32) and (29) are again (i)—(iii), and, as before, the two formulas produce the same estimate when the within-hospital sample means are all equal. The estimator (32) is calculable only when all of the ratios n,7°/o?, k in s, are known. For some response variables it may be known that these ratios are all quite large (or small), in which case an approximately optimal estimator can be calculated. For general values of the ratios when the n; and m are large, an approximately optimal estimator can be obtained by substituting estimates of the ratios for their actual values. This approach is not developed here. Instead the HDS-type estimator is considered, and questions of unbiasedness, stratification, and allocation are studied. The HDS Estimator Case of all discharges in scope and 100 percent response.— The HDS design is stratified, and the actual estimator is the sum of estimates of the form (28). The stratification variable is bed size. Suppose the hospitals are divided according to bed size into H strata (H can be 1), and let M, denote the number of hospitals in stratum h. Now for h=1,2, . . ., H and k=1,2, .. ., M, Bux is the bed size of hospital k, stratum h Nie is the total number of discharges from hospital k, stratum h, for [=1, 2, FO Ni; Zn is the response variable associated with discharge [ from hospital k in stratum h; Np Twe=Y, Zn is the total for hospital £, i= stratum h; and Mp, Th= Sy Thr is the total for stratum h. k=1 The underlying model is as before, except for obvious notational changes to indicate strata. For the present, attention is confined to the sim- plified problem with all discharges in scope and 100 percent response. The HDS estimator for this case is > (3 Bu) (> Tu S Bu). (33) kesy Where Toe = Nn y Zui mnie, sn is the sample of my Shik hospitals from stratum h, and sux is the sample of nme discharges from hospital k in stratum h. This Ho - estimator has the form > T) where R T= (3 Bu) (3 To / 2 Bu) Sh is a ratio-type estimator for the stratum h total T). a. Condition for unbiasedness.—The condition for unbiasedness, E(T—T)=0, applied to the estimator (33) is equivalent to 21 My, & (Zou) (seh) / 30) TR = > > E(Th). (34) Suppose E(T'nk)=E(Tw) for all k in sy. Under any model for which this is true, (34) is an unbiased estimator of the grand total T' if within each stratum the sample is “representative” in the sense that the ratio of the total expected value E(Y Ti) to Sh total beds > Bhi in the sample is the same as the Sh corresponding ratio for the entire stratum. If E(Th) is a J™-degree polynomial in Bx, the earlier results regarding defensive sampling apply. The estimator (33) is unbiased if a defensive first- stage sample sn(J) is chosen for h=1, 2, LH, and ET = ET: for all k in sn(J). If ETw= ETne=BnBnk for some constants By, then (33) is unbiased for any choice of the first-stage samples S1, $2, « . ., su. This result applies to the present Nhe model since ETw.=FE > Znct=ENnuOn= 0BB ni 1 and ETw= BN > Zou r= ENnOnx = 6BB nk. Shk b. Variance. — Under the present model the HDS estimator (33) is unbiased with Var fo S Var(T)) . =1 Using only the conditions that: (i) given Nu and Onk, the variables Zp ¢=1, . . ., Np are ex- changeable and (ii) NuOne k=1, , My are uncorrelated, it can be shown that the error variance for stratum A is Var (Tn — Th) = Var(Th) Sh > B ni ? > B ni Sh Y Var(Tn) Sh 22 3 E[Var(T um — The | Nuk, Oni) (35) The sum of the first two terms in this expression is the variance of T)—T, if Tw were observed for k in sp. The third term is the increase in error variance caused by estimation of Thy by Thx for k in sn and can be written in a more explicit form deter- mined by the relation E [Var(Twc — Tk | Nak, Oni) ] Nhk hk NZ, Nhk rl (1 a? (1— ou | (36) Expressions (35) and (36) are derived in the appendix. c. Design of survey. Allocation of second-stage samples within strata. — From (36) it is easily shown that, for a given sample of hospitals s, and a fixed total number of discharges ny to be sampled from stratum h, the error variance is minimized when nuk = nplNuk[ (1 — par) o? ef > Nue[(1 — Sh Phi Jo? ] 1/2, If the quantities (1 — pax) 02, , k in su, are approxi- hk mately equal, then optimal allocation is proportional allocation, npe/ Nuk = ni 2 Nu, for allk ins). Here the constant of proportionality is ny, / 3 2 Vix. If the constant must be chosen before the STRoiaEor of this ratio is known, then the total number of observations nj, is random. Nevertheless, if the 0? (1 —prk). k in sp, are all equal, then no other scheme for allocating the nj, observations can pro- duce a lower error variance. Choice of hospitals within strata. — When ot (1 — pn), k=1 and proportional allocation is used, the third term in (35) is a decreasing function of Sy Buk. This implies that if a first-stage Sk ., My, are all equal: sample s, for which > Bn is a maximum is sh optimal for estimating Th» in the single- stage problem (Tk observed for k in su), then it is also optimal for the two-stage problem. As in the single-stage problem, the choice of a suboptimal sample satis- Mp fying > Bui/mn = > Bnk/ My, might be jus- Sh tified on the grounds that it affords pro- tection against the errors in certain aspects of the regression model. Allocation of second-stage samples among strata.—For a given first-stage sample and proportional allocation of the second-stage sample among sample hospitals within each stratum, the third term of the error variance is H Sy h=1 Sy Bhi 5h My, 2 3 Bu 1 SE [Vn (2-1) (oh1=pw)) | Yh Sh @7) Here vy» is the sampling rate applied within sample hospitals in stratum A, i.e., nae/Nnx = yn for all k in sn. If the total expected number of second-stage units in the sample is fixed, say H * E $ y nm=E 3 Y yaNwe =n, h=1 Sp h=1 sp then what are the optimal rates vy}, v¥, . . ., v5? The answer is easily shown from (37) to be My, SY Bue [Y E(Nmoj (1 = pax) * 1 E1 Yh =X ! zm | *h 1/2 3 EN Sh h=1,2,.. ., H where \ is a constant determined by the restriction > YrE (= Nu) =n. *h If the 03,(1 — pax), k in sp, are all equal, then the y} are such that Mp yr Y BuilY Bu h=1,2,.. .,H Sp 1 are all equal. Note that when s, is such that Mp > Bhi/mp = > Bri/M Sp 1 for all h, the optimal sampling rates are determined by vysmn/Mp= constant. In other words, the optimal second-stage sampling rates are such that the overall sampling rate is the same in all strata. This is, in fact, the rule which is used stage sample. — Suppose that for h=1, 2, . . in the HDS. Optimal stratification and allocation of first- M 6) Bulma="3' Bu Ma; Sk 1 (ii) a fixed sampling fraction y, is to be used within all sample hospitals in stratum hk; and (iii) the rates vy;, . . ., yu are chosen so that the overall sampling rate is con- stant, i.e., ynmn/Mp= constant. Note that the previous analysis provides some justification for (i)-(iii). Under these conditions, if the variance of Thx is proportional to Bax (as is true when 7=0 and p=0), then the variance (35) H is of the form c, > MpBr/mn + c, for some 1 constants ¢; and c,. Therefore, the problem of optimal stratification and optimal alloca- tion of the first-stage sample (choice of my, my, . . ., my) is the same as in the single-stage sampling problem considered in Part I. Thus when the above conditions are approximately satisfied and it is not required that sp=s5(J) for J > 1 (condition wn H: 23 (i) means sp=sx (1)), the optimal number of strata is m, and optimal stratification must satisfy inequalities (15) and (16). If fewer strata (H < m) are to be created, then optimal allocation is given by the familiar rule mp/ (MB) V2 = constant. Using this allocation rule, optimal strati- H fication is achieved when > (MpBy)'? is 1 minimized. The allocation rule used in HDS is mun/Br= constant. Both allocation rules, mn/ (MyBr)V2= constant and mu/Br= constant, imply that the larger the average bed size Bn/My, the larger should be the first-stage sampling rate my/M). However, with the former rule this rate is propor- tional to (Bp/My)'2, while with the lat- ter rule the rate is proportional to By/Mh. Thus the former rule yields a more nearly constant first-stage allocation rate than does the latter. Note that with the latter, optimal stratification requires minimization H H of > Mp > Bn/m, which does not de- 1 1 pend on the way in which strata are formed. Thus when this rule is used, the choice of stratum boundaries appears to have little effect on the performance of the overall estimator. Effects of out-of-scope and nonresponse dis- charges. —1In this section it is recognized that some of the discharges from which the sample is drawn might be outside the scope of the HDS study. A two- valued variable § is used to indicate whether or not a discharge is in scope: &n=1 if discharge ¢ from hospital 4 in stratum h is in scope, and 8x=0 otherwise. The variables én ¢=1,2, . . ., Nu are treated as realized values of independent random variables, and mu; is the probability that n= 1. The response Zi can now be represented as the product of 8p; and a random variable Xj... Then the X-value is the characteristic of interest and nk Nhk Th= y Zh= > SniiX ni. Given the number Nh Nn of total discharges, > dni represents the 1 random number of in scope discharges from hospital hk. The expected response of each in-scope discharge from hospital hk is denoted by wn; i.e., par is the expected value of Xu, given that 8uu=1. Thus EZ y= 60n= EdniiXnii= mwhipun. The variances and covariances and responses of in-scope dis- charges are o%,, and px, 0%,,. Thus Var Zni=0%,= Var (SnxilXnri) = Thk0%,, + Tk(1 — Thi) and Cov (Znkts Zuri’) = pra = Th iPXnkOx pr The most direct estimator for this case, which is the analogue of the HDS estimator, is T= > Pos 1 in which -3 > SniiXnrit 2 om sh shk Y Nuit nnpann Sh CSB. 2 Sh — nuk) warns + where Thi= Nnk y Suit X nit kes Thee = > Snktl nk, Shk Shik and Mik = Sy ma > Snki- Shk Shk Note that this is the same estimator as (33). The response Zn has simply been expressed as the product 8x Xni. Similarly, the previous analysis and results remain valid; the parameters On, 07, , and pp in the previously stated formulas are simply recognized as functions of the parameters in the distributions of the more fundamental random variables 8 and Xp. Thus the previously derived error variance (35) can be expressed in terms of the parameters in the present, more detailed model, as follows: Var Z. -T)= > Var (Thr) Sh 2 Bri\? os Bhs 2 Var (Thr) - P(E +m(l—m) pu>— 7 pxo% } (38) The subscript hk is placed outside the braces in lieu of its being used repeatedly with N, n, m, 0% and py inside the braces. In this variance expression Var (Th) =Var (Nnmnepns) +E [{N (mo? +7 (1—m) u2) +N (N—1) mp, 0% ni] forall k=1,2, . . ., Mh. The situation is complicated by the introduction of nonresponse. A second indicator variable is employed to denote response status: for ¢ in sn, {mi=1 indicates response, and {mi =0 indicates nonresponse. Thus for ¢ in snk, Znki= {nkiSnkiXnkr is observable, and the problem is to estimate H Mp Np 2 2 > SniiX nit, the sum of X-values over all in- ope Sscharmes, The response indicator variables are treated as random, independent, and inde- pendent of all the other random variables present. For an in-scope discharge from hospital hk, the response probability Pr({u= 1) is denoted by ¢ns. It is assumed that each selected discharge can be classified as in scope or out of scope, even if it is nonresponding. That is, of the na: discharges selected in the sample from hospital hk, the number of in-scope discharges > nx is observable. Of the Shk LED dni in-scope discharges, only a random Shk number, m= Lnkidnir, will respond. A direct Shik Nhk estimate of Thy = 2 SniiX nit is clearly The= > CnwidnceX nia + (nhke— hse) fhonk Shik + (Nk — nak) Think (39) where fine=3 LucBrssXiw ig and =n [ne Shik The first term in (39) is the observed sum of the in- scope, responding sample discharges. The second term estimates the sum of X-values for the (np —nnx) discharges known to be in scope but unob- served because of their nonresponse. The third term is the product of (Nx — nak) nk, the estimated number in scope among the Nx — nar nonsample discharges, and the estimated average response funk. The estimate can also be written in the more compact form , 0 CnwedniiX ni 2 n hk Shk The= Nn = n'n ~ ~ = Nn Ahi Nhe Mhk If the stratum total T) is estimated by a ratio- type statistic employing these estimated hospital totals, and if the grand total is estimated by the sum of a H A these estimated stratum totals, T=Y 1, then the 1 resulting estimator is that used in HDS. Note that it is possible to have no responding in-scope discharges in the selected sample su, i.e., nj, =0. In such a case if the simplest natural course is taken and Th: is defined as zero, a small bias appears. Given the values of Nk, nk, nk, and ni, the expected value of Thy is simply Nix mhiini, while the expected value of Thi is Nuimrnepens [1— (1 — oni) (1— mnieponi) "m=". For all except extremely small values of mn: and on and small nay, the bias is clearly negligible. The error variance of T can be shown, by tedious but straightforward calculations, to be given 25 approximately by an expression of the same form as (35): Var (T-T) =! = Var (Thi) + 1 by Var (Tu) | +3 SE > E [Var (Tnx — Tax Nak» Oni) 1. The error incurred in using this approximation for the true error variance arises from the slight bias in T and is negligible whenever the bias is. Similarly, 26 the last term in this variance is given approximately by 2 Cn ser (0-7) (mo +7 (1—7) u2—n*pr0%) hue | when the nonresponse probabilities 1— en. are all small. Thus the earlier results concerning alloca- tion remain relevant when a small probability of nonresponse is present at the second stage of sampling. For the case of sizable nonresponse probabilities, the estimator should be reexamined and various alternate estimators considered in which 7x and unk are estimated by linear functions y nine and y ChikfLhk- Sh Sh REFERENCES [1] Royall, R. M.: Linear Regression Models in Finite Population Sampling Theory, in V. P. Go- dambe and D. A. Sprott, eds., Foundations of Statis- tical Inference, Holt, Rinehart, and Winston, Ltd., of Canada, 1971. [2] Royall, R. M.: On finite population sampling theory under certain linear regression models. Biometrika 57(2):377-87, Aug. 1970. [3] Ericson, W. A.: Subjective Bayesian models in sampling finite populations (with discussion). Journal of the Royal Statistical Society, Series B, 31(2):195-224, 1969. [4] Ericson, W. A.: Subjective Bayesian models in sampling finite populations: stratification, in N. L. Johnson and H. Smith, Jr., eds., New Developments in Survey Sampling. New York. Wiley-Interscience, 1969, pp. 326-357. [5] Kalbfleisch, J. D., and Sprott, D. A.: Applica- tions of likelihood and fiducial probability to sampling finite populations, in N. L. Johnson and H. Smith, Jr., eds., New Developments in Survey Sampling. New York. Wiley-Interscience, 1969. pp. 358-89. [6] Brewer, K. R. W.: Ratio estimation and finite populations: some. results deducible from the as- sumption of an underlying stochastic process. Australian Journal of Statistics 5(3):93—-105, Nov. 1963. [7] Scott, A., and Smith, T. M. F.: Estimation in multi-stage surveys. J.Am.Statist. A. 64327): 830-40, Sept. 1969. [8] Rao, C. R.: Linear Statistical Inference and its Applications. New York. Wiley, 1965. Chapter 4. [9] Cochran, William G.: Sampling Techniques. New York. Wiley, 1963. p. 174. [10] Op. cit., p. 131. [11] National Center for Health Statistics: Design of the NCHS Hospital Discharge Survey, by Simmons, W. R., and Schnack, G. A. Rockville, Maryland, 1969. Unpublished report. 27 APPENDIX Derivations of Conditions on Optimal Stratification with Equal Allocation and Defensive Sampling M,=M,= . .. =My at optimum.—1It will be shown that whenever M,, < My, forany h=1,2, . . ., H—1, the MSE (12) is reduced if the smallest unit in stratum h+1 is shifted into stratum h. The desired result follows directly from this fact. No generality is lost if attention is restricted to the case of h=1 and H=2. Let BW=< , . =< BM, +M,) he the size measures Bui, k=1, 2, . . ., Mu h=1, 2 arranged in nondecreasing order. Then BY, B®, , BM) are the sizes of units in stratum 1 and BWMi+1), BMy+2), , BOi+3,) are My My the sizes of units in stratum 2. > Bu=Y B® and 1 My M+ M, S Bu= 5 B®. The MSE is 1 M; +1 M, —m, My — my, Mi+M, 2 [eee (kK) ———— (k) o Sy B® + - Sy B 1 Mi+1 My My +My & |, S B® + M, S Bw] M, +1 M +M, S$ iy: 10) 1 because m= m.,. Now if the smallest unit in stratum 2 is shifted into stratum 1, the new MSE is (40) a? My+1 = [on +1) : B® + (My — 1) My+My Mi+M, 3 Bw] oe 2 Bk). M+2 (41) The difference of (40) minus (41) is proportional to My My+M, (My —M, — 2)BW1+1) — > B+ 3 B®) 1 Mi+1 which is = (M,—M,—2) B®™ +D—M,BM +1 +M,BM +1 =0 since M; < M.. The first inequality is strict unless all the B(*) are equal. Bi Bp for any h=1, 2, . . ., H — 1, the MSE is reduced if the largest unit in stratum h is shifted into stratum A + 1 unless such a shift forces violation of Mj, = My,,. As before, no generality is lost by restricting attention to the case of two strata. By the same basic argument used before, it can be shown that shifting the largest unit in stratum 1 into stratum 2 reduces the MSE by the factor M, MM, > B® + (M,—1)B™) — > B®) 1 My +1 — (Mz+ 1)BMY, which is positive when (M; —M,—2)B®) MM, > Bn -3 B0=5.~B. M +1 Since by assumption B, — B; < 0, the shift results in a positive reduction in the MSE if M, —1=M,+1. Note that M,—1 and M.,+1 are the sizes of the new strata. Derivation of Expressions (35) and (36) for Variance For convenience, the conditional expectation and variance of a statistic Y, given Onx, Npx k=1, . . ., My, are denoted by E(Y|C) and Var (Y|C), respectively. Then Var (Tn — Tx) = Var [E(Th — Ta |C)] + E[Var (Th —Tx|C)]. (42) ) My, Since Th — Th= > Bhi Sy Tm! S Bi 1 Sh Sh _ 3 Pip = sp > T ni ’ and, Sh 29 independent of Thy, the second term in (42) is equal given C, Thi, and Ty are and Thy, for k# Kk’, to f| S Var(Tw | €) +S Var (Ti | €) Sh Sh Mp S Bu : - 2 S Bu > Cov (Th, Ti | C) ke Sh M), 2 = > Bu 1 3X Bhi Sh + 3 Var(Tu | €)|. (43) Sh But from the exchangeability of the Z’s, given C, it is easily shown that Cov (Tus Tr: | C) = Var(Tuw: | C) Therefore, (43) can be rewritten E 3 Var(Tw | €) + Var(Twx | C) s sp “h My 3 Bu CSE Bh 3 Var(T | ©) My, 2 3 Bu 1 |S S Var(Tw | €) Sp ' Th My, 2 Y Bh + SB S$ (Var(Tw | €) — Var(Tw | €)) Sp *h =F | > Var (Thx | C) L +a 30 Mp 2 SY Bm —~ Tw | C + v} S Bn 2 Var nm | C) hn M), 2 Y Bu + SB SE [Var(Tu | €) : LL —Var(Tw | C)]. (44) If the relation Cov(Thi, Toi | C) = Var(Tu? | C) is applied to the final sum, then after some rearrange- ment (44) can be rewritten as > Thi E | Var Sn YS Bu — > Tw |C Sh Th Sh M, 2 y Bh 1 > Bhi Sh whe SE [Var(Tw—Tw | C)I. (45) Sh The first term in expression (42) for Var a, —Th) is > Bhi Var [E(T) — Th|€)] = Var{ =— 3 Nu®n > Bh Sp Sh ws > Nni® ni Sh > Th k = Var |E{> > Bhi Bn 3, Sh — > Tw. |C) |. (46) h Adding (45) and (46) yields (35). Now Var (Twi — Til €) =E[(Tw — Thi)? |C] N hk 2 = E| (Vue > Z niin — y Zn) lc] ’ nk ! which is, by exchangeability, the same for all samples sp; containing nny units. This quantity is thus the same as N 1 hk 2 iB 37 (Vu 3 Zudon = 3 zw) |c] Rh * where 3* indicates summation over all the ( Nhk samples spi of size np. Interchanging the order of summation and expectation in this last expression establishes (36). #* U. S. GOVERNMENT PRINTING OFFICE : 1973 515-212/59 31 Series 1. Series 2. Series 3. Series 4, Series 10. Series 11. Series 12. Series 13. Series 14. Series 20. Series 21. Series 22, VITAL AND HEALTH STATISTICS PUBLICATION SERIES Originally Public Health Service Publication No. 1000 Programs and collection procedures.— Reports which describe the general programs of the National Center for Health Statistics and its offices and divisions, data collection methods used, definitions, and other material necessary for understanding the data. Data evaluation and methods reseavch.—Studies of new statistical methodology including: experi- mental tests of new survey methods, studies of vital statistics collection methods, new analytical techniques, objective evaluations of reliability of collected data, contributions to statistical theory. Analytical studies.—Reports presenting analytical or interpretive studies based on vital and health statistics, carrying the analysis further than the expository types of reports in the other series, Documents and committee veports.—Final reports of major committees concerned with vital and health statistics, and documents such as recommended model vital registration laws and revised birth and death certificates. Data from the Health Interview Survev.—Statistics on illness, accidental injuries, disability, use of hospital, medical, dental, and other services, and other health-related topics, based on data collected in a continuing national household interview survey. Data from the Health Examination Survey.—Data from direct examination, testing, and measure- ment of national samples of the civilian, noninstitutional population provide the basis for two types of reports: (1) estimates of the medically defined prevalence of specific diseases in the United States and the distributions of the population with respect to physical, physiological, and psycho- logical characteristics; and (2) analysis of relationships among the various measurements without reference to an explicit finite universe of persons. Data from the Institutional Population Surveys. — Statistics relating tothe health characteristics of persons in institutions, and their medical, nursing, and personal care received, based on national samples of establishments providing these services and samples of the residents or patients. Data from the Hospital Discharge Survey.—Statistics relating to discharged patients in short-stay hospitals, based on a sample of patient records in a national sample of hospitals. Data on health resources: manpower and facilities. —Statistics on the numbers, geographic distri- bution, and characteristics of health resources including physicians, dentists, nurses, other health occupations, hospitals, nursing homes, and outpatient facilities. Data on mortality.—Various statistics on mortality other than as included in regular annual or monthly reports—special analyses by cause of death, age, and other demographic variables, also geographic and time series analyses. Data on natality, marriage, and divorce.—Various statistics on natality, marriage, and divorce other than as included in regular annual or monthly reports—special analyses by demographic variables, also geographic and time series analyses, studies of fertility. Data from the National Natality and Mortality Surveys.— Statistics on characteristics of births and deaths not available from the vital records, based on sample surveys stemming from these records, including such topics as mortality by socioeconomic class, hospital experience in the last year of life, medical care during pregnancy, health insurance coverage, etc. For a list of titles of reports published in these series, write to: Office of Information National Center for Health Statistics Public Health Service, HSMHA Rockville, Md. 20852 Completeness and Quality of Response in the North Carolina Marriage Follow - Back Survey U. S. DEPARTMENT OF HEALTH, EDUCATION, AND WELFARE Public Health Service Health Services and Mental Health Administration Vital and Health Statistics-Series 2-No. 56 For sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, D.C. 20402 Price 75 cents domestic postpaid or 50 cents GPO Bookstore Series 2 DATA EVALUATION AND METHODS RESEARCH Number 56 Completeness and Quality of Response in the North Carolina Marriage Follow - Back Survey A pilot survey to study response rates and quality of data from mail follow-back surveys linked to marriage records. DHEW Publication No. (HSM) 73-1330 U.S. DEPARTMENT OF HEALTH, EDUCATION, AND WELFARE Public Health Service Health Services and Mental Health Administration National Center for Health Statistics Rockville, Md. July 1973 NATIONAL CENTER FOR HEALTH STATISTICS THEODORE D. WOOLSEY, Director EDWARD B. PERRIN, Ph.D., Deputy Director PHILIP S. LAWRENCE, Sc.D., Associate Director OSWALD K. SAGEN, Ph.D., Assistant Director for Health Statistics Development WALT R. SIMMONS, M.A., Assistant Director for Research and Scientific Development JOHN J. HANLON, M.D., Medical Advisor JAMES E. KELLY, D.D.S., Dental Advisor EDWARD E. MINTY, Executive Officer ALICE HAYWOOD, Information Officer OFFICE OF STATISTICAL METHODS MONROE G. SIRKEN, Ph.D., Director E. EARL BRYANT ,M.A., Deputy Director Vital and Health Statistics-Series 2-No. 56 DHEW Publication No. (HSM) 73-1330 Library of Congress Catalog Card Number 73-600033 FOREWORD This is a report on a pilot survey of recently married persons that was conducted for the Na- tional Center for Health Statistics by the Univer- sity of North Carolina to test procedures for con- ducting follow-back surveys linked to marriage records. Dr. Bradley Wells and Dr. Elizabeth Coulter, Department of Biostatistics, School of Public Health, were the project director and dep- uty project director, respectively, and Dr. Monroe Sirken of the Center was the projectofficer. Mary Grace Kovar of the Center edited the final manu- script and worked with the Office of Information in preparing the report for publication. The methodology for conducting follow-back surveys was initially developed by the Center for surveys linked to death records and subsequently the methodology was applied to surveys linked to birth records. The developmental work ultimately resulted in a continuing statistical program for conducting sample surveys linked to birth and death records. The objective of the vital record survey program has been to expand the scope of national natality and mortality statistics beyond the items of information on the vital records themselves. There is also a need to expand the scope of national marriage statistics in order to measure trends and differentials in various phenomena as- sociated with the family, This need was recognized in a reporton''Needs for National Studies of Popu- lation Dynamics'' prepared by the U.S. National Committee on Vital and Health Statistics,! which states that ""A marriage follow-back survey would provide a great deal of the data that is currently lacking." It was also recognized in the report "Population and the American Future'' prepared by the Commission on Population Growth and the American Future, which recommended that the National Center for Health Statistics should: Undertake a crash program to qualify all States to participate in the marriage and divorce registration areas; to institute follow-back surveys for sample of mar- riages and divorces, such as the present natality and mortality follow-back sur- veys; to develop information sources on family formation and dissolution, and the fertility and other demographic conse- quences of family dynamics, The results of the pilot study in North Caro- lina are encouraging with respect to developing procedures for conducting follow-back surveys linked to marriage records. The overall response rate including personal interview follow-up of nonrespondents to the mail survey was about 80 percent. This rate is lower than the response rate (89 percent) in the national surveys linked to records of legitimate births, and it is also lower than the response rate (90 percent) in the national surveys linked to death records. The adequacy of information reported in the mar- riage follow-back survey compares favorably with that reported in follow-back surveys linked to birth and death records. In follow-back surveys, files of registered vital events serve as the sampling frames. In- formants who provided the information recorded on registration certificates are generally the sources of information queried in the survey. For instance, the brides and grooms are the informants for items recorded on marriage certificates and they would be the sources on information in the follow-back surveys linked to marriage records. Fortunately, the names and addresses of both the bride and groom are recorded on the North Car- olina marriage certificate, Although the addresses of both bride and groom are items of information recorded on the U.S. Standard Certificate of Mar - riage, these items do not appear on the marriage certificates being used in all States. In 12 States, neither the bride's nor groom's mailing address is on the marriage certificate. In 11 of the 12 States however, the local registrars are identified on the marriage certificates and it is possible that the mailing addresses of the bride and groom could be obtained from them. A comparable problem arises in surveys linked todeath records because the address of the death record informant is sometimes missing on the death certificate. The information is invariably obtained in the follow-back surveys linked to death records by writing to the funeral directors who are always identified on the death certificates. Before planning a national program of surveys linked to marriage records, a feasibility study should be conducted to test procedures for getting the addresses of brides and grooms from local registrars in those States where the-addresses do not appear on the marriage records. It would be appropriate to take that occasion to test additional procedures to enhance the survey response rates of brides and grooms. Monroe G. Sirken CONTENTS TF OTLWOLTL. ww 0 ir on sr om mw 0 35 mee 0 es 0 INtrodUCHion === === === mmm emo meme Background = === === == mmm mm mem mm eon eee ODbjJECtiVES == === === =m mm mm mmm ee eee eee eee Study DeSign-= === === === mmm mmm mm oe eee QUEStIONNAITES ~~ = === === === mm mmm mm eee The Study Population and Sample ~-----=---=---cmmmmmmmm Experimental Variables ---------=c ooo mmmmm momen Mailing Procedures====-======- == mmm mmm eme mmm Interview Follow-Up === === mmm m moe meee eee eee Response Rates === === mmm mmm momo o ooo Unweighted Mail Response Rates and Amount Added by Interview -------- Weighted Estimates of Mail Response Rates ---======---ooooooooooooo The Joint Effects of Duration and Questionnaire-------=-===--=ccoooo--o The Joint Effects of Questionnaires and Demographic Variables --------- Comparison of Respondents and Nonrespondents to Mail Survey From Interview Data ---======cccmm mere meme mm Quality Of Data-==== === mmm meee meme eee Adequacy of Mail Questionnaires Before and After Requery ------------- Completeness of Response to Individual Items --===-=-=====---occco--- Consistence of RESPONSES ===========m ooo mmmm mmm SUMMATY === === === = =m = = = mm mm eee ee meee References ---=-==mmmmmmmemcm cee cc em mmm mmm mmm mmm mmm me mmm mmm mmm mm LAST OF DELATIEA "TADLGE eve rw ee rer mo: rm mm mo mr rm 0 0 mt em Appendix 1. Forms Used in the Study---=-====-=======cccoooooooooooo License and Certificate of Marriage-1968------------ncmmommmocommnooo License and Certificate of Marriage-1969-=----------occoemmmooonmn Covering Letter for Basic Questionnaire-—---=-=---coommommmmmmoooooo Items Common to All Mail Questionnaires--=-----==ecooommcmomooo- Health Care Questionnaire Only----=======cmcmommmom moomoo Family Planning Questionnaire Only---=======-==mccooooommmooooooo Appendix II, Sampling Procedures, Methods of Estimation, and Standard EXrOrS-----cmom memo moo m eee Sampling Procedures -==--=-=-==cocmooommmooooooon TE HE Multinomial Model for Stratum Response Rates and Variances ----------- Limitations of Variance Estimates -----====ccommmmmmmcmce eee Weighted State Estimates of Response Rates and Standard Errors -------- Appendix III, Definitions of Certain Terms Used in This Report ----------- Appendix IV, Estimated Amount Added by Interview Follow-Up of Refusals and Nonrespondents --------=-------m-moommmm mcm mmmemom moomoo — ar no Ww wh NN NO 0 NNO O&O - o Ww ho NDN WH on Www 34 34 36 37 37 39 40 vi SYMBOLS Data not available-----ecemammcmmemaoo En Category not applicable-==c-ceomaoacoaoo Quantity zero---------ccmemcmmmomeoan - Quantity more than O but less than 0.05----- 0.0 Figure does not meet standards of reliability or precision-----=---cocoooooo COMPLETENESS AND QUALITY OF RESPONSE IN THE NORTH CAROLINA MARRIAGE FOLLOW-BACK SURVEY H. Bradley Wells, Ph.D., Elizabeth J. Coulter, Ph.D., and Linda S. Wienir,’ INTRODUCTION Background In 1956 the National Office of Vital Statistics, now a part of the National Center for Health Statistics, began a program of research on the methodology for mail follow-back studies linked to birth and death certificates. This led tothe os. tablishment of a National Mortality Survey i in1961° and a National Natality Survey in 1963,* and both have become effective means of supplementing national birth and death statistics. In 1967 NCHS, as part of its continuing meth- odology research program, contracted with the Department of Biostatistics at the University of North Carolina to conduct a pilot study of the fea- sibility of using mail follow-back surveys based on marriage records for collecting supplementary marriage statistics, The Research Triangle In- stitute's Division of Statistics subcontracted to trace S00 of the brides selected for the mail sur- vey and conduct interviews. The North Carolina State Board of Health agreed to make the mar- riage records and punchcards available, The 1968-69 North Carolina study was con- ducted in two phases: (1) a pretest for develop- ment of questionnaires (Spring 1968) and (2) a pilot survey to study response rates and to eval- uate the quality of responses (November 1968- June 1969). In both phases samples of brides married in the State were selected from marriage a . . . . . . . Ms. Wienir is now at Western Michigan State University. University of North Carolina records filed with the North Carolina State Board of Health and were sent mail questionnaires. In the second phase—the pilot survey—samples of both respondent and nonrespondent brides living in six central North Carolina counties were traced and personally interviewed as part of the effort to evaluate the quality of data collected and to examine potential biases among mail nonre- spondents, A brief report of the pretest results has al- ready been made” The present report is re- stricted to presentation of pilot survey results. Prior to 1967 two other mail follow-back sur- veys inked to marriage records had been done. Pratt’ used the method in studying records of marriages which occurred in the Detroit metro- politan area during 1960. Coulter” carried out a small pilot survey of recently married couples in North Carolina in 1966. Objectives The broad objective of the North Carolina study was to investigate the completeness and quality of data obtained in mail follow-back sur- veys of recently married brides, Specific objec- tives were to: Estimate differences in response rates by: Age, race, and previous marital status of the bride Time duration since marriage Questionnaire content Certified and regular mail delivery. Determine biases due to nonresponse. Investigate the quality of data by: Comparing the consistency of data from dif- ferent sources Examining the completeness of data for in- dividual items in the mail questionnaire. STUDY DESIGN Questionnaires Mail and interview questions were directed to the bride but included information about both the bride and the groom. Four mail questionnaires were pretested in a five-county area of central North Carolina, Re- visions were then made and the following three questionnaires were used in the pilot survey: A basic version which included the same demo- graphic content as the marriage record and ad- ditional questions on income, employment status, religion, residence prior to and after marriage, and household composition. A family planning version which included the same content as the basic version plus a one-page series of questions on number of children desired, whether currently pregnant, when the next child was expected, and contraceptive use by the couple. A health care version which included the same content as the basic version plus a one-page series of questions on current pregnancy status, pre- natal care, hospital care since marriage, and health insurance coverage, Facsimiles of the mail questionnaires are given in appendix I. The interview questionnaires were designed to collect the same information in essentially the same sequence as in the mail questionnaires, Ad- ditional items were added to the interview ques- tionnaire for control purposes and to obtain re- actions of the respondents to specific aspects of the mail survey, The Study Population and Sample The 48,162 marriage records filed with the North Carolina State Board of Health during the 12-month period February 1968-January 1969 constituted the pilot study population and sampling frame (table 1). In five-sixths of all marriages the bride was white, In two-thirds of the marriages the bride was white, never married, and under 30 years of age. Only 4.4 percent of the brides were 45 years or older at the time of marriage and 90 percent of those had been previously married. About one-fifth of the marriages occurred in Alamance, Durham, Forsyth, Guilford, Orange, or Wake Counties, Marriages in these six counties, all readily accessible to interviewers from the University of North Carolina and the Research Triangle Institute, were used to select the sample for personal interviews, Because all comparisons between mail and interview data are based on marriages in these counties, estimates for them are shown separately in the tables, Geographic stratification was done to assure sufficient numbers of mail respondents and non- respondents for interview follow-up in the six- county area, Stratification by previous marital status, race, and age of bride was deemed es- sential because of differential mail response rates observed in other studies. In the pretest, many older brides objected to the prenatal care and family planning questions, hence, in the pilot study, only the basic question- naire was sent to brides 45 years ofage or older. To simplify analysis and presentation of results, this report is restricted to results for brides under 45 years of age unless otherwise noted, thereby reducing the 32 strata in table 1 to 24 strata, Also due to small numbers, previously married brides under age 20 of races other than white are omitted in both geographic strata, fur- ther reducing the number of strata to 22 for this report. The sample design called for equal numbers in each stratum within each area, 144 ineach six- county stratum, and 216 in each rest-of-State stratum and this required different sampling frac- tions, A balanced design could not be achieved, however, because of the limited numbers insome strata, Effective sampling fractions in. the 22 strata ranged from 100 percent downward to 1.4 percent, Deviations from the sample design tended to occur in those groups which subsequently had low response rates, As aresult, somewhat tedious analytical procedures which are described in ap- pendix II along with the sampling procedures were required. Definitions of terms used in this report are given in appendix III, Experimental Variables Within each of the 22 demographic strata, four experimental variables were employed in further poststratification: Four time durations,—3,5,7, or 9 months between marriage and first mail follow-up. Three questionnaives,—Basic, family planning, and health care. Two alternate addresses.,—Either the bride's or the groom's as shown on the marriage license for first mail query. Two types of mail, —Certified and regular for the second query to nonrespondents from the first mailing, The time duration and version of the questionnaire to be sent to each bride were randomly assigned attime of sampling. Within each month of mailing, the choice of the bride's or groom's address for the first mailing was made by alternate assign- ment, The type of mail used for the second query was randomly assigned to each nonrespondent bride 15 days after the first query. Mailing Procedures Roughly 600 initial questionnaires were mailed on the first Monday of each month for 6 months, November 1968-April 1969. Response patterns for each month were similar and com- bined results are shown in the tables. Preliminary analysis also indicated that mail response rates were similar for addresses of brides and grooms; hence this variable is not considered in the pres- ent report, The first query was sent by first-classmail. Two weeks after the first mailing, all brides for whom no response had been received were ran- domly subdivided into two groups for the second mailing. To one group the second query was sent by certified mail and to the other it was sent by regular first-class mail. Two weeks later third queries were sent by regular first-class mail to all remaining nonrespondents regardless of what type mail had been used for the second query. A stamped, addressed return envelope was included in each query. When the Post Office returned a query in- dicating that it could not be delivered, another first query was immediately mailed to the alter- nate address on the marriage record if one was shown. After one or two undelivered letters (Post Office returns), if there was noother address, the bride was classified as a nonrespondent. Questionnaires returned with an indication that the sample bride did not wish to cooperate were classified as refusals and no further mail follow-up was made, For estimation purposesre- fusals and nonresponses are usually put together, Completed or partially completed returned questionnaires were classified as responses. Every response was edited within 3 weeks of receipt for completeness and internal consistency and a single requery was sent to the respondent asking for clarification of items judged to be in- adequate. Information from returned requeries was added to the original return and quality esti- mates are based on all data, Interview Follow-Up Six weeks after the first mail query each sam- ple bride was classified as respondent, nonre- spondent, or refusal. Five refusals which were especially strong were excluded, and then samples of mail respondents and nonrespondents and all other refusals for whom the most recent address was within the six-county area were taken for attempted follow-up and personal interview, Sampling for interviews was done separately within the mail respondent and nonrespondent groups for each month of mailing. A sample of 84 brides (42 refusals and nonrespondents and 42 respondents) was to be taken from each of the 6 months’ mail results—a total of 504, The total actually selected was 447 after excluding 43 brides 45 years or older. Of this total, 289 were found and interviewed, as shown in table A, RESPONSE RATES Unweighted Mail Response Rates and Amount Added by Interview Unweighted mail response rates for the major study variables are shown in table 2, Overall first-mail response was about 25 percent and was significantly low for: Brides 30-44 years of age Brides of races other than white Brides who had been previously married Brides married 9 months prior to the first mail Brides married outside six central coun- ties. However, after two follow-up mailings, the cumu- lative response was increased to 59 percent for those sent certified mail and to 52 percent for those sent regular mail at the second mailing. Response for all mailings was significantly low for: White brides among those sent certified mail Brides 30-44 years of age regardless of type of mail Brides who had been previously married re- gardless of type of mail. While overall refusal rates for the total mail survey were relatively low—3.9 percent for cer- tified mail and 2.9 percent for regular mail, there is some indication in table 3 that certified mail served as a stimulus to refusal as well as to response. The second-mail refusal rate for certified mail was significantly higher than that for regular mail—3.4 percent versus 1.2 per- cent, Second-mail refusal rates were significantly higher for certified than for regular mail for white brides, for those 30-44 years ot age, for those who had been previously married, and for brides who were sent the family planning questionnaire. Response rates for the sample are low in relation to the weighted estimates described in the next section, but even so most of the sample differences remain significant after the weight- ing procedure. Before discussing weighted esti- Table A. Number of brides and interview rates by results of the mail survey for six- county area: North Carolina Marriage Survey, 1968-69 Mail survey results Interview results Total Non- Respondent respondent Refusal Number Totalmmeemcmcm cmc ccmcceeeeeem 447 233 173 41 Interviewed=====cecmcccccccccceeeeam 289 187 92 10 Not interviewed----=-ccecccmcacaaaaa 158 46 81 31 Percent Interviewed-=---cccmcmmmcc cece 65 80 53 24 mates, however, it is appropriate to consider the increase in response rate due to personal inter - view of mail survey nonrespondents and refusals. Detailed analysis of the number of sample persons added by interview is not possible be- cause of the small numbers involved, only 173 nonrespondents and 41 refusals having been in- terviewed from the six-county area. For the un- weighted six-county sample, interview follow-up increased the certified mail response from 59 to 77 percent and the regular mail response from 54 to 80 percent (table 3), The amounts added by interview, however, are not significantly different by type of second mailing. See appendix IV for a more detailed discussion of weighted results, Although the differences are not statistically significant, there is some indication that the num- ber added by interview may be negatively cor- related with mail survey response rates. Weighted Estimates of Mail Response Rates Individual stratum estimates were weighted to obtain estimates of response rates which would have been expected with uniform sampling rates from brides married in North Carolina during February 1968-January 1969. In discussing table 4, differences due to other variables which might influence summary results were not considered. For example, the age distributions of never married and previously married brides are very different, the latter being considerably older than the former. In the following sections differences in response due to the joint effect of some of the major variables are examined in somewhat more detail using weighted estimates. The Joint Effects of Duration and Questionnaire Response rates specific for time duration since marriage and the version of the question- naire which are shown in table 5 generally fol- low the trends seen in table 4. For each mar- riage duration the family planning questionnaire response rate for certified mail was lower and the difference between certified and regular mail was less than for the other two questionnaires. However, the certified mail response rates for the family planning questionnaire were signifi- cantly lower than for the other questionnaires only at 7 months duration. The Joint Effect of Questionnaires and Demographic Variables Other weighted rates are shown in tables 6-8, Rates in these tables are interrelated and will be considered together in discussing several vari- ables. Marital status.--In all possible pairwise comparisons of response rates for brides who had not been married before with those who had been previously married within each type of mail group, all 15 rates in the first mailing, 13 of the 15 in certified mail, and 14 of the 15 in regular mail were higher for the brides who had not been mar- ried before (table 7). The three exceptions were among brides 30 years or older of races other than white, This, coupled with the summary rates in table 6 by kind of questionnaire and in table 8 by age, clearly shows that brides who had not been married before responded at significantly higher rates. Color.-—-Pairwise comparisons of white brides with brides of other races within age, ques- tionnaire, and type of mail groups (table 7) and the summary rates of tables 6 and 8 provide no con- clusive evidence, Brides of races other than white tended to respond to the first query at lower rates than white brides but responded at higher rates to second and third queries, especially with cer- tified mail, Among brides under age 20 who had not been married before, brides of races other than white responded to the family planning questionnaire at significantly higher rates to both certified and regular mail than did white brides (table 7). Rates for previously married brides age 30 or older were also generally higher for brides of races other than white than for white brides with both certified and regular mail, and a number of these differences were statistically significant (table 7). Brides of races other than white aged 30 or older who had not been married before re- sponded at a significantly higher rate to certified mail than white brides of the same age and mar- ital status (table 8). Version of questionnaive,— For certified mail the overall poor response rates tothe family plan- ning questionnaire in relation to the other two ques- tionnaires is due mainly to low response rates for white brides under age 20 who had not been mar- ried before and the heavy weight assigned to this group of brides in calculating weighted rates. As described above, the response rates of white brides to the family planning questionnaire were generally low, Age at marriage,— Response rates for brides aged 30-44 were significantly lower than for brides under age 30 in most of the triple com- parisons in tables 7 and 8. Except for white brides under 20 now married for the first time who re- ceived the family planning questionnaire, response rates generally were highest for brides under 20, slightly lower for those aged 20-29, and much lower in the 30-44 age group. Type of mail.—The overall significantly higher response for certified mail over regular mail was due wholly to the better response to the basic and health care questionnaires in most color-marital status groups (table 6). For brides aged 20-29 who had not been married before, a group which counts heavily in calculating weighted rates, regular mail yielded slightly (but not sig- nificantly) higher response for the family plan- ning questionnaire while certified mail response rates are considerably (but not significantly) higher for the two other questionnaires. Signifi- cantly lower response rates for the family plan- ning questionnaire than for the other question- naires with certified mail also appear for pre- viously married white brides under age 20 and previously married brides of other races aged 20-29 (table 7). Among brides of other races aged 30-44 years, response for certified mail was significantly better than for regular mail for the basic questionnaires sent to those who had not been previously married and family planning and health care questionnaires sent to those who had been previously married. Comparison of Respondents and Nonrespondents to Mail Survey From Interview Data Comparisons of mail respondents with mail nonrespondents on the basis of personal inter- view responses provide evidence of some slight differences. Brides who responded to the mail survey tended to have more years of schooling than those who didn't respond (63 percent com- pared with 58 percent had finished high school). Brides who did not respond to the mail survey tended to live in nuclear rather than extended families (67 percent of the nonrespondents com- pared with 58 percent of the respondents). Sixty percent of those who responded compared with 51 percent of those who did not respond to the mail survey reported no move since marriage at time of interview follow-up. QUALITY OF DATA Overall quality of data for sample persons was examined using three measures: 1. Adequacy of returned mail questionnaires (ex- cluding refusals and Post Office returns) 2. Completeness of answers to individual items on mail questionnaire, i.e., those for which a codable answer was reported 3. Consistency (agreement) of information col- lected by different sources. In general quality appears to be positively cor- related with response rates. Adequacy of Mail Questionnaires Before and After Requery Mail responses were edited to determine whether all priority items had been completed. Those questionnaires with one or more priority items missing were classified as inadequate and were requeried in an attempt to add the missing data. Priority items for requery were: Date and State of birth, education, usual activity before and since marriage, employment, income, sources of income, residence before marriage, hospitali- zation insurance coverage of the bride andgroom, and household composition after marriage. The percentage of the questionnaires judged adequate before requery varied from 52 percent for the family planning questionnaire to 56 per- cent for the basic questionnaire, The requery ef- fort increased the percentage judged adequate to 67 percent for the family planning questionnaire and 72 percent for the basic questionnaire (table 9). Only one inadequate section was required to classify the whole questionnaire as inadequate; hence the percentage of each section which was classified as adequate was considerably higher than the percentage of questionnaires classified as adequate, Levels of completeness for the whole questionnaire and for sections common to all ques- tionnaires were similar for the three question- naires. Although differences between question- naires are not significant, completeness in the common sections was consistently lower for the family planning than for the other two question- naires. Adequacy generally declined in successive mailings (table 9) and adequacy for certified and regular mail responses was very similar. Al- though adequacy levels before requery appear to be different for the three questionnaires, they are based upon relatively small numbers and hence are not statistically significant, Differences are less marked after requery. Adequacy was signifi- cantly higher for white brides than for those of other races (table 10). Completeness of Response to Individual Items Information on adequacy (or completeness) presented in the preceding section tend to ob- scure the relatively better levels of completeness for individual items on returned questionnaires. For individual items the only measure considered was completeness after requery. Because one purpose of the study was to examine quality, an- swers were not imputed for missing data. Completeness for an individual item refers to the proportion with a specific codable answer after requery other than 'mo answer'' or ''un- known." Completeness levels for common items were so similar on each of the three questionnaires that results were pooled. Generally item completeness was quite good for items common to all questionnaires. Com- pleteness was below 90 percent in only three of 22 items for the bride and in sevenof the 22 items for the groom (table B). In general completeness was slightly better for the bride, who presumably completed the questionnaire, than for the groom. Completeness was much lower for items on details of the groom's previous marriages than for other items. How- ever, items pertaining to age and employment have slightly higher completeness levels for the groom than for the bride. The single item below 90 per- Table B. Distribution of items according to level of completeness for 22 items for bride, 22 for groom, and 8 for cou- ple: North Carolina Marriage Survey, 1968-69 Percentage complete Bride | Groom | Couple Total==mm=== 22 22 8 95.0-97 .9=======u- 80.0-84,9========= Less than 80.0---- O o . o ' O & . O 1 1 1 1 1 1 1 ' ' 11 wow tI ND NMNNWSNON cent for couples was the one pertaining to tele- phone number. Completeness for five selected items is shown by major study variables in table 11. Except for the "work last week'' question, the level of com- pleteness declined with successive mailings, and there was little difference between certified mail and regular mail. Area differences except for "telephone number'' were quite small. The level of completeness improved with increased ed- ucation, The level of completeness for family planning items was mixed (table 12). Questions aboutnum- ber of children desired by the bride, whether the bride can have children if she doesn't expectany, current pregnancy status, and use of contracep- tion elicited a response of 90 percent or higher. Questions related to future plans—number of children actually expected, year next child ex- pected, and future use of contraception by couples who had not previously used it—had complete- ness levels of 81 percent or lower. Health care and health insurance questions had levels of com- pleteness of 95 percent or higher, Consistency of Responses Three potential sources for the same data made a number of consistency checks possible. Marriage records were available for both mail respondents (1,999) and nonrespondents (1,592), and interview records were available for 187 mail respondents and 102 mail nonrespondents including 10 refusals. Consistency checks were made between inter - view and mail survey and between vital record and mail survey, Comparison was restricted to those cases for which answers to the specific question were available in both records. The in- dex of consistency is the percentage of cases in which the codes assigned agreed. Consistency percentages are shown in table 13 for a number of items common to all ques- tionnaires. In general, consistency was slightly but not significantly better for mail survey and interview than for mail survey and vital record data for items available on all three record sources. Consistency levels were good or very good except for individual years of education and income. Consistency levels for the small number of respondents to the family planning and health care questions were moderately good except for poor consistency on questions on''number of chil- dren you (or your husband) would like to have" and health insurance for hospital care or doctor, Consistency generally appeared to be positively correlated with completeness of response to individual items. SUMMARY Response rates and quality of response were studied in a follow-back survey of marriages re- corded in North Carolina from February 1968 through January 1969. Three kinds of question- naires each with five to six pages were used in a mail survey of about 3,600 brides under 45 years of age which was conducted during the period November 1968-March 1969. As many as three mailings were made to each bride in the survey and personal interviews of samples ofre- spondents and nonrespondents to the mail survey were conducted to study the quality of the data. Response rates were significantly higher for brides being married for the first time than for those who had been married before, for brides under age 30 than for those 30-44 years of age, and for brides to whom the second mailing was sent certified than for those to whom it was sent by regular mail. The basic and health care questionnaires yielded significantly higher total response rates than the family planning ques- tionnaire when certified mail was used for the second mail, Total mail response rates were slightly lower for white brides than for those of other races even though first wave responses were considerably higher for white brides. First wave response rates were significantly higher at 5S months than at shorter or longer durations be- tween marriage and first mail query, and there was a slightly, but not significantly, higher re- sponse rate at 5 months for all waves combined. Interview follow-up of nonrespondents to the mail survey added an estimated 14-23 percent to mail response rates, yielding overall response rates between 75and 85percent. Interviews showed that nonrespondents tended to have slightly lower incomes and education levels than re- spondents but distributions were not significantly different. Completeness of information on returned questionnaires was quite good for most items except income for the bride, income for the groom, and selected family planning items. Completeness of information for certain items was significantly higher for responses to initial queries than for responses to second and third mailings and for white respondents than for respondents of other races. In general complete- ness appeared to vary in the same directions as response rates. Less effort was required to get the white brides to respond and, although their response rates generally were lower than for brides of other races, completeness of response was better, Consistency indexes comparing vitalrecords with mail survey data and mail survey data with interview data were quite good except for income, certain family planning items, and number of years of education, Overall, response rates and quality of data indicate that it is feasible to use mail follow- back surveys linked to marriage records to col- lect supplementary data from brides for whom this was the first marriage. Poor response to the family planning questionnaire with certified mail follow-up and poor response of older brides and those who had been married before demon- strates the need for additional research for im- proved survey techniques or for subject matter which would stimulate response from these groups. REFERENCES 1 ; . . . U.S. National Committee on Vital and Health Statistics: Needs for National Studies of Population Dynamics. Washing- ton, D.C. U.S. Department of Health, Education, and Welfare, 1970. Commission on Population Growth and the American Future: Population and the American Future, Washington. U.S. Government Printing Office, 1972. 3 National Center for Health Statistics: Design of Surveys Linked to Death Records, by M.G. Sirken, J.W. Pifer, and M.L. Brown. Public Health Service. Washington, D.C., Sept. 1962. * svcd Center for Health Statistics: Methods and re- sponse characteristics, National Natality Survey, United States, 1963. Vital and Health Statistics. PHS Pub. No.1000-Series 22-No. 3. Public Health Service. Washington. U.S. Government Printing Office, Sept. 1966 5 Lo . Wells, H.B.; Coulter, E.J.; Wienir, L.; and Sirken, M.G.: North Carolina survey of recently married persons: study de- sign and pretest results, American Statistical Association Pro- ceedings of the Social Statistics Section, Washington, D.C. 1968, pp. 297-307. 6 Pratt, W.F.: A Study of Marriages Involving Premarital Pregnancies. Microfilmed dissertation, University Microfilms, Ann Arbor, Michigan, 1965. 7 Coulter, E.J. and Greenberg, B.G.: A Pilot Mail Query Survey of Newly Married Couples—North Carolina 1966. Un- published manuscript, 18 page mineograph available. University of North Carolina, Department of Biostatistics, Chapel Hill, North Carolina. —_—000— Table 1. 10. 11. 12, 13. LIST OF DETAILED TABLES Distribution of brides in study population and sample by area and age, previous marital status, and color of bride: North Carolina Marriage Survey, 1968-69 ----- Unweighted cumulative mail response rate per 100 brides by selected character- istics: North Carolina Marriage Survey, 1968-69 === ceo cmmm comm cee Unweighted mail response rate and increase due to interview per 100 brides by selected characteristics and type of second mail: Six-county area, North Carolina Marriage Survey, 1968-69 === cece enema em Weighted mail response rate per 100 brides, by selected characteristics and type of second mail: North Carolina Marriage Survey, 1968-69 ------ceccoomcocccecacacox Weighted mail response rate per 100 brides by time duration since marriage, type of second mail, and version of questionnaire: North Carolina Marriage Survey, 1968 =69 == == mmm mm ee ee ee ee ee eee Weighted mail response rate per 100 brides by color and previous marital status of bride, version of questionnaire, and type of second mail: North Carolina Marriage Survey, 1968-69 - == om cm common emcee ccm mmm Weighted mail response rate per 100 brides by color, age, and previous marital status of bride, version of questionnaire, and type of second mail: North Caro- lina Marriage Survey, 1968-69 === ccc comm eee ccc mmm ————— Weighted mail response rate per 100 brides by color, age, and previous marital status of bride,and type of second mail:North Carolina Marriage Survey,1968-69-- Adequacy of mail questionnaire by wave of response, type of second mail, and version of questionnaire: North Carolina Marriage Survey, 1968-69 -=vmcmceetcocm- Adequacy of mail questionnaire by wave of response, type of second mail, and color of bride: North Carolina Marriage Survey, 1968-69 -==-mcecmmoccacccccmoen- Percent of selected items completed on the mail questionnaire, by wave of re- sponse, type of second query, version of questionnaire, and selected character- istics of the bride: North Carolina Marriage Survey, 1968-69 -=-mmcmeecccccacaon- Percent of selected items completed on family planning and health care question- naires: North Carolina Marriage Survey, 1968-69 -====m-m mmc coco cocccccc meee Percent of agreement of information obtained for selected items on all mail questionnaires with corresponding data on the vital record and the interview questionnaires: North Carolina Marriage Survey, 1968-69 -=mmmmoccccmcccccmemoanan 11 12 13 14 15 15 16 17 18 19 20 21 22 Table 1. Distribution of brides in study population and sample by area and age, pre- vious marital status, and color of bride: North Carolina Marriage Survey, 1968-69 Six=-county area Rest of State Age of bride Sow, | mnge | gewe | Srey White| Other | White | Other| White | Other | White | Other Number in study population Under 20 years=--=--==-e====-- 2,921 751 62 5115,572 | 2,416 255 8 20-29 years=-=-===c-ccceceaa= 2,862 938 607 72|11,002 | 2,433 2,416 181 30-44 years-----=ceccemmcemcea- 104 86 452 100 397 268| 1,866 291 45 years and over--=----e---- 22 17 307 80 123 52] 1,279 227 Percent of total study population Under 20 years===============- 6.1 1.6 0.1 0.0] 32.3 5.0 0.5 0.0 20-29 years--=-=c-c-cmcecmea= 5.9 1.9 1.3 0.1} 22.8 5.0 5.0 0.4 30-44 yearS=--==---emecceeaa- 0.2 0.2 0.9 0.2 0.8 0.5 3.9 0.6 45 years and over--=--==----- 0.0 0.0 0.6 0.2 0.2 0.1 2.6 0.5 Number in sample Under 20 years---------==-=-= 144 144 62 5 216 216 192 8 20-29 years=---==--cc-ccccea= 146 143 144 62 216 216 216 179 30-44 years=---=---ememcccea= 103 86 144 100 215 204 217 213 45 years and over------------ 20 17 72 73 49 44 70 45 11 Table 2. Unweighted cumulative mail response rate per 100 brides by selected charac- teristics: North Carolina Marriage Survey, 1968-69 Total Certified mail Regular mail number y Characteristic question- Flrst naires WW Second Third Second Third mailed wave wave wave wave Response rate per 100 brides Total=cemcemmcmnn nm ——————— 3,591 24.4 47.5 59.0 38.0 52.2 Area Six=-county area----=---cececaao- 1,283 27.4 46.4 59.0 40.7 54.3 Rest of State---meeccccccccaaaa- 2,308 22.8 48.2 59.1 36.5 51.0 Age of bride Under 20 years--------ememcecaa- 987 30.0 53.0 65.8 46.7 60.3 20-29 years---==--memccemcacoan 1,322 26.5 50.0 60.6 39.0 54,1 30-44 years------emmmcmemceaaa 1,282 18.1 40,9 52.2 30.4 44,1 Color of bride White====mmrrmnecencwcccee ean 2,015 25.9 46.2 56.0 37.8 50.9 OE IYEY we wom om mom om oo i mo oo 1,576 22.6 49,2 62.8 38.4 54.0 Previous marital status of bride Never married-------cemmceccaaoo 2,049 29.2 52,5 65,2 44.4 59.8 Previously married---------c--- 1,542 18.2 41.1 50.9 29.6 42.2 Time duration since marriage 3 monthg-=--ecmmcmcmm ceo 930 24.4 48.7 59.3 38.0 54,2 5 months ==-c-cccmmmm meee 870 27.4 50.6 60.9 41.4 55.2 7 months ---=-cc cmc ccc 865 24.4 46,2 57.0 38.2 51.1 9 months ----ccmmmmmeee ooo 926 21.8 44,8 59.0 34.6 48,2 Version of questionnaire BasiC--mmmmee mcm o 1,201 25,7 47.3 60.0 38.8 53.6 Family planning-----=-ceceeoooo-o 1,196 24,1 46.3 57.1 36.9 51,2 Health care----=-cececeeeoooooo 1,194 23.5 49.1 €0.0 39.3 52,7 12 Table 3. Unweighted mail response rate and increase due to interview per 100 brides by selected characteristics and Marriage Survey, 1968-69 type of second mail: Six-county area, North Carol ina Amount added by interview Total, mail Mail plus Poe . . ponse interview Nonrespond - Characteristic ents Refusals Certi- | Regu- Certi- | Regu- | Certi=- | Regu-| Certi=- | Regu- fied lar fied lar fied lar fied lar Response rate per 100 brides Total, six=-county area---========o= 76.7 80.4 59.0 54.3 16.3 25.5 1.4 0.6 Age of bride Under 20 years---------- 79.0 84.4 65.4 65.3 12.7 19.1 0.9 0.0 20-29 years---------=--=-- 81.2 81.3 59.9 55.6 19.9 24.6 1.4 1:1 30-44 years=-----==--=-- 69.2 76.3 52.6 44,0 14.8 317 1.8 0.6 Color of bride White-=-==ccmcmcccccaaaa- 73.8 79.2 59.0 54,2 12.4 24.4 2.4 0.6 Other------cemcmccccceaa- 80.3 81.8 58.8 54.4 21.3 26.8 0.2 0.6 Previous marital status of bride Never married----==----- 82.6 84.9 66.2 63.0 15.3 20.9 1.1 1.0 Previously married------ 68.3 73.2 48.4 41.1 18.0 32.1 1.9 0.0 Time duration since marriage 3 months ====ccemccaaaaa- 76.2 75.0 58.2 55.4 15.9 19.0 2:1 0.6 5 months--==-cccecaceca-- 74.5 77.1 58.7 58.9 14.8 17.2 1.0 1.0 7 months=---cemccccaaaa- 80.1 84.7 57.2 55.1 21.1 29.6 1.8 0.0 9 months=-==-cccmecaaa-- 74.8 86.6 61.5 48.2 12.7 37.6 0.6 0.8 Version of question- naire Basicm=--emecmmmm eee 70.6 80.9 54.5 57.2 14.8 23.7 1.3 0.0 Family planning--------- 81.3 78.2 62.3] 51.2 17.4 | 25.0 1.6 2.0 Health care==-=-ceceeoa- 78.9 83.0 60.1 54.5 17.5 28.5 1.3 0.0 Table 4. Weighted mail response rate per 100 brides, type of second mail: North Carolina Marriage Survey, 1968-69 by selected characteristics and All waves . First Characteristic t wave Certified | Regular Dijtevanca ail sail | eextified- m regular Response rate per 100 brides Total --eemmecccm ccc mm ccc mecca 30.3 66.6 59.0 +7.6 Area Six-county area-------ccemcmmcmcccmccaaaao 34.3 66.6 61.2 +5.4 Rest of State---emcececcmcccccccccccceaeo 29.3 66.7 58.4 +8.2 _Age of bride Under 20 yearS-==----ccmecccmcccceccccanoa- 30.7 68.6 62.2 +6.5 20-29 year§--=--=-cmmmmcccmcmcccmcc meee 32.1 68.3 58.7 +9,6 30-44 yearS-----m--ccmmmm cece cme eee 16.6 44.9 40.4 +4.4 Color of bride White--=-cccmcm mcrae em 30.7 66.3 58.1 +8.2 Other----ecem ecm eee cmee meee 27.8 68.5 63.2 +5.3 Previous marital status of bride Never married----=-c-ccmccccccccccccccc eee 31.7 69.8 62.0 +7.8 Previously married----=---cececccmnnnaanao 18.0 47.0 39.7 +7 .3 Time duration since marriage 3 months----cceceuna- mmm em 27.8 69.0 58.5 +10.5 5 months=--ceecm cme ccc eam 34.4 70.0 63:5 +6.5 7 months =---ccee mmr eee - 31.5 64.7 58.5 +6.2 9 months =--=c-cmcm meee em 27.3 62.9 55.4 +7.5 Version of questionnaire Basic=--cmmme mmo rere em 32.1 73.0 60.4 +12.6 Family planning---=-----cccccmcmmmcaccaaa 29.5 56.8 57.9 -1.1 Health care------cecmmmmmccc ccc eee 29.2 70.2 58.6 +11.6 14 Table 5. Weighted mail response rate per type of second mail, and version of questionnaire: 100 brides by time duration since marriage, North Carolina Marriage Survey, 1968-69 All waves Version of questionnaire and time First duration since marriage wave y Difference Gene fad Regu) ax certified- regular Basic questionnaire Response rate 3 months=--==-ececemc emcee ccc mec ccm meee 30.9 70.3 61.8 +8.5 5 months-------ccccccmmemcm ccc ccc cme em 36.0 79.6 67.4 +12 .2 7 monthg===-c-cceemmece ccc ccc cece mmm 31.6 75.7 56.8 +18.8 9 months--------cecmmcmcmm ccc c cece mmm 30.0 66.2 55.5 +10.7 Family planning questionnaire 3 months---=----ccecmmmm cece c ccc meme 28.7 62.6 55.0 +7.7 5 months=-=----cccmecm cmc ccc cece mmm 32.8 60.1 61.9 -1.8 7 months =-=-eeemcmc cece ccm c cece eee 29.1 47.9 57.5 -9.6 9 months------c-cmemmece ccc cc ccc meme 27.3 56.6 57:1 -0.4 Health care questionnaire 3 monthS=-e-meecccm meee c ere c meee eee 23.8 74.0 58.6 +15.3 5 months===-ececccmmm meee ccc meme mem em 34,8 70.3 61,1 +9,2 7 monthS§-------cccmmmcm cree ccc cca 33.7 70.4 61.1 +9.3 9 months------ccmcmmccm ccc cece cme = 24,6 65.9 53.6 +12.3 Table 6. Weighted mail response rate per 100 brides by color and previous marital sta- tus of bride, version of questionnaire, and type of second mail: North Carolina Mar- riage Survey, 1968-69 Never married Previously married Color of bride and version of All waves All waves questionnaire First First wave Certi- | Regu-| wave Certi-| Regu- fied lar fied lar mail mail mail mail White Response rate per 100 brides Basic--===--mcemmce mcm mmm em 35.2 78.9 63.4 18.6 40.9 44,7 Family planning-----------c-cmceccccacao- 31.7 56.6 | 58.9| 14.6 45.1 35.4 Health care---------cccmeccmcmcmccnnnnnan 31.9 73.9 61.9 21.2 51.9 37.4 Other Basic-----mmemecmeemec emcee meee ———— 29.5 72.2| 59.9] 120.4| 158.8( 151.9 Family planning-------c-c-eceemccmeanaano- 32.7 87.7 73.2] 13.3] 357.4] 138.2 Health care---=-=--c-cemcrmccmcmc meee ceaan 24.8 68.9 61.6 14.1 152.1 143.0 'Based on response experience for ages 20-44 years. 15 Table 7. Weighted mail response rate per 100 brides by color, age, and previous marital status of bride, Marriage Survey, 1968-69 version of questionnaire, and type of second mail: North Carolin a Never married Previously married Age and color of bride and version All waves ALL waves of questionnaire First First wave Certi- | Regu- | wave Certi- | Regu- fied lar fied lar mail mail mail mail UNDER 20 YEARS Response rate per 100 brides White Basice=mmmemmm mmm eee 34.1 75.4 | 63.9 24.6 54.2 53.5 Family planning--===---cccecmcccmmaaaaano 25.0 49.2 | 52.8 19,7 46.5 49.9 Health care---=--ccecccmmm mci cece 32.4 80.1 | 67.3 23.8 61.9 48.2 Other BasiCe==memmmmm meee 31.2 70.8 63.9 * * * Family planning---------cccccmmmcmcmaaana- 40.6 71.5 79.5 * * * Health care=-==----ccmmmmm mmc 27.6 75.8 60.4 * * * 20-29 YEARS White BasiC==mmmem meme eee 37.2 84.9 | 63.1 24,3 46.0 49.7 Family planning---=--=--ccccmmcmancaaaoo 40.8 66.6 67.4 13.9 49.9 32.0 Health care----==--cccmcmmmccmcccccceee 31.6 66.5 54,9 | 21.5 54.3 37.6 Other Basicmmmmm momma 28.4 73.9( 57.6 2). 7 58.2 51.1 Family planning-------cccccmommccacaaaaa- 26.6 65.5 70.0] 19.2 40.5 43.1 Health care--------cccmcmmmmcccce eee 23.4 64.0 | 64.7 17.9 45.9 53.3 30-44 YEARS White BasiC-mmmemmm meme meee 19,5 43,6 53.0 10.4 32.5 36.9 Family planning-------ceccccccmmmmnacaaaa 25.0 50.0 | 48.7 | 14.9 38.5 37.8 Health care--=--cccemmcmc cece ee 25.4 47.7 | 56.5| 20.5 47.4 35.7 Other JE Et 24,6 68.0 | 45.1 19.6 59.3 52.4 Family planning-=-----ccccccmmmmccccaaaao 19.6 53.5 | 47.6 9.6 67.9 35.2 Health care=--=---ccccmmcm ccc 12.9 55.0 43.7 11.8 55.9 36.7 16 Table 8. Weighted mail response rate per 100 brides by color, age, and previous marital status of bride, and type of second mail: North Carolina Marriage Survey, 1968-69 Never married Previously married Color and age of bride All waves All waves First First wave wave Certi- | Regu- Certi- | Regu- fied lar fied lar Response rate per 100 brides Whiteeeeeacaaamacacccccccacnacacaa—- 32,9 69.8 61.4 18.2 46.0 39.1 Under 20 yearS=-==ese-ccmccccacccacecaaa—-- 30.5 68.2 61.4 22.7 54.2 50.6 20-29 yearS=s=s-cescccccccecccececcccnan—- 36.5 72.7 61.7 19.9 50,1 39.7 30-44 yearsee---mccccccmcacacdcccacanao-a 23.3 47.1 52.7 15.3 39.5 36.8 Othere-cecccccmcccca ccc 29.0 69.6 64.9 | 115.9 156.1 Yas 4 Under 20 years=ee-ececacacecccccccccaa-a- 33.1 72,7 67.9 * * * 20-29 yearSee=-=-ecemce-cccccccccmecem————— 26.1 67.8 64.1 19.6 48.2 49.1 30-44 yearS=se-ecececceccccccccac acca 19.0 58.9 45.5 13.7 61.0 41.4 !Based on response experience for ages 20-44 years. 17 Table 9. version of questionnaire: North Carolina Marriage Survey, 1968-69 Adequacy of mail questionnaire by wave of response, type of second mail, and Version of questionnaire All Wave of response and type of second mail question- naires Family Health Basic planning care Number of responses Totaleeeccecccccccancacnncnccacccanan 1,999 683 648 668 First wave-eececacacaacecccacecnceccannnn= 878 309 288 281 Certified: Second Waveeeememcacaccnec acme cmnnnnnan—- 421 131 135 155 Third wave-eeeeeecacccccccccccccccaccaanaa 209 77 66 66 Regular: Second Waveeeeecuacaamceacacenmcennnannaan 240 78 75 87 Third wave-eeeeeccccccccaccccccccaacccaa= 251 88 84 79 Percent of questionnaires adequate before requery TOtal=eeccuucuccnnnnenunennnnnncccann- 53.8 55.8 51.8 53.6 First waveeeeececcecccccacccccccccacacaa 57.8 58.3 58.0 57.3 Certified: Second WavVeeeeeeccacceccccncc cence annanan= 52.5 60.3 48.1 49.7 Third waveeeeeecccccaaaccacaaccacccacana= 46. 40.3 45.5 54.5 Regular: Second waves==emeceeccccccccacccacaccncanna= 52.1 65.4 46.7 44.8 Third waveeeeeececccccacacccaccaccacccanaa- 49.4 45.5 46.4 57.0 Percent of questionnaires adequate after requery Totaleeeccccacncccccaccacccccacacanaa 70.1 72.3 67.0 71.0 First wave-e-=eeecccccccccccncacancanax 75.0 75.1 70.8 79.4 Certified: Second Waveee=mmecceccaccncccaccec ncaa nna= 69.1 77.1 65.2 65.8 Third waveeeeccccacaccccacccccccccacaaca- 63.2 61.0 60.6 68.2 Regular: Second wave=eeececccccccccccccncecenanan= 65.0 74.4 65.3 56.3 Third wave~swsssssneennsnnennnancnennuune 65.3 63.6 63.1 69.6 Table 10. Adequacy of mail questionnaires by wave of response, type of second mail, and color of bride: North Carolina Marriage Survey, 1968-69 Number of Percent of question- naire adequate Percent of question- naire adequate Wave of response and Tesponses before requery after requery type of second mail White | Other White Other White Other First waveeeeeecaacaa 522 356 66.9 44.7 82.4 64.3 Certified: Second wavee==meccccaca== 206 215 62.6 42.8 79.6 59.1 Third waveeeeececcccccacaa 99 110 56.6 37.3 722.7 54.5 Regular: Second waves=eeecaccccna= 119 121 61.3 43.0 77.3 52.9 Third wave=eecececccccaca= 131 120 61.1 36.7 72.5 57.5 19 Table 11. Percent of selected items completed on the mail questionnaire, by wave of response, type of second query, version of questionnaire, and selected characteristics of the bride: North Carolina Marriage Survey, 1968-69 Item Total Work last Number of mail week months at Day of Addi- Study variable re- address ay tional Tele- spond - before children hone ents marriage not in PROBS Bride | Groom houses hold Groom Bride Percent Total ==--mm mmm eee 1,999 89.3 88.0 85.2 91,7 90.8 80.2 Wave of response and type of mail First wave=-=-mmemcomcooccaaooooo 878 88.4 87.9 88.5 94.3 92.5 81.0 Certified: Second wave----=-cmmmommooomeo ooo 421 91.0 88.4 84.6 92.2 91.4 80.3 Third wave-----eemcmmce eee 209 88.0 7.1 80.4 84.7 88.0 78.9 Regular: Second wave-----emcmmemmeooooooo 240 92.5 89.2 84,2 90.8 90.8 82.5 Third wave----ememmceae ee 251 88.0 86.9 80,1 88.4 86.1 76.5 Area Six-county area--------cecmceeeoooooo 727 89.8 | 87.6 84.9 91.9 92.0 85.0 Rest of state---=ececeemmeceooooo_ 1,272 89.0 88.1 85.4 91.6 90.0 77.5 Color of bride White-=--ooeo mmm 1,077 92.5 89.7 89.4 94,7 93.9 86.5 ONT wee rm mm or si os 922 85.7 85.9 80.4 88.2 87.2 72,9 Age of bride Under 20 yearS=====mm-eecocomoocaoon 623 90.8 90.1 85.7 93.9 92,7 78.7 20-29 years----=mmmecemeccmcmmeman 758 90.8 87.3 87.1 93.3 92.2 81.8 30-44 years----mmcmmmmce mea 618 86.0 86.5 82,5 87.5 87. 79.9 Version of questionnaire BaSiC=mmmm meme ee ee 683 89.4 89.1 85.0 91.5 89.3 81.2 Family planning-------ccoeeoaaaaoo. 648 89.4 | 86.1 85.5 90.8 90.9 80.4 Health care----cemmcmemeo oo. 668 89.5 88.5 85,2 92.8 92, 79.0 Education of bride! 9 years or lesS---=-=emeecmmcmcmoooan 217 79.2 89.4 80.2 81.6 87.6 69.1 10-11 years----emmmcoomce occa 306 86.6 88.5 81.4 86.3 89.2 69.6 12 years=---emcmmmmmmeeee 576 92.2 87.7 87.0 94.8 91.5 81.4 13 years or more---=--eecoccmcaoooo. 405 94.6 | 88.4 90.1 95.3 93.8 91,9 IBased upon 1,504 brides because education was not available 495 vital records. 20 Table 12. Percent of selected items completed on family planning and health care ques- tionnaires: North Carolina Marriage Survey, 1968-69 Number of respondents Percent Version of questionnaire and question for whom 1 d question complete applicable Family planning Has the bride ever thought about number of children she would like to have? -----=-ecececcccmcmmm ccc ccccem meen cc mm mm 648 83.8 Number of children desired by: Bride---memmcmcmcmc meme eee emma mmmmmmmmeeecccemem—m———— 462 98.9 GY OOM====m=mmmmmeme meee cece ce eee me eeemeemeeem—e—————eae— 462 88.5 Number of children bride actually expectS-=-=-----=-----=-==c-=- 648 80.7 Does the bride think she can have children if none expected? =----mmmmmccmmme eee mmemmm mmm mmm mcmo moomoo 113 96.5 Is the bride pregnant? -----------eeecemmcmm ccc cmcmecoocaooan— 589 93.9 Year next child expected if bride not pregnant and thinks she can have children-----------cc--memmmmmccmeecnccec cee 408 76.5 Has bride or groom ever used methods to keep from having children? —==-----cecccccccc rem mmecmececmccc mmm mm mmm em 648 91.0 Specific methods of contraception used-=----=------c---ccc-o-- 358 100.0 Is future contraception by bride or groom anticipated if neither has used it?--------ceecmcc ccc cmcm cece cme meme mmm 230 70.4 Health care Is the bride now expecting a baby?--------cccececemmeananaaa- 668 98.7 Has the bride had a miscarriage since the present marriage?-- 668 96.1 Has the bride been in the hospital overnight since the present marriage? -----=------eemmmemmmemmememmcmommeo——ooo- 668 97.2 Do the bride and groom have insurance for payment of hospital billgs---=-t--cmmcmmmeccm ccc mc ccccmccc cme m mmm mee 668 97.5 Does any available insurance for hospital bills provide for costs of care for delivery of a baby?-----c-cccmccmmmcnncana- 462 94.8 Do the bride and groom have insurance to pay for bills of physicians? ----=cccmmccm cme cece rece reemeeee 668 94,8 Does any available insurance for bills of physicians provide for expenses of delivering a baby?---------cccmccmmcnccnnnan 400 96.5 21 Table 13. questionnaires with corresponding data on tionnaire: North Carolina Marriage Survey, the 1968-69 Percent of agreement of information obtained for selected items on all mail vital record and the interview ques- Mail survey and Mail survey and vital record interview Item Total, Total, item item Percent Percent reported | ooreement | ZSPOTtEd | 4orcenent sources sources Previous marital status: Bride---cccccmm ccm 1,946 98.4 184 98.9 GrOOM=====em ccc ccc ccc ccccmcec cece ecm 1,923 97.8 184 98.9 Number of times married: Bride-==-cccmcmm mmc e 1,934 97.3 181 98.9 GroOm===--= ce ccccccc ccc ccc ccc ce cm ——— 1,912 96.2 182 96.7 How first marriage ended (for previously married) : Bride-----cccmmmm mecca 666 98.0 64 96.9 (6 dele EE i 392 96.7 41 100.0 State of birth: Bride--=ccccc mmc ee 1,984 95.0 181 98.9 (63 dele) ER TT Tp pp—— 1,880 90.1 177 96.5 State of residence before marriage: Bride----cecccmm mcm e 1,984 95.0 186 98.9 GroOmM===-mcecm ccc cccccccccce cee 1,880 90.1 172 96.5 Year of birth of bride------eccemmccccccaaoao 1,826 95,7 166 97.6 BAGCAtLion OF DULAC =m mm mmm mmm mim wm ww www 1,495 72.0 185 77.8 Income: ! Bride==--emcccc ccc - — ve 173 69.4 (6 ele) EE 2% . 168 60.1 Year of birth of child delivered since marriagel == ceo omm eee ‘ow e 180 91.1 INot adjusted for time delay between mail survey and interview. 22 APPENDIX | FORMS USED IN THE STUDY LICENSE AND CERTIFICATE OF MARRIAGE — 1968 State of North Carolina LICENSE NUMBER COUNTY (C GROOM—NAME FIRST MIDDLE LAST ) RESIDENCE—STATE COUNTY CITY, TOWN, OR LOCATION INSIDE CITY LIMITS (Specify Yes Or No) 2a. 2b. 2c. 2d. STREET AND NUMBER | STATE OF RIRTH (If Not In US.A,, DATE OF BIRTH (Month, Day, Year) | AGE | Name Country) 2e. 3. 4a. 4b. FATHER—NAME STATE OF BIRTH (If Not In MOTHER—MAIDEN NAME STATE OF BIRTH (If Not In U.S.A., Name Country) U.S.A., Name Country) 5a. 5b. 6a. 6b. RACE—GROOM NUMBER OF THIS MARRIAGE IF PREVIOUSLY MARRIED EDUCATION—SPECIFY HIGHEST GRADE COMPLETED = First, SoD, LAST MARRIAGE ENDED BY DATE ELEMENTARY HIGH SCHOOL COLLEGE ETC. (SPECIFY) Death, Divorce, Or Annulment (Specify) “MONTH YEAR |(0,1,2,3,4, ...or 8) (1,2,3, or 4) 1,2,3,4, or 5) 7. 8. 9a. 9b. 10. BRIDE—NAME FIRST MIDDLE LAST MAIDEN NAME (If Different) 1a. 11b. RESIDENCE—STATE COUNTY CITY, TOWN, OR LOCATION INSIDE CITY LIMITS (Specify Yes or No) 12a. 12b. 12c. 12d. BRIDE STREET AND NUMBER STATE OF BIRTH (If Not In US.A., DATE OF BIRTH (Month, Day, Year) | AGE Name Country) 12e. 13. 14a. 14b. FATHER—NAME STATE OF ST Hof Not In MOTHER—MAIDEN NAME STATE OF BIRTH (If Not In lame Country) U.S.A., Name Country) 15a. 15b. 16a. 16b. RACE—BRIDE NUMBER OF THIS MARRIAGE IF PREVIOUSLY MARRIED EDUCATION—SPECIFY HIGHEST GRADE COMPLETED - BR >. LAST MARRIAGE ENDED BY | DATE ELEMENTARY | HIGH SCHOOL COLLEGE «4 ) Death, Divorce, or Annulment (Specify) “MONTH YEAR |(0,1,2,3,4, ...0r 8)| (1,2,3, or 4) (1,234, 0r 5) 19a. 19b. 20. BRIDE—PARENT'S ADDRESS 22. To any ordained mini of any religi d inati ini authorized by his church, or any Justice of the Peace or Magistrate, you are hereby authorized, at any time within 60 days from the date hereof, to celebrate the proposed marriage at any place within the said county. DATE ISSUED REGISTER OF DEEDS (DEPUTY /ASSISTANT) | CERTIFY THAT THE ABOVE MONTH DAY YEAR PLACE OF MARRIAGE—COUNTY STATE NAMED PERSON S WERE ARRIED ON 15a 15b. 15¢. OFFICIANT—SIGNATURE DATE SIGNED (MONTH,DAY, YEAR) OFFICIANT—Religious or Civil (Specify) 15d. 15e. 154. WIT —SIGNATURE 16a. 16b. The minister or other person celebrating this marriage is required within 10 days to fill out and sign RETURNED TO REGISTER OF DEEDS: both copies of this Certificate of Marriage, and return them to the Register of Deeds who issued the license. Failure to do so constitutes a misdemeanor and also subjects person celebrating the riag: , to a forfeiture of $200.00 to anyone who sues for the same. REGISTER OF DEEDS/DEPUTY OR ASSISTANT FORM VS.80 REV. 1/1/68 1/68-100M 23 LICENSE AND CERTIFICATE OF MARRIAGE State of North Carolina —1969 LICENSE NUMBER COUNTY GROOM-NAME FIRST MIDDLE LAST L RESIDENCE-STATE COUNTY CITY, TOWN, OR LOCATION INSIDE CITY LIMITS (Specify Yes Or No) EE 2a. 2b. 2c. 2d. STREET AND NUMBER BIRTHPLACE (COUNTY & STATE) DATE OF BIRTH (Month, Day, Year) AGE 2e. 3 40. 4b. FATHER -NAME STATE OF BIRTH ADDRESS (If Living) 5a. 5h. 5c. MOTHER-MAIDEN NAME STATE OF BIRTH ADDRESS (If Living) 5 6b. bc. RACE-GROOM NUMBER OF THIS MARRIAGE IF_PREVIOUSLY MARRIED EDUCATION-SPECIFY HIGHEST GRADE COMPLETED Sern: LAST MARRIAGE ENDED BY — DATE ELEMENTARY HIGH SCHOOL COLLEGE +4 ) Death, Divorce, Or Annulment (Specify) MONTH YEAR [(0,1.2,3.4,.. .0r 8) (1,23, 0rd) | (1,23 4 orb) \ 7. 8 9a. 9b. 10. (’ BRIDE-NAME FIRST MIDDLE LAST MAIDEN SURNAME (If Different) a 1b. El RESIDENCE-STATE COUNTY CITY, TOWN, OR LOCATION INSIDE CITY LIMITS (Specify Yes Or No) 12a 12b. 12¢ 12d. STREET AND NUMBER BIRTHPLACE (COUNTY & STATE) DATE OF BIRTH (Month, Day, Year) AGE 12e 13, l4o. 14b. FATHER-NAME STATE OF BIRTH ADDRESS (If Living) 150. 15b. 15¢ MOTHER-MAIDEN NAME STATE OF BIRTH ADDRESS (If Living) 16a 16b. léc. RACE-BRIDE HUMBER OF Tis MARRIAGE IF PREVIOUSLY MARRIED EDUCATION-SPECIFY HIGHEST GRADE COMPLETED ETC. (SPECIFY) LAST MARRIAGE ENDED BY DATE ELEMENTARY HIGH SCHOOL COLLEGE Death, Divorce, Or Annulment (Specify) [MONTH YEAR (0.1.2.3, 4... or 8) (1,23. 0rd | (1.2 3 4 or 5) 17 18 19a 19b. 20 DATE ISSUED REGISTER OF DEEDS (DEPUTY /ASSISTANT) To any ordained minister of any religious denomination, minister authorized by his church, or any Justice of the Peace or Magistrate, you are hereby authorized, at any time within 40 days from the date hereof, to celebrate the proposed marriage at any place within the above named county. reread | 2'o 21b. ( I CERTIFY THAT THE ABOVE YEAR Pi F MARRIAGE-CITY, TOWN, OR TOWNSHIP, COUNTY NAMED PERSONS WoRE MONTH DAY EA LACE OF MARRIAGE-C| OR TO! OFFICIANT-SIGNATURE TITLE ADDRESS \2'c. 21d. 2le. (SIGNATURE OF WITNESS SIGNATURE OF WITNESS LAL 3313 NAME OF WITNESS (Please Print) NAME OF WITNESS (Please Print) 22b 23b ADDRESS ADDRESS 22c. 23c. The minister or other person celebrating this marriage is required within 10 days to fill out and sign both copies of this Certificate of Marriage, a1 return them to the Register of Deeds who issued the license. Failure to do so subjects person celebrating the marriage to a forfeiture of $200.00 to anyone who sues for the same. FORM VS-80 REV. 1/1/69 24 DATE RETURNED TO REGISTER OF DEEDS: RECEIVED BY COVER LETTER FOR BASIC QUESTIONNAIRE (TEXT MODIFIED SLIGHTLY FOR OTHER VERSIONS) DEPARTMENT OF HEALTH, EDUCATION, AND WELFARE PUBLIC HEALTH SERVICE WASHINGTON, D.C. 20201 NATIONAL CENTER FOR HEALTH STATISTICS This questionnaire is being sent out by the University of North Carolina to help the U. S. Public Health Service gather certain new facts about couples recently married in North Carolina. The survey has been approved by: the Director of the North Carolina State Board of Health and is paid for by the U. S. Public Health Service. Your name was selected from the marriage certificates recently filed in North Carolina in such a way that answers from a relatively few recent brides would give an accurate cross-section for the whole State. But since only one out of every 10 brides is chosen, it is especially important that we get a reply from each particular person who received a questionnaire, Some of the questions we are asking are quite personal and your reply to these or any of the questions is entirely voluntary. However, we would like to point out two things. First, the replies will be used only for statistics, i.e. absolutely no use will be made of your reply except to put it together with other replies. Second, the information is really very badly needed and the only person who can give it to us is you. Some of the purposes for which it is to be used are listed below. 1. Learning where and with whom people live after they get married, a matter of interest in planning schools, housing, highways, and health and recreational facilities. 2. Learning about the background of the married couples, including their ages, education, and religious preferences, in order to plan better health and community programs. Let me repeat that all information you provide about yourself, your husband, or any member of your family will be kept completely confidential, as we are bound to do by official regulations of the U. S. Public Health Service. It will not be disclosed to any person or other government agency except for those working on the study, and will be used by them for statistical purposes only. Your cooperation in providing the U. S. Public Health Service with the requested informa- tion and in avoiding further and costlier follow-up procedures is greatly appreciated. By filling out and returning this questionnaire in the enclosed envelope you will be helping greatly to make this survey a success, amd your government will thereby be better able to serve your needs and those of your family. Sincerely yours, Theodore D. Woolsey Director File Number AT 25 ITEMS COMMON TO ALL MAIL QUESTIONNAIRES CONFIDENTIAL - All information which would permit identification of an individual, or of an establishment, will be held confidential, will be used only by persons engaged in and for the purpose of the survey, and will be protected against disclosure in accordance with provisions of 22FR. NORTH CAROLINA MARRIAGE SURVEY PART I. INFORMATION ABOUT YOU In this part we are interesied in obtaining some information about you such as where you were born, whether you were marnied before, whether you are working. This form was designed to be answered by the bride. 1. Where were your born? City County State or Foreign Country 2. How many brothers and sisters do you have? (include those who are now dead) Number 6. What was your usual activity just before your present marriage? [[] working M Attending School [C] Housework [CJ] other (Specify) 3. What is the highest grade (or year) of school that you have finished? (Circle highest grade, COMPLETED) None 0 Public or other { 1 2 3 4 5 6 regular school 7 8 9 10 11 12 College or University 1 2 3 L 5+ Other (Specify) 7. What is your usual activity since marriage? OO Working I] Housewife [J Attending School [J] other (Specify) 4. What is your religion? Protestant (Specify denomination) Roman Catholic Jewish None oogoo Other religion (Specify): 5. a. Have you ever been married before? PLease skip Zo A question 6 above How many times were you married before this present marriage? [J 3 or more Oo: O°: c. What was the date of your first marriage? Yes O NO =p o Year d. What was the date that your first marriage ended? Year e. Did that first marriage end by death, divorce or annulment? 0 Death f. How many children did you have by that first [CO] Dpivorce or annulment 8. a. Did you work at any time last week? O Yes = Db. Please check how many hours you worked: 0 35 or more [J 15 to 34 hours [C] 1ess than 15 hours c. If you did not work last week, do you have a job? O No [J ves d. If you did not work last week, were you looking for a job or on lay off? [ Yes [J wo J No —» 9. What is your own present annual total personal income? O None [J Under $1,000 [J $1,000 - $2,999 [J $3,000 - $4,999 [J $5,000 - $6,999 [J $7,000 - $8,999 O $9,000 or more [L0. From which of the following sources do you receive income? (Check as many as necessary) Wages, Salary (pay check) Parental help Military allowance for dependents Other (Specify) ogooo marriage? Number Go to question 6 J (Page 1) Form Approved (GO ON TO NEXT PAGE) Budget Bureau No. 68-R-0974 Expiration date Dec. 1969 26 PART II. INFORMATION ABOUT YOUR HUSBAND In this pant, we are interested in ment, and marital history. obtaining information about your husband such as his birthplace, employ- 14 you are now separated from your husband, you may skip this part. 1. Where was your husband born? 6. What was your husband's usual activity just before your present marriage? City County State or Foreign Country O Working 2. How many brothers and sisters does your husband OJ seenting Seheol have? (include those who are now dead) [J Armed Forces (Army, Navy, etc.) a [J Other (Specify) 3. What is the highest grade (or year) of school that | 7. What is your husband's usual activity since marriage? your husband has finished? —_ g . . LJ Working (Circle highest grade COMPLETED) 1 : None 0 Attending School Public or other ¢ 1 2 3 4 5 6 oping 7 2 . 10 0G 12 [CO] Armed Forces (Army, Navy, etc.) College or University 1 2 3 4 5+ UJ Other (Specify) Other (Specify) 8. a. Did your husband work at any time last week? 4. What is your husband's religion? Ul Yes =p Db. Plossy Tea Boe many Hours he [[] Protestant (Specify denomination) {] 35 or more O Roman Catholic Od 15 %o 34 pours [7] Jewish [[] 1ess than 15 hours J None c. If your husband did not work last 1 i 4 0 Other religion (Specify): week, does he have a job? [J Yes [J wo 5. a. Has your husband ever been married before? OJ No =p 3 d. If he did not work last week, was D ’ 3 3 3 [Yes 0 No ) Packs Seip bo NA he looking for a job or on lay off? a C] Yes J No b. How mahy times was your husband married before this present marriage? 9. What is your husband's present annual total income? Od: O 2 [J 3 or more [J None [] $5,000 - $6,999 c. What was the date of your husband's first 1] Under $1,000 J $7,000 - $8,999 marriage? O $1,000 - $2,999 O $9,000 or more a [J] $3,000 - $4,999 d. What was the date that your husband's first 10. From which of the following sources does your marriage ended? Year e. Did his first marriage end by death, divorce or annulment? [J Death f. How many children did your husband have by that first marriage? Divorce or annulment Number Go to question 6 husband receive income? [[] wages, Salary (pay check) Parental help [C] other (Specify) (Page 2) (GO ON TO PART III) 27 PART Il. MIGRATION In this section we would Like to obtain infommation about the places where you and your husband Lived before you got marnied and since you have been married. YOURSELF YOUR HUSBAND 1. a. Just before you were married, where did you 2. a. Just before your husband was married, where did live? (home residence not P.O. Box) he live? (home residence not P.O. Box) Street Street City County State or Foreign Country City County State or Foreign Country b. How long did you live there? b. How long did he live there? Months Years Months Years c. With whom did you live? Cc. With whom did he live? O Alone With other relative(s) J Alone J With other relative(s) [] with your parent(s) [] With other pessenis) 4 [J] with his parent(s) _] With other person(s) [C] with your children Go to question 2 [J with his children 3. Just after your marriage, with whom did you and your husband live? [J with your parent(s) [[] with other relative(s) J With his parent(s) [J Alone, just the two of you [I With your or your husband's children O Other (Specify) 4. Please list below each of the addresses at which you and your husband have lived 44nce your marriage. Street’ or RFD Present address City and County State or Foreign Country (If moved since marriage) Address before that Address before that Address before that Address before that 28 (Page 3) (GO ON TO NEXT PAGE) PART IV. INFORMATION ABOUT YOUR HOUSEHOLD In this part, information is asked about all the persons currently living in your household. 1. List below everyone who is living in your household at the present time. In addition to yourself, be sure to list your husband (if he lives at home), as well as your children (if any), other relatives and nonrelatives living with you. Do not include persons visiting you temporarily. For each person, provide the information requested below: Name Relationship to Yourself Date of Birth Marital Status Enter your name on the first line; enter|Relationship to you (husband, Specify one of the following: the names of all other persons who live |daughter, son, father-in-law, Single (never married), with you on the following lines: nephew, stepson, adopted (Month-Day- |Married, Separated, Widowed, daughter, lodger, etc.) Year) Divorced, or Annulled (First name) (Middle initial) (Last name) Yourself (14 mone space 4s needed, please continue on the back of pamphlet) 2. Who is the head of your household? 3. Is your husband presently serving in the Armed Forces on active duty? [J] Your husband LN Name of head O Yes OJ © [J another person =p 4, a. Have you ever had any babies or children in addition to those listed above? O Yes =p b. Please give the following information for each child Name of child Sex Month and Year|Is the child still . Te ————_) . . who is not living with you | (first name) (last name) |=—| of birth living? now. [OO Yes [] No [] Yes [J Wo [OQ Yes [J Xo (14 mone space is needed, please continue on back of pamphlet) [[] No == Go to the next page (Page 4) (GO ON TO NEXT PAGE) 29 PART V PERSON COMPLETING THIS FORM FULL NAME ADDRESS Street or RFD City State or Foreign Country TELEPHONE NUMBER DATE OF COMPLETION NOTES AND COMMENTS (Page 5) 30 HEALTH CARE QUESTIONNAIRE ONLY PART V. HEALTH CARE In this section and the next section, we are particularly interested in finding out about any recent on future medical care for pregnancy. 1. Are you now expecting a baby? [J Yes [J No —» Skip to question 3 2. a. Have you ever received medical care during this pregnancy? b. Who have you seen about care for this pregnancy? [1 physician in general practice [] Nurse [] Physician specializing in delivering [J midwife babies (Obstetrician) 0 Other (Specliy)d c. If you have received care for this pregnancy from a physician, where did you go for this [] Yes —» care? O Doctor's private office | Health Department O Hospital out-patient clinic O Other (Specify) d. If you have received care from a physician, during what month of your pregnancy did you first see him? Month of Pregnancy e. If you have NOT yet received medical care for this pregnancy, do you expect to receive care? [1 Yes [] No —b Skip to question 3 f. During what month of your pregnancy do you plan to receive medical care? a No > Month of Pregnancy g. Where do you plan to receive medical care for this pregnancy? Od Doctor's private office Od Health Department [] Hospital out-patient clinic [J other (Specify) 3. a. Have you lost a baby because of a miscarriage since your present marriage? "Yes [J] No —» Go to question 4 b. If so, please give the number of months you had been pregnant for each miscarriage. First Miscarriage Second Miscarriage Month of Pregnancy Month of Pregnancy 4. a. Have you been in the hospital overnight since you were married? Yes [J] No — Go to question 5 b. If so, what was wrong? (Briefly describe) 5. 2. Do you and your husband have health insurance to pay for all or part of a hospital bill? Yes [[] No =~» Go to question 6 5. If yes, would this insurance pay for all or part of the cost of care for the delivery of a baby? [J ves J we 6. ¢. Do you and your husband have health insurance to pay for all or part of a doctor's bill? J Yes [C] No —» Go to next page If yes, would this insurance pay for all or part of a doctor's bill for delivery of a baby? it Yes O No (Page 5) (GO ON TO PART VI) 31 PART VI. PERSON COMPLETING THIS FORM FULL NAME ADDRESS Street or RFD City State or Foreign Country TELEPHONE NUMBER DATE OF COMPLETION NOTES AND COMMENTS (Page 6) 32 FAMILY PLANNING QUESTIONNAIRE ONLY PART V. FAMILY PLANNING These questions relate to your plans for having children. 1. a. Have you ever thought about how many children 2. a. How many children do you think you will you would £{ke to have in the future? actually have in the future? OJ Yes =P bb. How many children would you Number like to have? b. If NONE, do you think you are able to have Number children? c. How many children would your O Yes [[] No =» Skip to question u husband like to have? Number [] No —p Go to question 2 - 3. a. Are you pregnant now? OJ Yes =~ Db. When do you expect your baby? Month Year [J No = c. When do you think you will have a baby? [J 1969 OO 1970 [1d 19m 1372 or later 4. a. Have you or your husband ever used any [[] Rhythm, safe period 3 s % methods to keep you from having children? J Rubber, ecnden, safe O Yes —® Db. Please check each method you J Diaphragm or your husband have used to »—& preg keep you from having children ] Jelly or cream [J No — c. Do you think that you or your [J roam (Emko, Delfen foam, etc.) husband will use some methods J Souchie to keep you from having children? OJ Yes 4. Please check each 1 Se emivectptive (the pill, Enovid, method you think in, etc. you or your husband )~ Od Coil, loop, intrauterine device (IUD) will use to keep you $13 : y from having children J Sterilization (tying tubes, etc.) [C] No wp Go to Part VI on next page [] withdrawal [] Other (Specify) PART VI. PERSON COMPLETING THIS FORM FULL NAME ADDRESS Street or RFD City State or Foreign Country TELEPHONE NUMBER DATE OF COMPLETION NOTES AND COMMENTS O00 33 APPENDIX II SAMPLING PROCEDURES, METHODS OF ESTIMATION, AND STANDARD ERRORS Sampling Procedures About 97 percent of marriage licenses issued in North Carolina are filed in the State Board of Health within 10 days after the end of the calendar month in which the marriage occurred. Two months after the month of marriage the records have been processed and punched cards are available for use in sampling as well as other processing of marriage data. Sampling for the mail survey was done separately for each principal month of marriage, i.e., using all licenses filed for a given month including roughly 3 percent which had occurred in an earlier month but were filed with the Register of Deeds during that cal- endar month, Table 1 shows the study population and the combined sample for the whole study period. In this appendix the details of the sampling procedure and the way the combined sample was obtained will be illustrated using data for the principal month of June 1968 marriage records. For each principal month of marriage the tabu- lating unit of the North Carolina State Board of Health sorted the punched cards into the strata shown in table I, counted them, and prepared a listing in State file number sequence within each stratum. The number of marriages required by the sample design” was se- lected at random within each stratum, ‘This number was either six, 12, or 18 marriages for the six-county area (Alamance, Durham, Guilford, Orange, Wake, and Forsyth Counties) or nine, 18, or 27 marriages for the rest of the State, depending on whether one, two, or three time duration subsamples were to be taken from that month's records. Table II shows how time duration subsamples were chosen from each principal month of marriage, e.g., February, March, December, and January each contributed only one subsample, while April, May, October, and November contributed two subsamples each and the remainder, June, July, August, and September, contributed three subsamples each, b ; Note that the required numbers for women 45 years of age and over were a third of those for the other ages because they were sent only one of the three questionnaires. Table I. Stratification of brides by previous marital status, race, and age of bride, and area: North Carolina Study Population, June 1968 Six-county area Rest of State Never Previously Never Previously Age of bride married married married married White | Other | White | Other | White | Other | White | Other Total-===---mecc mmm meen eee 902 196 136 30 | 3,870 592 599 65 Under 20 years-----=====m==============- 391 65 7 1 2,156 265 26 JL 20-29 years----=====---eeememmeeme—eaooao 497 123 59 411,657 292 235 16 30-44 year§---------mm-mmmecmmmmme memo 13 7 44 18 44 31 192 26 45 years and Over--=-------c---co-eononn 1 1 26 7 13 4 146 23 IBecause of small frequencies for brides other than white, years of age and in age groups under 30 years of age were combined to pling. 34 previously married brides under 20 form one stratum for sam- Table II. Time duration in months since marriage for subsamples by principal month of marriage and month of mailing: North Carolina Marriage Survey, 1968-69 Month of mailing Principal month of marriage 1968 1969 November | December | January | February | March | April 1968 Number of months February-----=--=-=-====--e-----cecommo-o== 9 - - - - - Marche=-===--c-cccmcmmcm meme mcccmmmmmm mmm - 9 - - - April------co-cmmmmmmmmmemmmomommmmmm momo 7 - 9 - - - MAY == =m rm om om om om mm 0 - 7 - 9 - - JUune=======m-=----mmemememmee-osse-s------- 5 - 7 - 9 - July===-==c-momcemmememmmmmmmem——ooomm—— m= - 5 - 7 - 9 August ---=====m--mm-em—-------somemmm—--o-oo 3 - 5 - 7 - September ---=-==-===-=-=-=-----o--o-o--=--- - 3 - 5 - 7 October ====---mm-emeeecmeemmmmm mmo omoam me - - 3 - 5 - November -==--===-==--emmmemmmmme—eoo—o— === - - - 3 - 5 December --====---mmee-cmemmemmm—mmeem——————— - - = - 3 - 1969 January----------=--===---------====--=----- - - » - - 3 Table III. Number of sample brides by time duration since marriage, month of mailing, area, and marital status and race of bride: North Carolina Marriage Survey, June 1968-69 Six=-county area Rest of State Time Sapagion Sines marriage, month Never Previously Never Previously of wai ing, Fns 288 ak marriage married married married married White | Other | White | Other | White | Other | White | Other 5 MONTHS SINCE MARRIAGE November 1968 mailing Under 20 yearS========scsceccccecemcce=== 6 6 3 } Iq 9 9 9 15 20-29 yearS===em-me=s=mcee-ccmcceso-ececee= 6 6 6 9 9 9 |f 30-44 yearSe=-=--mscceccececmmsssccenenaa= 4 2 6 6 9 9 9 8 45 years and OVere-eseeceeaccemee—carccaan 1 - 2 2 3 2 3 3 7 MONTHS SINCE MARRIAGE January 1969 mailing Under 20 yearS===--s=ee-ecec-e-ceeecana=- 6 6 2 } 1 9 9 8 } 15 20-29 yearSe====--e-cescccescesemcemee--= 6 6 6 9 9 9 30-44 years=-s=me=sssmm-e-e-oe--e-cea--oo-o- 5 2 6 6 9 9 9 9 45 years and OVere=s=-ee-e-e-e--e_ccac-aa= - 1 2 2 3 1 3 3 9 MONTHS SINCE MARRIAGE March 1969 mailing Under 20 yearS=es==s-see-sceemceeece-aa=- 6 6 2 } 1p 9 9 9 } 16 20-29 yearse==-smess-cemce-ecesemcma-oa-a= 6 6 6 9 9 9 30-44 yearSe==e==s=m-e-mmmme-eecececmemeaa= 4 3 6 6 9 9 9 9 45 years and Over=ee-esececcc-cncccmcnea= - - 2 2 3 1 3 3 IBecause of small frequencies for brides other than white, previously married brides under 20 years of age and in pling. age groups under 30 years of age were combined to form one stratum for sam=- 35 Three time duration subsamples (5,7, and 9 months in table II) were chosen from June 1968 marriages and were included in November, January, and March mail- ings, respectively. These are shown for illustration in table III. For the full strata 18 or 27 marriages were selected from each stratum of table I and were ran- domly allocated to the three time duration subsamples as shown in table III, Thus month of marriage is partially confounded with time duration since marriage and winter months were overrepresented at the extremes of 3 and 9 months' duration while spring and summer were over- represented at 5 and 7 months' duration. Strata which were not full were sampled at the rate of 100 percent.” For sampling purposes pre- viously married brides other than white who were under 30 years of age were considered as one stratum al- though for analytical purposes they were subdivided. Within each sample stratum of table III, brides under age 45 were subdivided into three subgroups, each to be sent one of the three versions of the ques- tionnaire. Brides over 45 were sent only the basic questionnaire, On the first mailing every other ques- tionaire was sent to the bride at her address and the next was sent to the bride at the groom's address, Thus when all 6 months of mailing were combined as indicated in table II, the "full" strata had 36 and 54 marriages for the six-county area and rest of the State, respectively, divided uniformly among the three questionnaires. Two weeks after the first mail query, which was always sent by regular mail, nonrespondents were ran- domly subdivided into two subsamples for testing the effect of certified versus regular mail on follow-up response, For the second mailing one subsample was sent certified mail and the other regular mail. Two weeks later a third questionnaire was sent by regular mail to all remaining nonrespondents regardless of what type mail had been used for the second mailing, This feature of the design made analysis of differences be- tween certified and regular mail response more com- plicated because of the built-in correlations between first wave and later results, Multinomial Model for Stratum Response Rates and Variances Type of response, timing of response, and type of second mail query were combined and condensed to create six multinomial "response categories' for clas- sifying sampled brides. “Note that month to month variations in the size of some strata may have resulted in less than 100 percent in the combined sample because no more than the required number were taken when strata were full. 36 All told, there were 24 area by marital status by race by age strata under age 45 years. Elimination of previously married brides other than white under age 20 years reduced this to 22 strata, Within each of these 22 strata there were 12 questionnaires by time duration strata, yielding a total of 264. Within each of the 264 sample strata the brides were classified as follows: Response category, time Stratum and type of response, and type of mail for second query Frequency | Proportion Total, all waves-= 'n,, A =1 FIRST WAVE (15 days or less) No second mailing l---Respondent===eeea-- - n, hi 2e=-0Othereeeeseccccaccaax n, A i2 Pia SECOND OR THIRD WAVE (16 days or more) Certified mailing 3---Respondente=eeamceux ny bis A 4ew-Othereeeececaccaaaa. fig bia Regular mailing S5--=Respondentece-ecececeac ng Pig 6==-Otherececeeaca- ————— Ne bie NOTE: », for each full stratum was 12 and 18 for the six-county area and for the rest of the State respectively. Where ng refers to the number of brides in the jth re- sponse category of the ith stratum (i= 1,2...264). i Mo 1 n the sample size for the ah stratum. jp ==. n Within each stratum, multinomial sample pro- portions and their covariance matrix were used to estimate first wave and all wave cumulative response rates separately for certified and regular mail and for the difference between regular and certified mail, Cu- mulative response rates can be expressed as follows: Through first wave (I): 7, =}, Cumulative response through third wave (III): Certified: N Tei = pi * w, (29) Regular: R . Ti =P tw, (bs) Where: w_; =reciprocal of the proportion sent certified mail 8 A 3 A = AZ py Ey Zo Py and w, = reciprocal of the 6 6 proportion sent regular mail = 2 baf zp. j=3"iif;=s ii Under the simplifying assumption that the weights, w,; and w,; are nonstochastic, estimates of variances (ignoring finite population corrections) were made for each stratum of table I as follows. 1 A A war) = + [5 04,0] A 1 A A A A A var (7 ;) = [2 (1p) + 20; (Py Pig) + w, #3) (14.3) | 1 2 var (7) = n [4 (1=B;;) + 2w,; (Pin pis) + wy, rs (1-2) And since A A 1 A A AA var A) == [02 Gig) 1B) ~2u, w,; (Fa Bs) n + (w,,)? Bs) (1-)g) ] Estimates from 'full'' sample strata, i.e. n,=12 or 18 for six-county area or restof State, respectively, were tabulated separately by area and type of second mail and for the difference between certified and reg- ular mail, There were 73 and 101 full strata for six- county area and rest of State, respectively. Averages are shown in table IV, Thus rather than using individual variances for each stratum, estimated average variances of rates within stratum were used as follows: var (A) = (0.1751)/n, var (P_) = var (7) = (0.5329) /n,, var (A, 7.) = (1.2276) /n; . Limitations of Variance Estimates Properties of the asymptotic estimates var (1) | are unknown. Assuming that the weights w_and w, (the inverses of sampling fractions for the second mailing) were nonstochastic may have caused under- estimation of sampling variances. Use of the arith- metic mean of all sample variances caused underes- timation. Ignoring finite population correction factors causes overestimation of sampling variances. Therela- tive extent to which these factors influence results of this study is not known, However, the estimates are as- sumed to be accurate enough for the purposes of this pilot study. Weighted State Estimates of Response Rates and Standard Errors Because of the small frequencies in each stratum it was not possible to interpret the response rates di- rectly. Therefore small stratum estimates were com- bined to obtain estimates for major variables using the Table IV. Average variance of cumulative response rates by mailing and area: North Carolina Mar- riage Survey, 1968-69 Type of second mail (all waves) First il Area and mailing Certified Regular Difference number of strata n, [var (%,)] n,, [var *.)] n, [var [)| ng. [var (*, -7.)] Six -county area: (73 strata, n, =12)======= 0.1851 0.5147 0.5267 1.2263 Rest of State: (101 strata, n, =18) ======= 0.1678 0.5411 0.5112 1,2285 Weighted mean variance-e==== 0.1751 0.5329 0.5177 1.2276 0.5301 37 Table V. Approximate standard error of unbiased rates of tables 4-8: North Carolina Marriage Sur- vey, 1968-69 All waves First Table and variable wave Certified| Difference or certified- regular regular Table 4 Total === mm mmm mmm meme eee meen 1.3 2.2 3.3 Six-county area----=--=---c-ccmmce ccm cccmmm———aa 2.1 3.4 5.9 Rest of State--=-ececcm came 1.8 2.9 5.4 Marital status of bride=------ceoomcmmm meee 1.9 3.1 5.6 Color of bride=-=—-=-emec cmon eeee ee 1.9 3.1 5.6 Age of bride=----cemmcm mmc eee - 2.3 4.0 6.0 Version of questionnaire--------ecccommoommom cee 2,2 3.7 5.8 Time duration since marriage----------cecommmcmccccceeeeeeo 2.5 4.3 6.7 Table 5 Version of questionnaire x time duration since marriage------ 4.4 1:5 11.5 Table 6 Version of questionnaire x color of bride x previous marital status of bride---==-=c-ccmcmm memes 3.0 5.0 7.3 Table 7 Version of questionnaire x color of bride x age of bride previous marital status of bride----=-------c=cccccccmaaao-o 4.7 7.8 12.0 Table 8 Color of bride x age of bride x previous marital status of bride === come mm ee eee eee en 2.9 4.6 7.1 Color of bride x previous marital status of bride~=----------- 1.6 78 5.6 corresponding proportion of brides in the study popu- lation (table 1) as weights. Equal weights were used in averaging over questionnaires or over durations (one- i=1 k R=% 2,7, where for convenience the wave subscripts I, ¢c,and 7 have beenomitted. Andthe & are weights based upon population proportions, third for questionnaires and one-fourth for duration), In general weighted estimates of a response rate were expressed as d . . The sample design gave essentially equal weight to each question- naire. However, time durations of 5 and 7 months were underrep- resented and hence equal weights yield unbiased State estimates. 38 questionnaires, and/or time durations as ap- propriate, Approximate variances were calculated as follows: k var (R) =Z ¢ var ,)- i=1 Standard errors for rates in detailed tables 4-8 are shown in table V above. APPENDIX lI DEFINITIONS OF CERTAIN TERMS USED IN THIS REPORT Principal month of marriage (occurvence).—The monthly period in which the vital record was filed with the State Board of Health, Area of occurvence,—The two study groups of counties within North Carolina in which marriages took place: the six-county area consisted of central counties of Alamance, Durham, Forsyth, Guilford, Orange, and Wake, and the rest of the State consisted of the other 94 counties in North Carolina. Time duration since marriage,.—The average elapsed time in months between the principal month of marriage and the month of mailing the initial survey questionnaire. Wave (time) of response, —First wave respondents returned a completed questionnaire within 15 days from the day of the initial mailing; second wave respondents returned a completed questionnaire within 15 to 28 days of the first mailing; and third wave respondents returned a completed questionnaire within 29 to 100 days of the initial mailing. Certified mail. —The type of additional postage (costing $.30) which was used for one-half of the first follow-ups (second wave). A receipt was signed by the addressee or someone at that address when the ques- tionnaire was delivered; otherwise, the addressee was notified to pick it up at the local Post Office. The Post Office returned letters which were not picked up ap- proximately 2 weeks after the initial notice. Post Office return, —A questionnaire whichwasre- turned by the Post Office stamped undeliverable, no forwarding address, no such addressee, nosuch address, unclaimed, or refused. Adequate response, —A returned questionnaire in which the information on all priority items was reported. Priority items common to all three versions of the ques- tionnaire included State of birth, education, usual ac- tivity before and since marriage, employment, income, sources of income, residence before marriage, house- hold structure after marriage, and date of birth for both bride and groom. Hospitalization coverage for the health care version was the only other priority item. Requery.—Special forms mailed to respondents who did not return an adequate response. Those items which were not completed properly were checked and the respondent was asked to complete and return the form. Colov.—The division of the population into two major groups on the basis of information reported on the marriage certificate. Races other than white include persons of Negro, American Indian, and Asian Indian races. Age.—Age at marriage based on date of birth, Previous marital status—The marital status of persons prior to the current marriage (never married or previously married) as reported on the marriage license. Income,—The present annual total income of the bride and the groom recorded separately. Household structure. —The type of group of one or more related or unrelated persons who occupy the same dwelling unit. A household with no relatives other than head, spouse, and children is classified as nuclear. A household including parents, relatives, and other persons is defined as "extended." Query.—The mailed questionnaires used in the sur- vey. Percent consistent, —Percent of consistent re- sponses to an item common to two record sources for which information was provided on each record. memes} (remem 39 APPENDIX IV ESTIMATED AMOUNT ADDED BY INTERVIEW FOLLOW-UP OF REFUSALS AND NONRESPONDENTS Both respondents and nonrespondents were sam- pled for interview follow-up. Respondents were inter- viewed to test the consistency and the quality of data elicited in the mail survey and on the marriage record. Results for respondents were reported in the text and are not included here, This appendix is limited to estimating the increase in response which would result from interviewing samples of refusals and nonrespond- ents, The number of cases on which the estimates are based is very limited, 41 refusals and 173 nonrespond- ents from the mail survey. These exclude those clas- Table VI. Distribution of mail survey sample by interview sampling and eligibilit sified as sampled, i.e., not eligible because the most recent address, either on the refusal or on the mar- riage record, was outside the six-county area. These cases are shown in table VI along with the notation which will be used to explain the estimation procedure. Because certified and regular mail categories were established only at the time of the second mailing, this sampling fraction as well as the interview rate must be taken into account in estimating the amount which would have been added if the mail sample had been carried out completely with either certified or regular mail, vey wave and response category: North Carolina Marriage Survey, 1968-69 status and mail sur- Mail survey wave and response category Second and third waves Interview sampling First wave and Total C ified i Sligarelity ert ed mail Regular mail status Non- Non- Non-=- Re- Re- Re- Re- re- re- re- spond- d- | Refusal d- pons fusal | spond- Spon Sings spond- spond Refusal spond- ent ent ent Total---- | 1,283 351 8 13 206 28 229 170 18 260 Not sampled---- 711 205(n;) L(ng) 8(ng) 113(n; 4) 3(nq) 138(ny;) 85(nyq) 1(nyg) 157(n35) Sampled: Not eli- gible------- 125 45(n,) L(ng) L(nyq) 26(ny,) 4(nyg) 11(n,,) 20 (nye) 3(ngq) 14(nq,) Eligible: Inter- i wa 289 | 79(ng) | 2(ny) 2(ny,) 54(n;¢) 6(nyq) 36(nyq) | 54(nyy) 2(ng;) 54(ngq) viewed=-====- 158 22(n,) 4(ng) 2(ny,) 13(ny¢) 15(n,0) 44(ny,) 11(n,g) 12(ng,) 35(ng4) 40 Within each of the 6 months of mailing interview subsamples were selected at random (and at different rates) from the three categories—respondent, refusal, and nonrespondent—without regard to the wave (or time) at which the result had been categorized. Sampled cases were then classified as eligible for interview if the address was in the six-county area, Eligible cases were then classified as interviewed or not interviewed depending on results, At that stage the results were tabulated in the detail shown in table VI for all cases. Using the notation above the total number of 36 cases in the six-county area = n, = 1283, i=1 12 Total cases, all classes, in the first wave = 3 n, = 372, i=1 Total, all classes, second and third waves combined are: 24 i ail = for certified m RA = 463 . 36 for regular mail = X », = 448, etc. i=25 It was assumed that interview rates among those not eligible would have been the same as among eli- gibles if they had been traced. Thus the amounts added by interview were estimated as follows: n, 7 A; = > Zn n, +ng 5 = amount added by interview of first wave refusals, ny, 12 4, = —— In; T i Way yy ff 9 =amount added by interview of first wave nonre- spondents . M19 2 36 36 2 #3 n. Tn In. / In Nig tn, 19 ! 7! 130 13 - amount added by interviewing certified mail re- fusals, / nya 24 36 36 24 A," Zn, z n, z n./ x n, nog t Nog 21 1 13 13 = amount added by interviewing certified mail non- respondents, Nay 32 36 36 36 Ag =| —— Zn, n; Zn; / En nay + Nay 29 1 13 25 = amount added by interviewing regular mail re- fusals, and Ny 36 36 36 36 Ag = —— In, / Tn zn, / Zn, n 35+ Nag 33 1 13 25 Estimates based upon the data in table VI, i.e., relative to the totals, are: Amount added by interview of: Percent First wave refusals, 4, ==e=ccee-- 0.2 Certified mail refusals, 4; =e==-- 1.2 A, Ply ceamimm—— (1.4) First wave nonrespondents, 4, ====- 0.5 Certified mail nonrespondents, 4, = 15.8 Ay + Ay mmemencncan (16.3) First wave refusals, A, =ee=veecee- 0.2 Regular mail refusals, 4, ==-===-- 0.4 Ay +t Ag mecmcemenen (0.6) First wave nonrespondents, 4, ==--- 0.5 Regular mail nonrespondents, dz = 25.0 Ay +t Ag ceccmamcaan (25.5) Thus the total amounts added by interview of re- fusals and nonrespondents are: Certified mail: 4, +4, +4,+4,=17.7 percent, and Regular mail: 4, + 4,+ 44 + A, = 26.1 percent Similarly, estimates of amounts added by inter- view were made by race, age, and marital status. These are shown in table VII with weighted results from table 8. It is clear that the largest estimates of amounts added by interview are for those groups making up the smallest fractions of the study population of brides, For example, never married white brides accounted for 70 percent of the marriages in the State. When weighted according to the proportions in the study pop- 41 Table VII. Estimated amount added by interview of Marriage Survey, 1968-69 refusals and nonrespondents: North Carolina Weighted percent Amount added mail response by interview Total Previous marital status, color, and age of bride - - - 8 eseel Regular boul Regular Certl Regular Never married White Under 20 yearg=----==-=—eeeoemmmemeaaan- 68.2 61.4 13.1 13.3 81.3 74.7 20-29 years--=====--mcmeemeemmmemmmeooo 72.7 61.7 6.1 9.5 78.9 70.2 30-44 years--------cemmmemmmmmceemeeaao 47.1 52.7 5.3 14.1 52.4 66.8 Other Under 20 years-----—=-----mmeccccmcaaaa- 72.7 67.9 20.4 19.0 93.1 86.9 20-29 years=--=------=-eccemcccoeacanao- 67.8 64.1 3.7 - 81.5 64.1 30-44 years---—---e-emmccmcmmemmceeeeo 58.9 45.5 24,2 46.8 83.1 92.3 Previously married White Under 20 years=---------ececccmmeaoaao- 54,2 50.6 6.6 37.6 60.8 88.2 20-29 yearS=-=----=--emeemmmmemmmeeeee 50.1 39.7 9.9 11.6 60.0 51,3 30-44 years=-=--=-mcmceccmmeemeemmeem- 39.5 36.8 25.7 32.7 63,2 69.5 Other Under 20 years---=----e-ccceecoecoaaaax * * * * * * 20-29 year§=----------m-memcmmmmmeeaoo 48.2 49.1 37.0 23:7 85.2 72.8 30-44 years--------c-mmmccmcnemeeeaa- 61.0 41.4 14.3 37.9 75.3 79.3 ulation the total amounts added in a random sample would be: Certified: 11.0 percent Regular: 13.4 percent 42 Even so it would appear that combined mail-in- terview response rates of 80 percent or higher are possible except for white brides who were previously married and/or over 30 years of age. 000 # U., S, GOVERNMENT PRINTING OFFICE : 1973 515-214/84 Series |, Series 2. Series 3, Series 4, Series 10. Series 11. Series 12. Series 13. Series 14, Series 20. Series 21. Series 22. For a list of titles of reports published in these series, write to: OUTLINE OF REPORT SERIES FOR VITAL AND HEALTH STATISTICS Originally Public Health Service Publication No. 1000 Programs and collection procedures.— Reports which describe the general programs of the National Center for Health Statistics and its offices and divisions, data collection methods used, definitions, and other material necessary for understanding the data. Data evaluation and methods vesearch.—Studies of new statistical methodology including: experi- mental tests of new survey methods, studies of vital statistics collection methods, new analytical techniques, objective evaluations of reliability of collected data, contributions to statistical theory. Analytical studies.—Reports presenting analytical or interpretive studies based on vital and health statistics, carrying the analysis further than the expository types of reports in the other series. Documents and committee reports.— Final reports of major committees concerned with vital and health statistics, and documents such as recommended model vital registration laws and revised birth and death certificates. Data from the Health Interview Survey.— Statistics on illness, accidental injuries, disability, use of hospital, medical, dental, and other services, and other health-related topics, based on data collected in a continuing national household interview survey. Data from the Health Examination Survey.—Data from direct examination, testing, and measure- ment of national samples of the population provide the basis for two types of reports: (1) estimates of the medically defined prevalence of specific diseases in the United States and the distributions of the population with respect to physical, physiological, and psychological characteristics; and (2) analysis of relationships among the various measurements without reference to an explicit finite universe of persons. Data from the Institutional Population Surveys.— Statistics relating to the health characteristics of persons in institutions, and on medical, nursing, and personal care received, based on national samples of establishments providing these services and samples of the residents or patients. Data from the Hospital Discharge Survey.—Statistics relating to discharged patients in short-stay hospitals, based on a sample of patient records in a national sample of hospitals. Data on health resources: manpower and facilities, — Statistics on the numbers, geographic distri- bution, and characteristics of health resources including physicians, dentists, nurses, other health manpower occupations, hospitals, nursing homes, and outpatient and other inpatient facilities. Data on mortality.—Various statistics on mortality other than as included in annual or monthly reports—special analyses by cause of death, age, and other demographic variables, also geographic and time series analyses. Data on natality, marriage, and divorce. — Various statistics on natality, marriage, and divorce other than as included in annual or monthly reports—special analyses by demographic variables, also geographic and time series analyses, studies of fertility. Data from the National Natality and Mortality Surveys. —Statistics on characteristics of births and deaths not available from the vital records, based on sample surveys stemming from these records, including such topics as mortality by socioeconomic class, medical experience in the last year of life, characteristics of pregnancy. etc, Cffice of Information National Center for Health Statistics U.S. Public Health Service Rockville, Md. 20852 600 DHEW Publication No.(HSM) 73-1330 Series 2 -No.56 Sd Nl 57 ] JITAL and HEALTH STATISTICS DATA EVALUATION AND METHODS RESEARCH IN it¥, = fo) NCHS nd 9 h/4 ~ 1 Net Differences in Interview Data on Chronic Conditions and Information Derived From Medical Records U.S. DEPARTMENT OF HEALTH, EDUCATION, AND WELFARE Public Health Service Health Services and Mental Health Administration Vital and Health Statistics-Series 2-No. 57 For sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, D.C. 20402 Price 85 cents domestic postpaid or 60 cents GPO Bookstore Series 2 DATA EVALUATION AND METHODS RESEARCH Number 57 Net Differences in Interview Data on Chronic Conditions and Information Derived From Medical Records A methodological study of the completeness and accuracy with which chronic conditions are reported by health plan enrollees in household interviews as compared with infor- mation recorded by physicians. DHEW Publication No. (HSM) 73-1331 U.S. DEPARTMENT OF HEALTH, EDUCATION, AND WELFARE Public Health Service Health Services and Mental Health Administration National Center for Health Statistics Rockville, Md. June 1973 NATIONAL CENTER FOR HEALTH STATISTICS THEODORE D. WOOLSEY, Director EDWARD B. PERRIN, Ph.D., Deputy Director PHILIP S. LAWRENCE, Sc.D., Associate Director OSWALD K. SAGEN, Ph.D., Assistant Director for Health Statistics Development WALT R. SIMMONS, M.A., Assistant Director for Research and Scientific Development JOHN J. HANLON, M.D., Medical Advisor JAMES E. KELLY, D.D.S., Dental Advisor EDWARD E. MINTY, Executive Officer ALICE HAYWOOD, Information Officer DIVISION OF HEALTH INTERVIEW STATISTICS ELIJAH L. WHITE, Director ROBERT R. FUCHSBERG, Deputy Director RONALD W. WILSON, Chief, Analysis and Reports Branch KENNETH W. HAASE, Chief, Survey Methods Branch COOPERATION OF THE BUREAU OF THE CENSUS Under the legislation establishing the National Health Survey, the Public Health Service is authorized to use, insofar as possible, the services or facili- ties of other Federal, State, or private agencies. In accordance with specifications established by the Health Interview Sur- vey, the Bureau of the Census, under a contractual arrangement, participates in most aspects of survey planning, selects the sample, and collects the data. Vital and Health Statistics-Series 2-No.57 DHEW Publication No. (HSM) 73-1331 Library of Congress Catalog Card Number 73-600034 FOREWORD A continuing concern and effort of the Na- tional Center for Health Statistics has been to better assess the effectiveness of its survey data collection mechanisms. Through the means of household interviews, examination surveys, and record surveys, a large variety of data, some of it overlapping, has been collected. Pro- gram plans and objectives have madeitimpera- tive that research be conducted to evaluate the strengths and weaknesses of the various sur- veys and thus to concentrate the efforts on those objectives best performed in each particular survey. Important questions with respect to inter - view surveys have continued to be How complete is the reporting of chronic conditions by house- hold respondents? and What is the value of con- dition data collected by household interviews? A large-scale study was conducted in collabora- tion with the Health Insurance Plan of Greater New York to compare the information collected in household interviews with that found in ex- isting medical records. (See "Health Interview Responses Compared with Medical Records," Vital and Health Statistics, PHS Pub. No. 1000- Series 2-No. 7.) This study probed many facets of the agreements and disagreements to be found in such comparisons, It also indicated the need for a more sophisticated study plan which would utilize a prospective record source designed to control for differences in communication be- tween physician and patient, for the duration of the condition, and for some measures of theim- pact of the condition as correlates of the meas- ures of completeness of reporting in health in- terviews, Such a study was planned as a contract proj- ect with the extensive collaboration of the Stan- ford Research Institute, the Kaiser Foundation Health Plan (Southern California Region), South- ern California Permanente Medical Group, the U.S. Bureau of the Census, and the National Center for Health Statistics. The first report (Series 2, No. 23), is a description of the study, in which the chronic illnesses and impairments reported by a sample of persons in household interviews were compared with the chronic ill- nesses and impairments found in specially pre- pared medical records. The study population consisted of a sample of members of a prepaid medical and hospitalization plan. The general objectives of the study were: 1. Ascertaining the extent of reporting by respondents in household interviews of conditions for which medical care was sought over a period of 12 months, 2. Relating the extent of reporting of con- ditions to some measures of communi- cation between physician and patient; to the relative impact of the condition in terms of duration and number of physician visits; and to type of treat- ment, 3, Experimenting with different versions of the health interview questionnaire. This is the second report from the study and it deals primarily with overreporting and under- reporting of specific chronic conditions in house- hold interviews. Dr. William G, Madow of the Stanford Re- search Institute served as project officer for this study and was responsible for the prepara- tion of this report. Mrs, Louise Bollo served as nosologist, and Mrs. Geraldine Gleeson per- formed major editorial service in preparing the report for publication. Elijah L. White. Director, Division of Health Interview Statistics SYMBOLS Data not available-=ccccammcccc eee Category not applicable--=eemmcoecacaoaan Quantitv Zero---------ccmmcmmccmmmmmmeoo Quantity more than O but less than 0.05---- Figure does not meet standards of reliability or precision (more than 30 percent relative standard error)--------- CONTENTS Page Foreword ---===-=-mmcmmmcmmcmm ce mee meee emm mmm mmemmeommomom momo iii Objectives and General Findings-------==----=-memmecooccmmommcoonnonn 1 Background -----------c-mmmmmm mmm mmm omeooooooeomooeooommmo oo 1 Content of Earlier Report----------c-covmomomommmococaocooo comme 1 Planning and Conducting the Study----=-=-=--==-=-mcccccmcomomnoonnoaon- 2 Earlier Research on Health Interview Data---------=-==c-m-"-vo-coo--- 2 Data Collection for the SRI Study-------=---====-cecmommcmcmcacacoano- 2 Analysis of the Data--------=cc-moommmmmmmmccomccm moomoo mmm mm ooo 3 Chronic Conditions by Type of Medical Services Used-----======---------- 3 Type of Medical Services-----=-=-=-==mmmmmccmmoooooccommooom momma 3 Distribution of Chronic Conditions-----=--=-=c-c-eeccmmcmmcccanaocoomoo 4 Net Differences in Interview Reports and Medical Records--=--=-=-====-=------ S Underreporting and Overreporting of Chronic Conditions---------------- S Net Reporting Differences------==-=-=m-mmemmcmoocommcmm mmm em mmm 5 Completeness of Reporting by Frequency of Physician Visit§------------ 6 Completeness of Reporting by Presence or Absence of Medication------- 7 Completeness of Reporting by Sex and Age-----===--=--=-=--co-zonocm- 7 Completeness of Reporting by Educational Status----------------------- 8 Comparability with HIP Study Findings------===-==----c--cocmm-uoomonn- 8 SUMMATY=========== o-oo oooeooooo-oo-ooo---oe- 8 List of Detailed TableS-========m-cmcccmccmmcmm mm mmmmmmmmmmmmcmommmoeo 10 Appendix I, FOorms---------cccoiommmmmmmoemm mm mcmmmmommm moomoo 27 Version One of Questionnaire----------=cececemmmmmmooomoooocooonoomn 27 Version Two of Questionnaire-------------coecommmoomocccooaooonomo- 35 Version Three of Questionnaire-----------c--cocmmmmmmmmmocaooooooommn 43 Physician Visit Record---==-======--=ccommmmommo mmm mmm meen 51 Sample of Completed Physician Visits Record Summary---------------- 52 Appendix II, Diagnostic Recode------------=-mcoomoeemcmmmcnennncnenee 53 Appendix III, Sampling Design-----------=mm-cmcmmomoooommmmoomnomooonn 56 Introduction-=-===-=- =m mmm emmmmmmmmmmmooooooo- 56 Family Account Numbers and Medical Record Numbers at KFHP-------- 56 Population =========m ooo mmo meme mmmmm moomoo 56 Selection and Assignment to Interviewers of the Interview Sample-------- 56 Interview Sample for Which PVR's Were Not Used------=-------------- 58 NET DIFFERENCES IN INTERVIEW DATA ON CHRONIC CONDITIONS AND INFORMATION DERIVED FROM MEDICAL RECORDS William G. Madow, Ph.D., Stanford Research Institute OBJECTIVES AND GENERAL FINDINGS Background As a part of its continuing program of studies designed to evaluate the accuracy and complete- ness of diagnostic information obtained by house- hold interview, the Health Interview Survey con- tracted with the Stanford Research Institute to do a study comparing responses to health interviews with medical records. The population used for the study was a sample of the members of the Kaiser Foundation Health Plan (KFHP) Southern Cali- fornia Region--a large prepayment medical plan providing services through the Southern California Permanente Medical Group (SCPMG) and hospital- ization through Kaiser Foundation hospitals. The data collection phase of the study consisted of com- pleting medical records created specifically for the study and then interviewing the persons for whom these records were maintained, Content of Earlier Report The general findings of the study, together with conclusions and recommendations pertinent to theinterview survey, are presentedin an earlier publication issued by the National Center for Health Statistics entitled "Interview Data on Chronic Conditions Compared With Information Derived from Medical Records' (Vital and Health Statistics, Series 2, No. 23), That report also in- cludes a description of the backgroundandobjec- tives of the project and some of the problems en- countered during the conduct of the study. As in other evaluative studies of this kind, a principal finding was that a certain proportion of conditions listed in the medical records were not reported in the household interviews. In general the unreported conditions tended to be those for which there was little, if any, impact on the per- son involved. Respondents reported more fully on conditions important to them and reported less well on conditions of lesser subjective impor - tance, Most researchers, while doing their utmost to reduce errors of response, are aware of the fact that there are errors of reporting in two directions: understatement and overstatement. The costs and difficulties of eliminating response bias are often so great that researchers take ad- vantage of the extent to which understatements and overstatements balance one another in indi- vidual estimates, i.e., the size of the net response bias. In some instances, relationships may hold up even when individual estimates are subject to fairly large errors of reporting. By considering only whether conditions found in the medical rec- ords had been reported in the household inter- views, the earlier report considered the gross error of reporting in only one direction. No at= tempt was made to investigate the extent to which underreporting of conditions found in the medical records was balanced by overreporting of condi- tions in the household interview, The general purpose of the present report is to investigate the extent and magnitude of net dif- ferences in the conditions reported in household interviews and those recorded in the Physician Visits Record Summary,! On the whole, while differences do exist, there is a tendency for the gross errors to balance out and for the net bias tobe relatively small, partic- ularly in view of the frequent vagueness and un- certainty in diagnosis and the lack of precision with which patients understand diagnoses. Even though only a small number of comparisons have been made here, it seems that the differences between chronic conditions in the medical rec- ords (PVRS's) for the study year and those re- ported in the household interviews are small enough that the findings can be used for some evaluative purposes, The large sizes of thegross errors, however, still require attempts to im- prove the data, PLANNING AND CONDUCTING THE STUDY Earlier Research on Healtn Interview Data Recognition of the fact that information on illness collected by household interview does not reflect a complete and accurate account of all chronic conditions present in a population led to a number of research studies during the early years of the National Health Survey, In one of these studies,? carried out by contractual ar- rangement with the Health Insurance Plan of Greater New York (HIP), the use of medical serv- ices for a condition during a given year was established from records maintained by HIP, and information collected by interviews was examined in relation to this criterion source, The record source in this study was the reporting document 1A form—the Physician Visit Record—was filled out by the physician for each sample person after each visit to SCPMG during the study year. At the end of the study year, the Physician Visit Records were summarized for each person; this summary is called the Physician Visits Record Summary. For more detailed information on this form, see Series 2, No. 23, page 7. 2¢“Health Interview Responses Compared With Medical Records,” Series 2, No. 7. (Med 10) which HIP physicians submitted to the central office in accordance with operational pro- cedures of the Plan, These records, consisting of single-line entries on an administrative form, were used instead of the entries on the patient's clinical chart because the wide geographic dis- persal of the medical groups and the variety of recordkeeping systems precluded the examination of all physician entries for a given individual. Since the Med 10 form gave no medical history, evaluation of symptoms, nor weighing of differen- tial diagnoses, conditions and their chronicity were inferred from the records. While studies carried out within HIP have indicated that the Med 10 is a reliable document for statistical purposes, it was somewhat less than ideal for useas a cri- terion relating to the presence of diagnosed chronic conditions, While the retrospective study conducted by HIP yielded valuable information, it was felt that its findings should be confirmed in a different popu- lation and that other aspects of interviewing prob- lems could be investigated in a prospective rec- ord-check study. This plan for further research led to the arrangement with the Stanford Research Institute (SRI) to conduct such a study. Some of the comparative features of the two studies are dis- cussed later. Data Collection for the SRI Study An important innovation in planning the pro- spective study undertaken by SRI was the creation of medical records to be used especially for the study—the Physician Visit Record (PVR)—which was filled out by the physician following each phy- sician-patient visit, In preparing the PVR, sum- marized in the Physician Visits Record Summary (PVRS), the physicians were asked to enter any diagnosis (condition) impression or symptom that was considered, noted in the record, or mentioned by either the physician or the patient, The con- dition category noted in the record referred to conditions that the doctor had entered in the pa- tient chart regularly filled out after each visit. It was quite possible that the patient had various conditions never mentioned in his meetings with the physician during the study year; such condi- tions would not have been entered on the PVR and thus would not have been summarized onthe PVRS, The physician aid not always enter onthe PVR all conditions that he noted on the patient chart dur- ing a visit, Some patients received part or all of their medical care outside SCPMG. Conditions reported by such patients would not necessarily appear in their medical records maintained at SCPMG. In estimating net differences it therefore seemed de- sirable to limit the study to data for persons who reported that they had used only SCPMG as a source of medical assistance during the study year. The study included only those conditions which were entered in the medical records or about which the respondent said he had spoken with a physician during the year.Because of these limitations the basic comparison in this reportis between conditions found in the medical records \ (PVRS's) created for this study and conditions \ reported in the household interviews conducted | after the completion of the study year. 7 The interviews, which were conducted by the U.S. Bureau of the Census acting as collecting agent for the Division of Health Interview Statis- tics, National Center for Health Statistics, per- tained to conditions which were diagnosed or for which medical treatment had been received during the study year. The formats of the questionnaire used in the interviews, the Physician Visit Rec- ord, and the Physician Visits Record Summary are shown in appendix I of this report, A descrip- tion of the sample design can be found in appendix III. The questionnaire and corresponding PVRS for each patient were sent to the Division of Health Interview Statistics, which undertook demographic and medical coding of the study data. Transcrip- tion sheets were prepared, and chronic conditions on the questionnaire and PVRS were identified, compared, and matched. Analysis of the Data Once the chronic conditions had been identi- fied, they were assigned the three- and four-digit diagnostic codes of the Seventh Revisionofthe In- ternational Classification of Diseases (ICD). The codes were summarized into a classification of 50 diagnostic categories similar to the Recode 3 used in the Health Interview Survey with each of these 50 classes consisting of chronic conditions with specified codes. (See appendix II.) If a chronic condition on the PVRS anda con- dition on the questionnaire had ICD codes within the same recode class, the conditions were as- signed match code A. If a chronic condition on the PVRSanda con- dition on the questionnaire had ICD codes that were not within the same recode class but appeared to be associated, the conditions were assigned match code B. (It is recognized that code B is not sharply defined.) Chronic conditions on the PVRS which were not assigned either match code A or match code B were assigned code C, A code C condition on the PVRS had no associated condition on the question- naire, If a chronic condition on the questionnaire was not assigned either match code A or B, it was as- signed code D, meaning that there was no associ- ated condition on the PVRS. Code D conditions about which the respondent reported that he had seen or spoken to a physician in the preceding 12 months were analyzed separately as D12 condi- tions, Code D conditions for which the respondent did not report receiving medical services during the 12-month period are D+ conditions. CHRONIC CONDITIONS BY TYPE OF MEDICAL SERVICES USED As mentioned earlier, inorder to consider the net effects of reporting errors, it was necessary to limit the study to persons who had utilized only SCPMG for medical services and only Kaiser Foundation hospitals for hospital services, Those interviewed were asked about the physicians they had contacted and hospitals they had used for their medical services during the study year and were then asked to authorize examination of their med- ical records. Type of Medical Services Table 1 shows the distribution of chronic con- ditions in the medical records and household interviews and of persons with chronic conditions according to the type of medical service utilized. The four classifications of these data are based on what physicians and hospitals the respondent reported using during the study year: (1) SCPMG and Kaiser Foundation hospitals only, (2) other physicians and hospitals in addition to SCPMG and Kaiser Foundation hospitals, (3) only non-SCPMG physicians and hospitals, (4) no physician or hos- pital services at all. The tabulation excludes per - sons reported by both the medical records (PVRS's) and the questionnaire as having no chronic conditions. Approximately 67 percent of all conditions re- corded in the PVRS's and/or reported in the inter - views were for persons who reported inthe inter- view that they had received only SCPMG services and whose utilization status was verified in the PVRS. An additional 15 percent of the conditions were for persons who had received services from other medical facilities as well as from SCPMG, according to entries on the questionnaire and the PVRS. For approximately 9 percentofthe conditions both PVRS ‘and interview indicated that either no SCPMG services or nomedical services whatever had been received. It should be pointed out that the 245 persons (shown in table 1) who reported the receipt of services from "SCPMG only" according to the PVRS but reported '"No utilization'' in the house- hold interview could have used the services of physicians and hospitals other than SCPMG. Ac- cording to the medical records these respond- ents were seen by an SCPMG physician or were hospitalized in a Kaiser Foundation hospital for 596 conditions during the study year. These re- spondents may also have had conditions that were diagnosed by physicians or in hospitals outside SCPMG, but rather than have a separate category for "SCPMG and possible others' it was decided to categorize them as "SCPMG only." An additional 172 conditions were recorded in the medical records (PVRS's) of persons who reported in the interview that they had used only non-SCPMG medical services. These may be viewed as reflecting a memory defect. The classification of utilization for deter- mining whether the person had utilized SCPMG was based on his PVRS, but the respondent's statements with respect to outside utilization or no utilization were accepted when they were not in conflict with the medical records (PVRS's). There were 470 conditions reported by re- spondents who said they had used only SCPMG services for which no record of SCPMG usage during the study year could be found on the PVRS's. Again, this seems to indicate a memory failure. Distribution of Chronic Conditions Of the 15,417 conditions found in either the medical records or in the household interview questionnaire, 14,099, or approximately 91 per- cent, were reported in the same manner in both sources with respect to the utilization of medical services. Approximately 88 percent (4,445 out of a total of 5,027) of the respondents reported the same utilization of medical services as was re- ported in the medical records. For persons who either reported at leastone chronic condition or had a chronic condition re- corded on their PVRS, table 1 shows the number of chronic conditions per person according to the utilization of medical services as reported bythe respondent and recorded on the PVRS, Persons who, according to both PVRS and questionnaire, utilized not only SCPMG but also other treatment facilities had about 20 percent more conditions per person than those who utilized SCPMG only, and persons in both of these categories had at least 50 percent more conditions per person than those who utilized only non-SCPMG sources and those who had no medical or hospital services. The distribution of chronic conditions by type of match (A, B, C, D12, or D+) according to utilization as recorded on the PVRS and reported in the interview is shown in table 2. More than one-third of the 15,417 conditions reported in the interview and/or recorded in the PVRS were conditions reported in the interview only for which the person did notreportreceiving medical services during the 12-month period preceding the interview (D+ conditions). Since the absence of medical treatment during the year pre- cludes their being recorded in the PVRS's, these conditions are not considered in the rest of this report, Of the 6,140 conditions recorded inthe PVRS's (match categories A, B, and C, shown in the first line of table 2), 3,359, or approximately 55 per- cent, were also reported in the interview (match categories A and B). When conditions are re- stricted to those of persons who received SCPMG services only, the percentage is slightly less, 54 percent, NET DIFFERENCES IN INTERVIEW REPORTS AND MEDICAL RECORDS Underreporting and Overreporting ot Chronic Conditions An important consideration in this study is determining if there is an interchange between the conditions identified as C conditions (reported only in the records) and D12 conditions (reported only in the interviews), i.e., whether the words used by the patient and doctor in describing the same condition are sufficiently different from one another to preclude an A or B match. Table 3 provides some whether a person tends to have equal numbers of C (underreported) and D12 (overreported) con- ditions. The table shows, for persons who utilized only SCPMG services, the distribution of condi- tions by number of D12 conditions and number of C conditions. (In both cases, the maximums shown in the table are correct; i.e., no individual had more than five C conditions or eight D12 condi- tions.) Clearly the association between C and D12 conditions is not great, Of the 3,401 persons who utilized only SCPMG services, 998, or 29.3 per- cent, had neither a C nor a D12 condition. An ad- ditional 228 persons, 6.7 percent, had the same number of C and D12 conditions, and 1,476 per- sons, 43.4 percent, had numbers of C and D12 conditions that differed by one. Among the sources of difference may be the reporting of an ailment as a single condition in one source and as more than one condition in the other source. Nonetheless, it is not reasonable to assume from the findings of this study that re- spondents were reporting C conditions as D12 conditions because of their failure to understand the nature of their conditions. The basic measures of completeness of re- porting in this study are presented in table 4 and summarized as indexes of reporting differences in table 5. It is evident from table 4 that the number of conditions reported in the medical records (5,279) for persons who had received only SCPMG serv- ices during the 12 months prior to interview was roughly 12 percent higher than the number of con- ditions reported in the household interview (4,714). This difference is due entirely to condi- information on. tions onthe PVRS only (C conditions) or reported in the household interview only (D12 conditions), and there are approximately 30 percent more C than D12 conditions. However, if one compares the percentages for all conditions reported in the PVRS's and all conditions reported in the house- hold interviews, the agreement is fairly good, and the differences that do exist seem to be logical. For example, one of the larger differences is in the category ''mental illness, specified types, not elsewhere classified." There is little difference in the less specific category "ill-defined mental and nervous trouble," and, in addition, when the questionnaires were examined, the entries found in the medical records for such conditions had names such as "anxiety," or ''tension,' or other words that might not appear to a patient perma- nently living with such conditions to be the med- ical reasons for which he had consulted a doctor. On the whole, considering the tendency to under - report mental illness and other illnesses that the respondent believes to be socially unacceptable, lack of agreement between medical records and interview data in this instance is not unexpected. Net Reporting Differences In a study of this kind it is difficult to estab- lish a proper denominator to compute a single index which will reflect the net reporting dif- ferences. As an alternative, two indexes are presented in table 5, One is the proportion of con- ditions found in the medical records but not | reported in the interviews (an estimate of under- reporting), and the other, the proportion of con- ditions reported in the interviews but not found in the records (an estimate of overreporting). Since these indexes, shown for each diagnostic category, are derived from the data shown in table 4, they are based on information about per- sons who used SCPMG services only for condi- tions that had been medically attended during the 1-year period covered by the PVRS's, When both of the indexes for a particular diagnostic category are comparatively low, itcan be expected, on the basis of this study, that con- ditions within the category will be reported in an interview with a fair degree of accuracy. If, in addition, the indexes are of the same general magnitude, the gross prevalence produced from interview data will approximate the unbiased es- timate of the true prevalence level in the popu- lation. If, on the other hand, either or both of the indexes for a category are high, then the estimates from the interview must be considered as suspect, even though two high indexes of the same magni- tude will produce an approximate gross preva- lence, Conditions with low indexes of underreporting and overreporting, which might be expected to be reported with a fair degree of accuracy and com- pleteness in a household interview, include dia- betes, vascular lesions of the central nervous system, heart conditions, diseases of the gall- bladder, and absence of fingers and toes. In evaluating this material it should be kept in mind that overreporting may have resulted from the fact that record data were limited to those conditions for which a person had seen an SCPMG physician during the year, Thus, respondents could have reported in the interview conditions of long duration or even presently inactive conditions they had many years ago which were not noted in the current medical records. This possibility may ex- plain some part of the overreporting in such cate- gories as tuberculosis, rheumatic fever, sinusitis, bronchitis, and severe visual impairment. Furthermore, data on the accuracy of chronic condition reporting in household interviews, shown in tables 4 and 5, should be interpretedin the light of some findings presented in the earlier report on this project (Series 2, No, 23). Some of the more pertinent paragraphs from the earlier publication that describe several of the shortcomings of the study follow. Communication between physician and patient seemed to vary considerably from condition to condition, Often in the discussion, reference is made to the fact that something was or was not entered during the visits at which the physician reported the condition on the PVRS. ... For 31.3 percent of the 6,140 conditions recorded on the PVRS, the physician stated that during no visit during the study year had he told the patient the actual diagnosis or a diagnosis codable to the actual diagnosis, Similarly, for about S51 percent of the con- ditions the physician stated that during no visit had the patient told him either the ac- tual diagnosis or used a term codable to the actual diagnosis-—i.e., neither a formal diag- nostic statement, lay terms, nor symptom statements related to the diagnosis had been used by the patient during his visits to the physician. Sometimes in speaking to a patient a physician emphasizes the condition from which the patient is suffering and sometimes he does not, For 54 percent of the conditions, the physician claimed that during no visit had he made a particular point of the diagnosis in discussing the condition with the patient, The physician was asked to enter on the PVR whether the patient reported having pain or emotional stress or spending at least 1 day in bed during the week preceding the pa- tient's visit, Approximately 70 percent of the conditions were such that at no visit did the physician indicate on the PVR that the pa- tient had had pain or emotional stress during the preceding week. For about 10 percent of the conditions, the physician stated that the patient had said he had spent at least 1 day in bed during the preceding week. Even though the percentages quoted above pertain to all conditions in SCPMG records (A, B, and C conditions intable 2), itis reasonable to as- sume that they also apply to conditions among per - sons receiving SCPMG services only. Completeness of Reporting by Frequency of Physician Visits Shown in tables 6 and 7 are distributions of conditions by number of physician visits they caused during the study year based on information from the PVRS's (table 6) and by number of physi- cian contacts reported in interviews (table 7). The data used for physician contacts were those stated by the respondent on the questionnaire, For 366 conditions table 7 shows no physician contact, but data from the PVRS's show that at least one SCPMG physician had been consulted. If these 366 conditions are included with the 1,592 conditions for which one contact was reported (table 7), the comparison of the percentages in the two tables, while certainly not perfect, is sufficiently close to provide information for evaluative purposes. This is true with respect to all conditions and also with respect to the specific comparison of C and D12 conditions. There is some tendency for the number of physician contacts reported for conditions inthe interview to be higher than the actual number of visits recorded in the PVRS because some of the physician contacts may have been by telephone rather than by personal visit. A high proportion of the conditions that were underreported (C conditions in table 6) and over- reported (D12 conditions in table 7) consisted of those for which a single physician visit or con- tact was made during the study year. As the num- ber of visits increased, the percent of conditions underreported or overreported declined sharply. This pattern indicates that increased opportunity for communication with the physician improves the ability of a respondent to report his condi- tions in an interview with accuracy and complete- ness, Completeness of Reporting by Presence or Absence of Medication In tables 8 and 9 the distribution of conditions included in the medical records and those reported in the household interviews is shown by type of match according to whether or not the personwas taking medicine for the condition. Approximately 56 percent of the conditions for which the medical records indicated that no medication had been prescribed were not reported in household inter - views (C match conditions in table 8). Only 33 percent of those conditions for which medication had been prescribed were not named during the interviews. The impact of frequent medical at- tention and regular medication, shown in tables 6 aud 8, is effective in reducing the amount of underreporting in the household interview, Overreporting of conditions was not unduly influenced by whether or not the respondent was taking medication. About 38 percent of those con- ditions reported in interviews as requiring medi- cation were not found in the medical records. The comparable proportion for those conditions with no medication during the study year was 45 per- cent (table 9). Completeness of Reporting by Sex and Age The distribution by sex of conditions recorded in PVRS's and reported inhousehold interviews according to match code indicates that the propor- tion of conditions in all categories was much higher among women than among men (table 10). How- ever, there was very little difference between the sexes in the underreporting of conditions; about 46 percent of the conditions for men shown in the medical records were not reported in the inter- view, while a very comparable percentage among women was 47 percent, However, the amount of overreporting was somewhat less among males than among females, No evidence of approximately 37. percent of the conditions reported intheinter- view by males was found in the records; among females the comparable percentage was 45 per- cent, Slightly more than two-thirds of all of the conditions recorded in the medical records for both sexes were among persons 35-64 years of age. About three-fourths of the recorded condi- tions not reported in the interview were among persons in this age range (table 11). When con- ditions in this age group are considered by type of match, the proportion of those in the records that were not reported in the interview was about the same among men and women, However, ding parable percentages shown in table 12 indicate that the proportion of conditions overreported in this age range was substantially higher for women than for men, The seemingly high percentage of conditions underreported by women 17-24 years of age, shown in table 11, can be attributed to the small numbers of total conditions among persons inthis age group. The instability of the numbers in this age group may also account for the high rates of overreporting (table 12). For persons 65 years and older, the accu- racy and completeness of reporting was substan- tially greater among women than among men, The percentages summarized from tables 11 and 12 and shown in table A indicate that the A and B match rates were much higher for women and that they underreported and overreported conditions less frequently. Table A. Proportion of chronic conditions re- ported among persons 65 years and over, by sex Proportion | Proportion Proportion | of condi- of condi- of A and B tions in tions re- Sex matches records ported in based on that were interview medical not re- that were records ported in | not in the interview records Male----==-- 56.8 43.2 38.2 Female----- 71:1 28.9 34.6 Completeness of Reporting by Educational Status Of all the conditions recorded in the medical records or reported in the interview, essentially one-half of them were among persons with 9-12 years of education. The distribution of all con- ditions by education was quite similar to that for recorded conditions that were not reported in interviews (table 13) and for reported conditions for which there was no confirming evidence in the records (table 14). This would indicate that edu- cation did not influence the amount of underre- porting and overreporting in this study to any appreciable degree with regard to total preva- lence of chronic conditions, Most of the disparity in reporting noted among educational groups when they are considered by type of match can be at- tributed to the small numbers of conditions in some of the groups. Comparability With HIP Study Findings It is not possible to compare the net differ- ences in interview data and medical record infor - mation derived from the HIP study and those of this study because the procedure used in the HIP study did not provide for the measurement of overreporting., Any comparative estimates of the amount of underreporting in the two studies are very rough approximations because it is not pos- sible from available data to restrict the HIP study group to those who had received only services under that insurance plan. However, gross figures indicate that approximately 56 percent of the con- ditions defined as unqualifiedly chronic inthe HIP records were not reported in interviews, while a comparable estimate in the SRI study was 47 per- cent, Among disease categories for which dataare available, comparatively low rates of underre- porting were found in both studies for asthma, hay fever, diabetes, heart conditions, bronchitis, ulcer of the stomach and duodenum, and diseases of the gallbladder. Those conditions which were grossly underreported in both studies include be- nign and unspecified neoplasms, anemia and other blood disorders, mental illness, respiratory dis- eases other than bronchitis and tuberculosis, skin diseases, and menopausal and other genito- urinary disorders, Findings from both studies in- dicate that underreporting occurred less fre- quently among older persons than among children and young adults, and also among those with 10 or more physician visits during the study year than for those who had seen a physician less frequently. SUMMARY A study designed to measure the accuracy and completeness of the reporting of chronic condi- tions in health interviews was carried out by the Stanford Research Institute during the early years of the National Health Survey. The sample popu- lation was selected from members of the Kaiser Foundation Health Plan, a large prepayment medi- cal plan providing medical services through the Southern California Permanente Medical Group and hospitalization through the Kaiser Plan. Medical records were compared with interview responses from persons for whom the records were maintained, The study design provided for the creation of medical records specifically for this study in order that the project could be con- ducted on a prospective basis. Interviews were conducted following the completion of physician records maintained on sample persons during a 12-month period. This report has presented the findings in the phase of the study dealing with the comparative amounts of underreporting and overreporting of chronic conditions in health interviews. The fol- lowing statements summarize the principal find- ings of the study. The total number of chronic conditions re- corded in the medical records and/or reportedin interviews amounted to 15,417, For this phase of the study the following categories of conditions were excluded: 4,499 conditions that had not been treated exclusively in SCPMG facilities, 3,633 conditions that were reported in interviews as having been treated in SCPMG facilities prior to the 12-month period covered by the study, and 103 conditions for which SCMPG utilization status was not available. These exclusions reduced the group to 7,182 chronic conditions, Of this number, 2,811 conditions were recorded in the medical records and also reported inthe interviews, 2,468 conditions were recorded in the records but not reported in interviews, and 1,903 conditions were reported in interviews but not recorded in the medical records. Reporting indexes derived from these figures (shown in table 5) indicate that the estimate of underreporting in interviews was 46.8 percent and the estimate of overreporting of conditions was 40.4 percent, When conditions were classified into 50 broad disease categories, it was found that certainkinds of conditions with comparatively low indexes of both underreporting and overreporting might be expected to be reported in an interview with a fair degree of accuracy and completeness. In- cluded were diabetes, vascular lesions ofthe cen- tral nervous system, heart conditions, diseases of the gallbladder, and absence of fingers and toes. \ \\ High indexes of underreporting with rather low proportions of overreporting were noted for such conditions as benign and unspecified neo- plasms, mental illness of specified type, menstrual disorders, and skin diseases. These resultswere not unexpected in the reporting of conditions which might cause embarrassment or reluctance on the part of the respondent. High indexes of overreporting with a lower degree of underreporting were found in the re- porting of hay fever, asthma, tuberculosis, head- ache and migraine, hypertension, hemorrhoids, rheumatic fever, sinusitis, bronchitis, visualim- pairments, hearing impariments, and speech de- fects. It is quite possible that respondents were : reporting conditions of long duration or even conditions they had many years ago which were not noted in current medical records. i A high proportion of the conditions that were underreported or overreported consisted of those for which a single physician visit or contact was made during the study year. The sharp increase in the accuracy of reporting as the number of physician visits increased indicates that oppor - tunity for communication with his physician im- proves the ability of a respondent to report his conditions in an interview. Regular medication for a condition also increases the probability that it will be reported in an interview, - The percentage of underreporting of condi- | tions was about the same for men and women; | however, women have a greater tendency toover- | . ! report their conditions. For persons 65 years and older, the accuracy and completeness of reporting | were substantially higher among women than among men. Education did not influence the amount of underreporting and overreporting in this study to any appreciable degree, —— 00 O0—mm Table 1. 10. 11. 12. 13. 14. LIST OF DETAILED TABLES Number and percent distribution of chronic conditions and persons with chronic conditions and number of chronic conditions per person for persons having at least one condition by utilization of medical services as reported in medical records and interviewS=-=----cccccccccccccc ccc cece emcee cece cece cme — em = Number and percent distribution of chronic conditions by utilization of medical services as reported in medical records and interviews, according to type of Number and percent of persons using SCPMG services only by number of underreported (match code C) and overreported (match code D12) chronic conditiong=---=---==-== Number of chronic conditions for persons using SCPMG services only, by type of Percent of chronic conditions underreported and overreported in interviews for persons using SCPMG services only, by type of match-r-=---cccccccccaccccnnnaaana Number and percent distribution of chronic conditions recorded in medical records for persons using SCPMG services only by number of physician visits for the con- ditions, according to type of match==----eccccececcccccccccccccccccccceccccannaa-" Number and percent distribution of chronic conditions reported in interviews for persons using SCPMG services only by number of physician contacts for the con- ditions, according to type of match=-=-----eccccccccccccccccnceecc cece ccnccccaa- Number and percent distribution of chronic conditions reported in medical records for persons using SCPMG services only by type of match, according to whether or not medication was prescribed==--=--m-cmomccm cece cceeeeeememeemeee mene Number and percent distribution of chronic conditions reported in interviews for persons using SCPMG services only by type of match, according to whether or not. medication was prescribed-=-----ccccmmmcccccce cece ce cece cme mene Number and percent distribution of chronic conditions reported in medical records and in interviews for persons using SCPMG services only by sex and type of match- Number and percent distribution of chronic conditions reported in medical records for persons using SCPMG services only by age and type of match,according to sex-- Number and percent distribution of chronic conditions reported in interviews for persons using SCPMG services only by age and type of match, according to sex---- Number and percent distribution of chronic conditions reported in medical records for persons using SCPMG services only by education of respondent and typeof match- Number and percent distribution of chronic conditions reported in interviews for persons using SCPMG services only by education of respondent and type of match-- Page 11 12 13 14 16 18 19 20 20 21 22 23 24 25 Table 1. Number and percent distribution of chronic conditions and persons with chronic conditions and number of chronic one condition by utilization of medical services as reported in medical records and conditions per person for persons having at least interviews Utilization as reported in: Chronic All con- || All condi= | A11 con- | ALL ditions || persons! ditions | persons! Medical Interview per records person Percent Number distribution SCPMG only SCPMG only 10,322 3,156 3.27 67.0 62.8 SCPMG and others SCPMG and others 2,350 594 3.96 15.2 11.8 Non-SCPMG only Non-SCPMG only 825 361 2,29 5.4 7.2 No utilization No utilization 602 334 1.80 3.9 6.6 SCPMG only .No utilization 596 245 2.43 3.9 4.9 SCPMG and others Non-SCPMG only 172 67 2.57 1.1 1.3 Non-SCPMG only SCPMG and others 80 50 1.60 0.5 1.0 No utilization SCPMG only 470 220 2.14 3.0 4.4 Total -==c-eccccccccccncncnnann 15,417 5,027 “ne 100.0 100.0 1Excludes persons with no reported conditions either on the PVRS or on the question- naire. n Table 2. Number and percent distribution of chronic conditions by utilization of med- ical services as reported in medical records and interviews, according to type of match Utilization as reported in: T Type of match ool Reported in CLORE interview only . A B C Medical : Interview records p12! D+ Number of conditions SCPMG only SCPMG only 10,322 1,902 855 2,323 1,996 3,246 SCPMG and others SCPMG and others 2,350 344 161 285 819 741 Non-SCPMG only Non-SCPMG only 825 xe 463 362 No utilization No utilization 602 . wes eo 20 582 SCPMG only No utilization 596 28 26 145 10 387 SCPMG and others Non-SCPMG only 172 32 11 28 48 53 Non-SCPMG only SCPMG and others 80 ; 10 70 No utilization SCPMG only 470 : . 250 220 Totale=emcmc mmm cmc e 15,417 2,306 1,053 2,781 3,616 5,661 Percent distribution of conditions SCPMG only SCPMG only 67.0 82.5 81.2 83.5 55.2 57.3 SCPMG and others SCPMG and others 15.2 14.9 15.3 10.2 22.6 13.1 Non-SCPMG only Non-SCPMG only 5.4 “hw 12.8 6.4 No utilization No utilization 3.9 0.6 10.3 SCPMG only No utilization 3.9 1.2 2.5 5.2 0.3 6.8 SCPMG and others Non-SCPMG only 1.1 1.4 1.0 1.0 1.3 0.9 Non-SCPMG only SCPMG and others 0.5 ‘e . 0.3 1.2 No utilization SCPMG only 3.0 . 6.9 3.9 Total=-=meeem cece eee 100.0 100.0 100.0 100.0 100.0 100.0 Excludes conditions for which information on medical attention past 12 months was not available. NOTE: Definition of type of match: conditions reported on PVRS and in interview which matched. conditions reported on PVRS and in interview which appeared to be associated. conditions recorded in PVRS only. D12 = conditions reported QW > nnn D+ = conditions reported in interview only about which res in interview only received during the about which respondent said he had con- tacted a physician during the preceding 12 months. tacted a physician during the preceding 12 months. pondent said he had not con- Table 3. Number and percent of persons using SCPMG services only by number of under- reported (match code C) and overreported (match code D12) chronic conditions Number of conditions recorded on PVRS only (C match) Number of D12 conditions! Total 0 1 2 3 4 Number of persons All conditions--=--c=ceceeca= 3,401 1,750 1,057 428 117 41 8 O--mmeemceccccccccccc ccm 2,123 998 755 263 68 35 4 I tt EE EEL 873 511 194 124 34 6 4 RE ttt 256 145 68 32 11 - - Jem em cmc mmc cme ccc emcee ee 96 61 26 7 2 - - ttt 36 23 13 - - - - Jeceeccec ccm me creme em emcee —————— 10 7 1 2 - - - fremmmmmmmmemmem cece c cece cee cea 1 1 - - - - - Jmmmmemmccccccc ccc ccc mcm —— 5 4 - - 1 - - Beceem mcm ccc cme ee 1 - - - 1 - - Percent of persons All conditiongs--=-ceccccccnnux 100.0 51.5 31.1 12.6 3.4 1.2 0.2 Qrmemecccc cece ccc ccc mc ccc 62.4 © 29.3 22.2 7.7 2.0 1.0 0.1 locememccccccccce eee ————ememeeee—- 25.7 15.0 547 3.6 1.0 0.2 0.1 2mmmmemcemccccc cee c cme 2:5 4.3 2.0 0.9 0.3 - - Jem mmm cmc crm cc ccm m—————— 2.8 1.8 0.8 0.2 0.1 - - LR LLL EEE LEE EEE EEE EEE EEE 1.1 0.7 0.4 - - - Semmceeeae Femmes ces cccc cece ——— 0.3 0.2 0.0 0.1 - - - femme mmcmccm ccc cccem cece eee 0.0 0.0 - - - - - Jemmmmmemce ccc ccc cee ce 0.1 0.1 - - 0.0 - - Bemmmmmmmcccm cm ccccc cece meee eee 0.0 - - - 0.0 - - Excludes conditions for which information on medical attention received during the past 12 months was not available. NOTE: Definition of type of match: A = conditions reported on PVRS and in interview which matched. B = conditions reported on PVRS and in interview which appeared to be associated. C = conditions recorded in PVRS only. D12 = conditions reported in interview only about which respondent said he had con- tacted a physician during the preceding 12 months. 13 Table 4. Number of chronic conditions for persons using SCPMG services only, by type of match Type of match Reporced Reported Chronic condition and recode number! Reported Reported ia Deen in BPoreey medical in apse. records Beales) terviews records A+B+D12) and in- Be 8 only? (A+B+C) ( terviews 9a y (p12) (A+B) (©) Number of conditions EH ITII———————— 5,279 4,714. 2,811] 2,468 | 1,903 01 Tuberculosis (active) (inactive), all SiteS=eememeccccccccc ccc cccc ccc ccna canaa 1 6 1 - 5 02 Other chronic infective and parasitic diseases====ccccccccccccccc ccc 86 62 38 48 24 03 Malignant neoplasmsS-======ccccccccaccacanaa 49 57 30 19 27 04 Benign and unspecified neoplasms=========-- 299 164 130 169 34 05 Hay fever, without asthma-=--===-ccecccaa-- 164 228 120 44 108 06 Asthma (with or without hay fever) } (bronchial) (not otherwise specified) ----- 39 55 27 12 28 07 Other allergic disorders, not elsewhere classifiable=====ccccccmmccccancccaccanaa- 117 114 64 53 50 08 Diseases of the thyroid gland-=---===eeea-- 61 65 39 22 26 09 Diabetes (mellitus)=s=---cecccccccccccacaa- 88 72 71 17 1 10 Anemia and other diseases of the blood and blood-forming organs, 3 mo.,t-======--- 40 45 15 25 30 11 Vascular lesions of the central nervous SySteMe==eeccccccccccaamcccccacccccncenaa - 28 30 24 4 6 12 Headache and migraine, chronic----==e-caa-a- 90 119 56 34 63 13 Specified mental disorders, not elsewhere classifiable~==ccccccccccccccccncccccaaaa. 381 180 152 229 28 14 TIll-defined mental and nervous trouble, not elsewhere classifiable, 3 mo.,+======-- 89 86 38 51 48 15 Diseases of the heart, not elsewhere classifiable (chronic rheumatic) (arte- riosclerotic) (hypertensive) -====cececacaa- 238 245 189 49 56 16 Hypertension, not elsewhere classifiable, without heart involvement-=---=-cceceaaaa- 227 285 184 43 101 17 Varicose veins-==-ceccccccacccccccccccaaaa- 81 82 39 42 43 18 Hemorrhoidg=====--ecccccccaccanccacccacaaaaa 131 192 87 44 105 19 Rheumatic fever; arteriosclerosis, not elsewhere classifiable; other chronic diseases of the circulatory system-==-==--=-- 33 48 13 20 35 20 Chronic sinusitiSe===--cccccccccccccccaaaa- 19 91 19 - 72 21 Chronic bronchitige==-ecccccccccccccccccaaa 24 61 19 5 42 22 Other chronic diseases of the respiratory SySteMe==m=eeeccmccccaccccnccacccnnccnnaaan 151 128 66 85 62 23 Ulcer of stomach and duodenum---==-ececceea- 111 112 67 44 45 24 Hernia (abdominal cavity) =-===-=eeeccaceaca-- 78 67 40 38 27 25 Diseases of the gallbladder, chronic==-===-- 27 34 23 4 11 26 Other chronic diseases of the digestive SySteMemmeemmecemcceccec ccc cccc ccc cccaa- 267 198 130 137 68 See notes at end of table. Table 4. Number of chronic conditions for persons using SCPMG services only,by type of match—Con. Type of match Reported Reported Chronic conditions and recode number?! Repoteed Reported ia bok in Repersed medical in iaege- records medical terviews records A+B+D12) and in- Soy only? (A+B+C) ( terviews ji y (p12) (+B) (©) Number of conditions 27 Disorders of menstruation-------=----e---- 170 102 86 84 16 28 Menopausal symptoms, except psychosis----- 98 47 21 77 26 29 Urinary calculi; prostate disorders; other chronic genitourinary conditions--------- 384 211 131 253 80 30 Chronic skin diseases-=====c==cceccccac--- 429 195 148 281 47 31 Arthritis and chronic rheumatism---------- 178 238 122 56 116 32 Other chronic musculoskeletal disorders--- 175 107 75 100 32 33 Fractures, 3 mo.+, no residual specified-- 6 16 4 2 12 34 Other injuries, 3 mo.+, no residual specifiedm-==ccecccenccanaa" mmmemmmemm—————— 9 15. 2 7 13 35 Severe visual impairment---=-=------------ 2 8 2 - 6 36 Other visual impairment--=-=====-==-ccc---- 80 95 57 23 38 37 Hearing impairments 50 103 36 14 67 38 Speech defects====-=--- 4 7 4 - 3 39 Paralysig----------e-cmecocmcoccoconooonooo 32 36 20 12 16 40 Absence, fingers, toes, only---==-=----=-==== 4 4 4 - - 41 Absence, major extremitieg---------c-c---- - - - - - 42 Impairments (except paralysis and ab- sence), back or spine-=--==----c--c-cc-oo-- 124 138 75 49 63 43 Impairments (except paralysis and ab- sence) , upper extremities and shoulders-- 15 19 7 8 12 44 Impairments (except paralysis and ab- sence), lower extremities and hips with any other site=--==-=-==-cececc-ceooccoc=x 64 108 30 34 78 45 Impairments (except paralysis and ab- sence), multiple not elsewhere classifi- able, and ill-defined, limbs, back, trunk- 43 33 26 17 7 46 Other impairments§==-=-=-==--cc-cccccccc=u= 5 13 4 1 9 47 Other chronic conditions, not impairments and not in recodes 48-50=====-=c-c-ce-oo- 83 54 36 47 18 48 Chronic diseases of eye, not impairments-- 224 192 118 106 74 49 Chronic diseases of ear, not impairments-- 98 73 61 37 12 50 Chronic organic nervous system conditions- 83 74 61 22 13 IThe recode categories 1-46 are the same as those used in the Recode 3 for the Health Inter- view Survey. Recodes 48-50 were included in Recode 47 in the original recode. 2Excludes conditions for which information on medical attention received during the past 12 months was not available. NOTE: Definition of type of match: A = conditions reported on PVRS and in interview which matched. B = conditions reported on PVRS and in interview which appeared to be associated. C = conditions recorded in PVRS only. D12 = conditions reported in interview only about which respondent said he had contacted a physi- cian during the preceding 12 months. 15 Table 5. Percent of chronic conditions underreported and overreported in interviews for persons using SCPMG services only, by type of match Type of match 1 Repoveed Reported Chronic condition and recode number in inter- medical records Views only only _C : | (+7570) A+B+D12 Percent Percent under- over=- reported | reported All chronic conditions-=e=eeecccccccacmcmcccccacccccccacnan 46.8 40.4 Ol Tuberculosis (active) (inactive), all siteS-em=emccccccamaa-- - 83.3 02 Other chronic infective and parasitic diseaseS-ceeeeceeeeee-- 55.8 38.7 03 Malignant neoplasmg==e=erececccescnmmnunnnarennnccncncnnnnannn 38.8 47.4 04 Benign and unspecified neoplasms=-=-=-eeecececcccceccacacacaan 56.5 20.7 03 Hay fever, without asthmaresesrscammrncmsmnccanmncnmnnnmnmnem 26.8 47.4 06 Asthma (with or without hay fever) (bronchial) (not other- wise specified) =cemcecmcacaccccccccacccncececececccccaee 30.8 50.9 07 Other allergic disorders, not elsewhere classifiable-e==eem== 45.3 43.9 08 Diseases of the thyroid gland=e==eeeceeceeecceeccccaceccncane- 36.1 40.0 “09 Diabetes telat) Sees esssmscsesccacacmceeecessececcececmae—— 19.3 1.4 10 Anemia and other diseases of the blood and blood-forming Organs, 3 mo,t====eeecceccccacccccccmccccccccccccaccnae———-= 62.5 66.7 11 Vascular lesions of the central nervous systeme=e=eee===e-= ——- 14.3 20.0 12 Headache and migraine, chronice=e=e= Seesecccccccaccemanacea-- 37.8 52.9 13 Specified mental disorders, not elsewhere classifiable-eee==n= 60.1 15.6 14 1Ill-defined mental and nervous trouble, not elsewhere classi- fiable, 3 MO.++=eemcmcecccccccecccacacacacacmcecdcccccmaa—- 57.3 55.8 15 Diseases of the heart, not elsewhere classifiable (chronic rheumatic) (arteriosclerotic) (hypertensive)eemememme-eececco-= 20.6 22.9 16 Hypertension, not elsewhere classifiable, without heart involvement =m eee acco cece 18.9 35.4 17 Varicose veins===eececccmcccccccccommccmcecccccccccccmcana- 52.9 52.4 18 Hemorrhoids==ee-mmme ccm mcccccccccccccccmce——- 33.6 54.7 19 Rheumatic fever; arteriosclerosis, not elsewhere classifi- able; other chronic diseases of the circulatory system==----- 60.6 72.9 20 Chronic SinuSitiS==emeeceeccccacmcmecececcccmeiocceeenn - 79:1 21 Chronic DronchiCigs w= scosme sewmmmuneensms smn. 20.8 68.9 22 Other chronic diseases of the respiratory system-e=--e-ceece-- 56.3 48.4 23 Ulcer of stomach and duodenum-=e==eeeme-eceecoccccccoccmncaenn 39.6 40.2 24 Hernia (abdominal Cavity) memecmmc mmm cee ceaeeee 48,7 40.3 25 Diseases of the gallbladder, ChIONIC~wwwewwscunssemsnn wm 14.8 32.4 26 Other chronic diseases of the digestive System==--cceccccacaax 51.3 34.3 27 Disorders of menstruation=ressceeescmsmemsceeeccccacanaenmn——— . 49.4 15.7 28 Menopausal symptoms, except pSychoSiS==eesececccccecemmceeeean 78.6 55.3 16 See notes at end of table. Table 5. Percent of chronic conditions underreported and overreported in interviews for persons using SCPMG services only, by type of match—Con., Type of match Reported R Co in Reported Chronic condition and recode number! medical in inter- views records ive TE (sBsor 3 A+B+C ) Percent Percent under- over- reported | reported 29 Urinary calculi; prostate disorders; other chronic genito- urinary conditions-=---==e-eememceoceccccmccommmomomomooon=s 65.9 37.9 30 Chronic skin diseases-=-==e-cccccmecemmcnoccosmocnnm moomoo 65.5 24.1. 31 Arthritis and chronic rheumatism-=-===-=--ee-eccceccceenccccnn- 31.5 48.7 32 Other chronic musculoskeletal disorders-==--------c-ce-=---=- 57.1 29.9 33 Fractures, 3 mo.+, no residual specified--------e-eccc-co-n-- 33.3 75.0 34 Other injuries, 3 mo.+, no residual specified--------=c===--- 77.8 86.7 35 Severe visual impairmentes-scecsccmcccccccccccccoommononooo-- - 75.0 36 Other visual impairmente--e-cececceccccceccccccccccccoonmnmmon- 28.8 40.0 37 Hearing impairments=------sseeeceee-ccccoeoccocseosommooonooos= 28.0 65.0 38 Speech defectS======e-ececmmemcmcccoccmcocnmommoomomnos oon - 42.9 ParalysiS-e=====-ececcccccesccmmeceocococsesosmmmomnooononon- 37.5 44.4 40 Absence, fingers, toes, only=====-=e-ececccccemcomoococoooo== - - 41 Absence, major extremities===-s===eec-ceccccccccmmcmmnnonmoomo- - - 42 Impairments (except paralysis and absence), back or spine---- 39.5 45.7 43 Impairments (except paralysis and absence), upper extremities and shoulders=e==--= mmm mmmmmeesecce=- cmmecmeemescem———— 53.3 63.2 44 Impairments (except paralysis and absence), lower extremities and hips with any other site==e====eeecc-occcccccccmoooonnco= 53.1 72.2 45 Impairments (except paralysis and absence), multiple not elsewhere classifiable, and ill-defined, limbs, back, trunk- 39.5 2142 46 Other impairmentSes==-==ececcccccecm-eecmecccceceonoonmmmnn==- 20.0 69.2 47 Other chronic conditions, not impairments and not in recodes 48-50-mceccecececememcceem-e-esceses-e-e-eess-s--ssesses==es 56.6 33.3 48 Chronic diseases of eye, not impairments=---=----e---ce=c=c---- 47.3 38.5 49 Chronic diseases of ear, not impairments====-==--=-=-cecce-=-- 37.8 16.4 50 Chronic organic nervous system conditions===--=-ese=---ece-c--= 26.5 17.6 1The recode categories 1-46 are the same as those used in the Recode 3 for the Healeh Interview Survey. Recodes 48-50 were included in Recode 47 in the original recode. 2Excludes conditions for which information on medical attention received during the past 12 months was not available. NOTE: Definition of type of match: "conditions reported on PVRS and in interview which matched. conditions reported on PVRS and in interview which appeared to be associated. conditions recorded in PVRS only. D12 = conditions reported in interview only about which respondent said he had con- tacted a physician during the preceding 12 months. ow» wun 17 Table 6. to type of match Number and percent distribution of chronic conditions recorded in medical records for persons using SCPMG services only by number of physician visits for the conditions, according Type of match All Number of physician visits conditions A B C Number of conditions 5,279 1,930 881 2,468 2,616 630 448 1,538 1,133 438 158 537 554 249 109 196 331 177 76 78 226 139 28 59 342 230 56 56 54 48 4 2 18 14 2 2 5 5 - - Percent distribution of conditions 100.0 100.0 100.0 100.0 49.6 32.6 50.9 62.3 21.5 22.7 17.9 21.8 10.5 12.9 12.4 7.9 6.3 9.2 8.6 3.2 4,3 7.2 3.2 2.4 6.5 11.9 6.4 2¢3 1.0 2.5 0.5 0.1 0.3 0.7 0.2 0.1 26 Or mMOres====-cecccecccecaa= Meesemccccccccccanaa “0.1 0.3 - - NOTE: Definition of type of match: A = conditions reported on PVRS and in interview which matched. B = conditions reported on PVRS and in interview which appeared to be associated. C = conditions recorded in PVRS only. D12 = conditions reported in interview only about which respondent said he had contacted a physi- cian during the preceding 12 months. Table 7. Number and percent distribution of chronic conditions reported in interviews for persons using SCPMG services only by number of physician contacts for the conditions, according to type of match Type of match Number of physician contacts! .... 2 A B D12 Number of conditions All contactS-==--=ceccmmeccccccccccanecan- 4,535 1,831 801 1,903 0 memcemmmmmmmmememmemmeececeeeeeeeeese———————— 366 220 146 - lememececcccc cece cece mmm meen remem ccc ——— 1,592 376 219 997 718 284 106 328 440 215 78 147 420 185 67 168 147 75 15 57 457 264 87 106 249 127 53 69 82 52 15 15 64 33 15 16 Percent distribution of conditions All contacts======ceeecccccemmmcceecccconn= 100.0 100.0 100.0 100.0 8.1 12.0 18.2 - 35.1 20.5 27.3 52.4 15.8 15.5 13.2 17.2 9,7 11.7 9.7 7.7 9.3 10.1 8.4 8.8 i 4.1 1.9 3.0 10.1 14.4 10.9 5.6 5.5 6.9 6.6 3.6 16-25-==c==emmemmmmemmcmmeccemmeemeeseceeoeeososooe- 1.8 2.8 1.9 0.8 26 Or MOYE===m==mme-ecece-eeemecececoe-c-s-cc====- 1.4 1.8 1.9 0.8 1A physician contact occurred if a physician was seen or spoken to about the condition. Condi- tions for which the respondent did not know (or could not estimate) the number of physician con- tacts are excluded. 2Fxcludes conditions for which information on medical attention received during the past 12 months was not available. 3Includes conditions for which the respondent reported no physician contacts in the interview but for which it was known from entries on the PVRS that there had been at least one contact dur- ing the study year. NOTE: Definition of type of match: A = conditions reported on PVRS and in interview which matched. B= conditions reported on PVRS and in interview which appeared to be associated. t = conditions recorded in PVRS only. D12 = Conditions reported in interview only about which respondent said he had contacted a physi- cian during the preceding 12 months. 19 Table 8. Number and percent distribution of chronic conditions reported in medical records for persons using SCPMG services only by type of match, according to whether or not medication was prescribed Medication status in medical records m Type of match conditions A B C Number of conditions Tota lmwwmn 100 0 i i 5,279 1,930 881 2,468 Medication=memmmmm cence eee 2,121 979 437 705 No medicationmm—smmeemmeesnmeee meee mncnn eee 3,158 951 44 1,763 Percent distribution of conditions TORR Lawman cr ee ion mmm ih mmm mmm sm A 100.0 36.6 16.7 46.8 Medications mmm on om cee eee 100.0 46.2 20.6 33.2 No medication=-=-semme meme eee 100.0 30.1 14.1 55.8 NOTE: Definition of type of match: A = conditions reported on PVRS and in interview which matched. B = conditions reported on PVRS and in interview which appeared to be associated. C = conditions recorded in PVRS only. D12 = conditions reported in interview only about which respondent said he had contacted a physi- cian during the preceding 12 months. Table 9, Number and percent distribution of chronic conditions reported in interviews for persons using SCPMG services only by type of match, according to whether or not medication was pre- scribed Type of match All Medication status reported in interview condi tions : A B D12 Number of conditions? Totaleeeeaccmacaccc acc cc cece cece eae 4,780 1,909 865 2,006 Medicationeeemceeccerecccaccceaccccmce ceca 2,254 966 426 862 No medication-eeeecccccceccecceccocccccacaanannas 2,526 943 439 1,144 Percent distribution of conditions SITE Lom EE A 0 0 i ei 100.0 39.9 18.1 42.0 Medicationeemmmemmmemeccceecemaccmcanmconanaannn 100.0 42.9 18.9 38.2 NO Tedicatliimm mma mune nes ewes vemos memes 100.0 37.3 17.4 45.3 Excludes conditions for which information on medical attention received during the past 12 months was not available. Excludes conditions for which the medication status could not be obtained from the inter- view data. NOTE: Definition of type of match: A = conditions reported on PVRS and in interview which matched. B = conditions reported on PVRS and in interview which appeared to be associated. C = conditions recorded in PVRS only. } D12 = conditions reported in interview only about which respondent said he had contacted a physi- cian during the preceding 12 months. 20 Table 10. Number and percent distribution of chronic conditions reported in medical records and in interviews for persons using SCPMG services only by sex and type of match Conditions in medical records Conditions in interviews Sex All Type of match All Type of match condi- condi- tions A B Cc tions A B p12! Number of conditions Both sexes====cemcececcccacanx 5,279 1,930 881 | 2,468 4,81711,930 881 | 2,006 Male-=-=-=ccccmmccccceccnaaa= m—————- 2,168 813 363 992 1,873 813 363 697 Female-=--=cecmmecncccncacana" F. - 3,111 1,117 518 | 1,476 2,944111,117 518 | 1,309 Percent distribution of conditions by sex Both sexes===e=ememecccccccaax 100.0 100.0 | 100.0 | 100.0 100,0/| 100.0 | 100.0 | 100.0 Male--=--=c-mcmmmcmmemmcccnn cence 41.1 42.1 | 41.2 | 40.2 38.9|| 42.1 | 41.2 | 34.7 Female-==-=emcccmcecencenccccecnaan~" 58.9 57.9 | 58.8 59.8 61.1|| 57.9| 58.8 | 65.3 Percent distribution of conditions by type of match Both sexes=-==eememeccccccanax 100.0 36.6 16.7 | 46.8 100.0|| 40.1 | 18.3 | 41.6 Male-===m=ceccecccccecccm nnn 100.0 37.5 16.7 | 45.8 100.0 || 43.4) 19.4 | 37.2 Female--===-=-cecccmmccemmenceeneacax 100.0 35.9 16.7 | 47.4 100.0 |} 37.9 17.6 | 44.5 - lgxcludes conditions for which information on medical attention received during the past 12 months was not available. NOTE: Definition of type of match: conditions reported on PVRS and in intervi Ow >> nnn conditions reported in PVRS only. D12 = conditions reported in interview only about which respondent sa cian during the preceding 12 months. ew which matched. conditions reported on PVRS and in interview which appeared to be associated. id he had contacted a physi=- 21 Table 11. Number and percent distribution of chronic conditions reported in medical persons using SCPMG services only by age and type of match, according to sex records for Male - Female Age All “Type of match All Type of match condi- condi - tions A | B Cc tions A B Cc Number of conditions 17 years and over--=-=-ecceaa- 2,168 813 363 992 3,111 1,117 518 1,476 17-24 years 88 40 9 39 167 58 18 91 25-34 years 166 51 43 72 398 155 50 193 35-44 years 415 136 75 204 820 277 131 412 45-54 years 622 260 107 255 765 260 117 388 55-64 years 426 138 61 227 608 193 125 290 65 years and over------cc-ccceacaaa- 451 188 68 195 353 174 77 102 Percent distribution of conditions by age 17 years and over----=--eceeeca- 100.0 || 100.0 | 100.0 | 100.0 | 100.0 | 100.0 | 100.0 ; 100.0 17-24 years==-=-ececoccccccccencaaa- 4.1 4.9 2.5 3.9 5.4 5.2 3.5 6.2 25-34 years----=ececmcecccmceccacao- 7.7 6.3 11.8 7.3 12.8 13.9 9.7 13:1 35-44 years----- 19.1 16.7 | 20.7 | 20.6 26.4 24,8 | 25.3 27.9 45-54 years-- 28.7 32.0 29.5 25.7 24,6 23.3 1 22.6 26.3 55-64 years==---c-ccmcccccmcecce—aan 19.6 17.0 | 16.8 | 22.9 19.5 17.3] 24.1 19,6 65 years and over--=-eeececmccccanaa- 20.8 23.1 18.7 19.7 11.3 15.6 14.9 6.9 Percent distribution of conditions by type of match 17 years and over-----eeececa-- 100.0 37.5 | 16.7 | 45.8 | 100.0 35.9| 16,7 47.4 17-24 years=---eccceccmmmcccccccaaas 45.5 10,2 | 44.3 100.0 34.7 | 10.8 54.5 25-34 years==---cecccmcmccceeeeeeen 30.7 25.9 | 43.4 100.0 38.91 12.6 48.5 35-44 years---=-cccmcccmcccccccaeaan 32,81 18.1 | 49.2 100.0 33.8 16.0 50.2 45-54 years-- 41.8 | 17.2 | 41,0 | 100.0 34.0 | 15.3 50.7 55-64 yearse=---ccmcoccccmccceceaea- 32.4 | 14.3 |/53.3 100.0 31.7 | 20.6 47 7 65 years and over-------ecececccocao- 100.0 41.7 15.1 |\ 43.2 100.0 49.3 | 21.8 28.9! NOTE: Definition of type of match: A = conditions reported on PVRS and in B = conditions reported on PVRS and in C = conditions recorded in PVRS only. D12 = conditions reported in interview cian during the preceding 12 months. 22 interview which matched. interview which appeared to be associated. only about which respondent said he had contacted a physi- Table 12. Number and percent distribution of chronic conditions reported in interviews for per- sons using SCPMG services only by age and type of match, according to sex Aen 7 i Co Male Female Age . Zo , ) All Type of match All Type of match —— “on condi- condi- - ‘ tions A B p12! tions A 2 p12 Number of conditions 17 years and over=----==---=-=-= 1,873 813 363 697| 2,944 || 1,117 518 | 1,309 17-24 years==========-c--=-c-c--o--- 116 40 9 67 183 58 18 107 25-34 years-- : 177 51 43 83 388 155 50 183 35-44 years----=--=-=-========- 302 136 75 91 781 277 131 373 45-54 years----=-=------eece-ce-ac-- 573 260 107 206 692 260 117 315 55-64 years----=-mm-m=m=---e-cccecee-- 291 138 61 92 516 193 125 198 65 years and over-------=---=-=-==-- 414 188 68 158 384 174 77 133 Percent distribution of conditions by age 17 years and over-----=------- 100.0 || 100,0| 00.0] 100.0| 100.0 | 100.0 | 100.0 | 100.0 17-24 years------=-=c-c-ccmmmammnon- 6.2 4.9 2.5 9.6 6.2 5.2 3.5 8.2 25-34 years-----=--c-c-mo--ocm-aooo- 9.5 6.3] 11.8 11.9 13.2 13.9 9.7 | 14.0 35-44 years----=--=---m-meme-ocoo-o- 16.1 16.7 | 20.7| 13.1 26.5 24.8 | 25.3 | 28.5 45-54 year§---e-m-a-mcemccememanoon= 30.6 32.0| 29.5] 29.6 23.5 23,3 | 22.6 | 24.1 55-64 yearS---------c-=c--e-camcaaa- 15.5 17.0] 16.8] 13.2 17.5 17.3 | 24.1 | 15.1 65 years and over--------=-=-------- 22.1 23.1} 18.72; 22.7 13.0 15.6 | 14.9 10.2 Percent distribution of conditions by type of match 17 years and over------------- 100.0 43.4| 19.4] 37.2 100.0 37.9 17.6 44.5 17-24 years 100.0 34.5 7.8 57.8 100.0 31.7 9.8 58.5 25-34 years 100.0 28.8 24,3] 46.9 100.0 39.9 12.9 47.2 35-44 years 100.0 45.0| 24.8) 30.1] 100.0 35,5 | 16.8 | 47.8 45-54 years 100.0 45.4 18.7 6.0 100.0 37.6 | 16.9 | 45.5 55-64 years 100.0 47.41 21.0| 31.6f 100.0 37.4 | 24.2 | 38.4 65 years and over------------------- 100.0 45,4 | 16.4 38.2 100.0 45.3 | 20.1 | 34.6 lgExcludes conditions for which information on medical attention received during the past 12 months was not available. NOTE: Definition of type of match: A = conditions reported on PVRS and in interview which matched. B = conditions reported on PVRS and in interview which appeared to be associated. C = conditions recorded in PVRS only. D12 = conditions reported in interview only about which respondent said he had contacted a physi- cian during the preceding 12 months. 23 Table 13. Number and percent distribution of chronic conditions reported in medical records for persons using SCPMG services only by education of respondent and type of match All Type of match Education of respondent condi- tions A B c Number of conditions All educational groupS---=---ececcaaa-o- 5,254 1,921 880 2,453 No education====e==ccececmmmccccccccccacaeao 72 25 24 23 1-4 yearsS======cmcccccmmccccccceccmemaeao- 84 26 15 43 5-8 yearse=eme=-cmemccccmcmmcccccccceeaeas 922 353 166 403 9-12 years=-==--=-cccmmcmmmmmcmmecceeeeeeeeo 2,684 995 443 1,246 13-14 years=====cmcccmcc mm ccceccceeeeeo 708 273 108 327 15-16 years===-eemccccccmmmcccccccccccaceaaao 468 166 63 239 17 years or more-------c-ccemeccccccecccccaaao- 316 83 61 172 Percent distribution of conditions by education All educational groups----=-=ce-cecccaa- 100.0 100.0 100.0 100.0 No education===-=-c-ccccmcccmccccccacaaaaaoo 1.4 1.3 2.7 0.9 1-4 years=-=-ccccccmmmmccmmcmcceeeeeeea eo 1.6 1.4 1.7 1.8 5-8 yearse===--cecccmcmccmcccccmcccceeccaaa 17.5 18.4 18.9 16.4 9-12 yearsS=====---ccmcmcmmccceceeeceooooas 51.1 51.8 50.3 50.8 13-14 years=====ecceccccccacm mcm 13.5 14,2 12.3 13.3 15-16 years====-=ccecccmmm mmm 8.9 8.6 7.2 9.7 17 years or more----=-ececccccccmccnccaaaaaaa 6.0 4.3 6.9 7.0 Percent distribution of conditions by type of match All educational groups---=-eececaccaca- 100.0 36.6 16.7 46,7 No education====emcccc coca ccccccccemeeao 100.0 34,7 33.3 31.9 1-4 years==emececccmcm cme 100.0 31.0 17.9 51.2 3-8 years=eemceemeccccccccccmcmeeeeeeeeaao 100.0 38.3 18.0 43,7 9-12 years=====smcceccccmcmmcmeemcecoeeos 100.0 37.1 16.5 46.4 13-14 years====-eccccmcmm cmc 100.0 38.6 15.3 46,2 15-16 years====ememccccccmcm ccna 100.0 35.5 13.5 51.1 17 years or more------cecccaccccccacccccaaoao 100.0 26.3 19.3 54.4 NOTE: Definition of type A = conditions reported on B = conditions reported on C = conditions recorded in D12 = conditions reported tacted a physician during of match: PVRS only. in interview only 24 about which respondent the preceding 12 months. PVRS and in interview which matched, PVRS and in interview which appeared to be associated. said he had con- Table 14. Number and percent distribution of chronic conditions reported in interviews for persons using SCPMG services only by education of respondent and type of match Type of match All Education of respondent condi=~ tions A B p12! Number of conditions All educational groupsS===-====--c----- 4,791 1,921 880 1,990 No education=====-cecccccmccccccccnccncccan- 89 25 24 40 1-4 yearS====e-emccccccececcccec cece ee—aa—a 92 26 15 51 5-8 yearS--==--emmcecccccmceceeccccce meme 802 353 166 283 9-12 yearS======mccccccccccccccmeemem——ae——— 2,473 995 443 1,035 13-14 yearS==-===ccececcccccccccececccee———— 678 273 108 297 15-16 yearS===-=ceeececccceccccccccccc ca ———— 393 166 63 164 17 years Or mMOre=======-eccceccccccccccaaaa= 264 83 61 120 Percent distribution of conditions by education All educational groups=========e-ee--- 100.0 100.0 100.0 100.0 No education=====-ececceccccccccmccacccccana= 1.9 1.3 2.7 2,0 1-4 years====ee-ececcecceccccccce ceca 1.9 1.4 1.7 2.6 5-8 yearS====--eeemecccccccccccccec ec 16.7 18.4 18.9 14,2 9-12 yearS===m=-eemecceccccscccmccccmcee———— 51.6 51.8 50.3 52.0 13-14 years===e--cecccccceccce ccc mcm —————— 14.2 14,2 12.3 14.9 15-16 years=====ceemmemccccccccccccnccc naa 8.2 8.6 Tue 8.2 17 years Or more===-=---cecccccmececceceeeaa= 5.3 4.3 6.9 6.0 Percent distribution of conditions by type of match All educational groups=--=---=----c--- 100.0 40.1 18.4 41.5 No educatione====--scccccccccccccncnccnccana- 100.0 28,1 27.0 44.9 1-4 years====-eccemccccccccmc mcm ———— 100.0 28.3 16.3 55.4 5-8 years----smeccecccccccccccccccm eee nme 100.0 44.0 20.7 . 35.3 9-12 yearS====mmmccmccmccccccccccccccmea———— 100.0 40,2 17.9 41.9 13-14 years======ecececceccecccccencccccaaaa 100.0 40.3 15.9 43.8 15-16 yearS=====-e-cececcccmccccccnccenanna= 100.0 42,2 16.0 41.7 17 years Or more======---eccccecccccomccanna= 100.0 31.4 23.1 45.5 Excludes conditions for which information on medical attention received during the past 12 months was not available. NOTE: Definition of type of match: tacted a physician during the preceding 12 months. said he had con- A = conditions reported on PVRS and in interview which matched. B = conditions reported on PVRS and in interview which appeared to be associated. C = conditions recorded in PVRS only. D12 = conditions reported in interview only about which respondent 25 “ en on ee ee en eee APPENDIX i. FORMS Version One of Questionnaire Form Approved: Budget Bureau No. 680-R620-F9. 1 The National Health Survey is authorized by Public Law 652 of the 84th Congress (70 Seat. 409; 42 U.S.C. 305). All information which would 1 permit identification of the individual will be beld strictly confidential, will be used caly by persons engaged ia and for the pumposes of the ¢ survey sod will not be disclosed of released to ethers for any other pusposes (22 FR 1687). Lom aA U.5. DEPARTMENT OF COMMERCE 1. Questionnaire BUREAU OF THE CENSUS ACTING AS COLLECTING AGENT FOR THE U.S. PUBLIC HEALTH SERVICE of NATIONAL HEALTH SURVEY Craton 2 ADDRESS a. Address or description of location 3. Aseigamest No. b. Mailing address if aot shown ia (a) Include city aad State 4, Serial No. F neh of these | represents your tote! Tomi iy | Income for the pest 32 mo months, thet is, your's, os, [= } (Sw Card Cand. de Income rom all all sources, such ss weges, selaries, v from property, eo ronan help reletives, 6. What Is the telephone number here? Telephone No. [J None 7. 1f sample person has aot been i lewed but i lew has been completed for other related bets, ask: As | mentioned eerlier, in eech household we ask seme special questions about one porsen for himeolf only. In this case, it is (Sample p ). Whet Is the eerliest time | would be able te see him (or her)? Date Time (Banter best time te call). . . . . 8. RECORD Item 1 Com. 2 Com. 3 Com. 4 Com. s Com. oF cALLS Eatire Date HOUSE. household be cme ——- le ce mmm] |emmm—- cm mmm=] |mee——-- HOLDS Time Record of Col.No. Date Collbecks SP for he wn «= BEERS NEE $0 | ee = w= ny —emmemmam|] Jeamememmam] | e-e-e--- Somple Person a Time | 9. REASON TYPE A TYPE 8 ' Tyee z’ FOR NON- [] Refusal OC (sposity. o.a.. Sample JowHiy [0] incesview not obeained for nin [C1 No one at home - repeated calls moved te te) Sample Person (8): [CJ Temporarily absent (8pecily reason) [C] Other (Spesity) 10. Signature of interviewer 11. Code FOOTNOTES 27 1. a. Whot is the name of the head of this household? (Enter name in appropriate Last name Last name QO column) b. What are the names of all other persons who live here? (List all persons who live here) Titer nota and totaal wie] oe Se fw oT mg. rm c. Is there anyone else who lives here who is now First name and initial First name and initial temporarily in a hospital?.....civiiiinenens LINo [] Yes (Lier) d. Away on business?. wees [CINo [J Yes (Liat) 0. On a visu vu vvnns cvnnmunsve nny ceeves [[JNo [] Yes (Liat) f. 1s there anyone else staying here now? ...... [JNo [] Yes (List) 2. How ore you related to the head of the household? (Enter relationship to Relationship Relationship head, for example: head, wife, daughter, grandson, mother-in-law, etc.) 3. How old were you on your last birthday? Age [J Under 1 year Age [] Under 1 year 4. Race (Check one box for each person) [CJ White [] Negro [J Other [OJ whiee [7 Negro [] Other 5. Sex (Check one box for each person) = Mile | Cl Mae | If 17 1d k: Under 17 years Under 17 years t yuats older over, as 0 MaiTed Separated [] Marie ya ced 6. Are you now married, widowed, divorced, separated, or never married? CJ Widowed CJ Never CC] Vidowed CC] Never (Check one box for each person) [C] Divorced Married CJ Divorced Married If 17 years old or over, ask: [J Under 17 years [J Under 17 years © 7. a. What is the highest grade you attended in school? Elem: 12345678 Elem: 12345678 (Circle highest grade attended or check ‘‘None'’) ] High: 1234 High: 1234 College: 1234 5+ College: 123 45+ None 2] Nene b. Did you finish the -- grade (year)? CJYes ~~ ™ Ne ~~ 77 [JYes "°° ° No T° 7 If 17 years old or over, ask: [J Under 17 years [J Under 17 years 8. a. What were you doing most of the past 12 months — [] Working [] Working (For males): working or doing something else? (For females): keeping house, working, or doing something else? If ‘Something else'’ checked, and person is 45 years old or over, ask: b.' Are you retired? [[] Keeping house [C] Something else [CJ Keeping house [J] Something else [1] Yes [J Ne NOTE: Beginning with Question 9, you must interview the sample person for him= self. Check the appropriate box and follow the indicated order of asking the questions. [C] Sample Person home and available — ask SAMPLE PERSON Q. 9-19 0 Sample Person not at home or not available — continue interview for during the past 12 months? If "Yes," ask: b. How many times were you in a nursing home or rest home during that period? 9. W ou sick at any time LAST WEEK OR THE WEEK BEFORE? (That is, the Ts Sd which ended this past Sunday night.) [J Yes [J No [J Yes [J No a. What was the matter? b. Anything else? 10. Last week or the week before did you take any medicine or treatment for any condition (besides . . . which you told me about! [CO Yes [C) No| [7] Yes CI Ne a. For what conditions? b. Anything else? 11. Last week or the week before did you have any accidents or injuries? [] Yes [J No |[] Yes CJ No a. What were they? b. Anything else? 12. Did you ever have an (any other) accident or injury that still bothers you or . affects you in any way? : fury y 3 Yes a Ne a. In what way does it bother you? (Record present effects) b. Anything else? BN 13. Have you had any of these conditions DURING THE PAST 12 MONTHS? Cd Yes Ore (Read Card A, condition by condition; record any conditions mentioned.) Yes No 14. Do you have any of these conditions? nl OD (Read Card B, condition by condition; record any conditions mentioned.) 15. AT THE PRESENT TIME do you have any other ailments, conditions or problems with your health? Y ’ 0 Yes One a. What is the condition? (Record condition itself if still present; otherwise record present effects.) _b. Any other problems with your health? 18. a. Have you been in a hospital at any time during the past 12 months? If ‘Yes,’ ask: C1 Yes LI Ne b. How many times were you in the hospital during that period? No. of times 19. a. Have you been a patient in a nursing home, rest home, or any similar place [J Yes CI Ne No. of times For non-sample persons 17 years old or over, show who responded for $ 9-11. For persons under 17 show who responded for them, [J Responded for self e| [] Responded for self © Col. was respondent J Last name Q Last name 3) |Last name 4 ] Last name First name and initial "| First name and inicial | First name and initial | First name and initial | Relationship Relationship Relationship Relationship e Age Age Ase [C] Under 1 year A» [J Under 1 year [CJ] Under 1 year [CJ Under 1 year [C] White [J] Negro [] Other [J White [] Negro [] Other [C] White [] Negro [] Other [CJ White [] Negro [] Other Male Male Male Male Female Female Female Female . nder Under Under 17 Under 17 Married Vy Tene, ie? VES me | Mand Separated g Mal arm ed [J Widowed [J Never Widowed Never [C] Widowed Never Widowed Never [C] Divorced Married [CO] Divoeced Married [C] Divorced Married [CJ Divotced Married [C] Under 17 years [J Under 17 years [J Under 17 years [C] Under 17 years Elem: 12345678 ©[=F 12345678 © Elem: 12345678 © FR 12345678 Oo High: 1234 High: 1234 High: 1234 Highs 1234 College: 1234 3+ College: 1234 35+ College: 123 45+ College: 123 4 5+ Molle, ew | Che - | Cone | Nowe Yes ~ Cine =" "7 Yes ~~ T(CTNeT TTT TO Yes TT TNe TTT TT Yes TT TNT" "7 7 Under 17 years Under 17 years 5 Under 17 years Under 17 years Working Working Working Work [C]) Keeping house [CJ Keeping house [C] Keeping house Keeping house [C] Something else [CJ Something else [J Something else [CJ Something else fmm mm mmr rr mmm rm rrr rrr rrr ede rrr ree mm [3 Yes CI Ne | [Yes CJ No| [1] Yes [CI No| [CO] Yes CI Ne and Tables C-1, H and P for himself. THEN ask Q. 9-11 and Table C-2 for non-sample persons. noo-sample persons only. 3] Yes No |] Yes CI No| [O] Yes [CI No | [C] Yes ZI No [ Yes CI No | [Yes (EJ No| [] Yes CI No | [] Yes CN [Yes [CONo | [J Yes [No | J Yes [No | Yes CINe [CJ Responded for self Col.—was respondent [CJ Responded for self © Col. —was respondent [J Responded for self Col. —was respondent [CJ Responded for self Col.—was respondent 29 TABLE C-1 (For i i Sol: Quass Name of condition as | Did you | Ask for all illne: nd i ! SF ONLY), ill onc line of Table Of bor ot Lon reported in Questions VER present effects of old injuries: fuk istic gany Coll tert) freon 83 for way ay in an; : i pes time Y | (a) If doctor talked to: A i 6 years 3 She k uli ro What did the doctor say » Iopaiment, » Sind . adoctor| it was? . . . did he give It or toe |Allergy* T abou} a medical nome? a Symptom vision, or Asthma “Condition’ i (b) If doctor not talked to: eye troublef Cyst wDiseage § Rested ofsingl cary and | What wos the couse of . . . Shrangsivg Sova “Thotble “ required. oe Con you What kind of 3 Auk for wll biuth | see we at of... Is it? 3 ar ot} Jopurice during aig *For an allergy or hot nw Sol wes ordinary stroke ask: newspaper | How d the What kind of | int wi rere Way di 38 injury was It? pi (stroke) affect you (a) | (b) (c) (d) (e-1) (2) (e-3) (e-4) [] Yes, x x 1 Cl Ne [C] Yes x [CI No [] Yes| : CJ Ne Bye * [Neo . [] Yes 3 CJ No aye” * [C] No [] Yes 4 CI No XC] Yes * x [CJ Neo [] Yes 5 |@ OI Ne * C3 Yeni 2 x CI Ne [C] Yes 6 CO Ne x[[C) Yes x x No [C] Yes = yr 7 [CJ Ne 3 [J Yes » x ! _ i . CJ Ne TABLE H (For SP ONLY): Fi ; ol. Ques- | You said that you were in the hospital H ; LY): Fill one line of Table H for each hospitalization ER II i . ' calendar and ask the Sy anyer v | ae hospital -- 3 ton dy enter the hospital (the JSein hs pape nome? hex do you know the madies! 2 (If exact |How mony of | How (If medical name not k (Enter month, day and year; i number not |these - - oat =) of Not Poe respondent's description) eer $ y year; if exact know nights . a date not known, obtain estimate.) accept best in the post pe ge Roablyst in or aro hod, Er ILL A as ay | ® art o "”g n | © estima) 12 lar the wouk night? required in Table C-1) same detail as (| ® ia a 0 ®) y | Y i iehts | Nia! Fon) a - ; ’ ear Nights Nights | Nights [None | Yes | No (h) © Co PY ) ; | | | 1 | | T 1 | | | \ I I 2 ! I J ! | | | | T : | | 1 1 T T : ® ! | | 1 | | | | | | FORM NHS-8:13:1 (9-4-62) 30 each condition reported in Questions 9-17 for the Sample Person. Ask only for: CL If 6-16 yrs. [|How mony During that |Ask ONLY if | Ask When did you first When did you | Ask only if Impairments and injuries Id ask: doys did 2 wees “None ONLY notice ... last see or doctor seen Aud for: ow many eep periodhow [checked in if "Yes" Check fi talk to a during the Absceens forlamiseton ys did Jou fom many days a (g) and a {Che os st box biog, J past 12 es euralgia ee ze * . id ’ 1 a | or business [keep you Bleeding Newlis ze from hast work in Bo Last Jeex pid you Enter month | During the Boil Sor or the week [all or most ave to and year if | past 12 oils res we before? of the day? EEK cut down during past | months about Cancer Soreness weel before? BEFORE did | for as 12 months; how many Sp Timor Ber nuiliee i poy - or . Fase you os gs ones wise times have 3 L ays, of ays o cut down on| a day chec| Infection Weakness cl ! entered in [the things fa “before 12 Talked 9 o 5 What part of the body is affected? None Cols. ® usually do months”’ or | doctor v Show detail for: and sm o 0), ip Enevai™ about... ? & : . o Col. 0x 1 Ear or eye - (one or both) Heod - (Skull, scalp, face) Back - (Upper, middle, lower) Arm « (Shoulder, upper, elbow, lower, wrist, hand; one or both) . % Leg - (Hip, upper, knee, lower, (8) (h) (i) 1 [2] le, foot; one or both) ssi SN smu fr 00 grates one un tt, 57 oe mee) (e-5) Days ! None\ Days None Days None[ Yes ! No Yes No (k) a) (m) * | | 0 | [) last 2 wks, [] before |M/Y. ) | | } | 2wkse3me. 12 CJ B. 12mo. " | I | | (Z] 312 months months |] Never No. of times = t t + + I | I | (]last2 wks. [] before |M/Y_______ I | I I (C] 2 wks.-3mo. 12 |] B. 12 mo. 1 | | I |) 312 months months |) Never No. of times 1 t + t x 1 | | I [7 1ast 2 wks. [7] before |M/ Yee I | . — DOES | | J [J] 2wks.-3mo. 12 (18. 12mo. [ro a i s } (CJ 312 months months|[] Never | | ' | (J last 2 wks. [] before |M/Y. | 1 1 | |) 2wkse3mo. 12 CB. 12mo. | | | I {(C] 312 months months|[] Never No. of times x | | | ) | |COtast 2 wks. [7] before |M/Y. I | I | | (] 2 wks.-3mo. 12 (JB. 12mo. | | 1 | | [J 3-12 months months |] Never No. of times T x | | } ! 1 [tase 2 wks. before [MY 1 I I | | [) 2 wks.-3 mo. 12 (J B. 12mo. I 1 1 | | (] 312 months months|(] Never No. of times T T x ' y T 1 | Sast 2 wks. 3 petore [MY | | | | | [J] 2 wks.3 mo. 12 (CJ B. 12mo. | I I I | 1 1 | I [C] 3-12 months months. (CJ Never No. of times reported in Questions 18 or 19. (If no hospitalization reported go to Table P) during ¢ If “Yes,” ask: a. What was the name of the operation? b. Any other operations? Were any operations performed on you is stay at the hospital Name of hospital | What is the name and address of the hospital you were in? (Enter full name of hospital, street or highway on which it is located, city and State; if city not known, enter county.) 1 | City and | State | Street | State | Street . ' City and | State NOTE TO INTER- VIEWER After Completing Table H go to Table P USCOMM-DC TABLE C-2 FOR NON-SAMPLE PERSONS ONLY: ‘Fill one line of ; | \ A In Col. (1) | Ask only If oy oes Name of condition as Ve ok los Sh dian. ol old i urlest fu hdd ke Sruurs g ' oy fe Dade of No. | teported 415 Suasr!e0) » ony (8) If doctor talked tor An Impalement, fine sad that Includes the words: per talk too Whet did the dester say of poot vision, [Allergy® Tumol " en doctor wos? = did he give If" a Symptom ore Asthma "Condition oy rn lings [Sr (Ble! (b) J tacngt 00) aad Ad What wos the souse of . , . ? you A kd = (e=5) as Aeeual For an allergy or stroke Ask for all Injuries during fea 4 past 2 weeks! nary (Mow does the allergy Whet part of the body was print with | (otroke) atfest you hurt? glosses? What kind of Inlury was It? Anything else? (a) | (b) (e) (C] (e=1) (e=2) (e=3) (e=4) 2 Yeo Yes * x 1 CO Ne CJ No 2 Yes C)Yes * x 2 CJ Ne CJ Ne 3 Yes CYes © ’ 3 CI Ne Ne CC Yes CC Yeo x % 4 CJ Ne CI Ne CO Yeo CC) Yes * $ 3 CJ Ne CJ Ne 1 Yeu CC) Yes * ! 6 [Ne CJ Ne Yes CIYes * 3 7 CI Ne CI Ne 2 Yeo CC] Yeo I } 8 CI Ne CI Ne ] Yes Yes * ) 9 CJ Ne CJ Ne 3 Yeo CO Yes x 3 10 CI Ne CI Ne G0 TO FRONT OF QUESTIONNAIRE FOOTNOTES 32 Table C-2 for each condition reported for each Non-Sample Person J] 2wks.-3 mo. 12 (J 312 mouths months ‘Ask only for: If 6-16 years [If 17 Ask ONLY if Ask only if |When Impairments sad injuries aid ak] [Old ol mare [During thet | Rae checked | “Yes Ta [Mra dey And for: ask: period how in columns (f) or |columa (i): |... Abscesses Inflammation Now a How meny |meny deys (8) and (bh) Did you have i Aches Newalgia ...keep |deys dd dd... LAST WEEK OR [te cut down | {Check the first Bleed Neuritis you from Ad keep you THE WEEK for as much x which Blood Clot Pains yhoo! last you eg in all or |BEFORE did es @ day? applies) . Sores week or your job most of ++. COUSO YOU Cancer Soreness week before? [or business [dev? to ot down on Cyst Tumor lost the things yeu Growth Ulcers Eater number [or the week |If any ‘‘days’’| quqlly Yet Infection Weakaess of davy = ,{ before? 2uivied . Yea chedhad What part of the body is affected? & “kip Cae Enter number Bi us) = ask Col. Gn If Show detail for: Col. (b) of days oe 1. (k No!" skip to ra (one or both) check Col. (k) ale. ae Ae : Opp. es loge Col. (h) Joa aD baad; one of Leg = ( er, knee, lower, i, wp ! one or both) ® @® ® ® 0 erp Be fmm mmm n meen den cm. (e=3) Days [Noe Days) Noae |Days | None | Yes ! NY Yes No (k) | 1 | | i [last 2 wks. [J before ! 1 ' ! Oavks3me. 12 | | CJ ¥12mosths moaths | | | | : (last 2 wks. [CT] before | | | | ! 0) 2wke3mo. 12 1 ! : [J >12 months moaths : ' : | [last 2 wks. [J before i | | | (J 2wke.-3 mo. 12 HN ! 1 | | (] 312 moaths moaths | ) | | : last 2 wks. [CO] before | | | | | OJ 2wks.-3mo. 12 } | | ! ; \ (J 312 months months ! | ! ! ' (last 2 wks. [CO] before ! 1 | Cl2vks-3mo. 12 | | | | [J 312 months months : 1 | 1 (J last 2 wks. [] before ; 1 : 2 wke-3mo. 12 | ! ! | [J ¥12months moaths : 1 | | (last 2 wks. [] before | | | | \ CJ 2wks.-3 mo. 12 | 1 ! \ | (J 312 months months i : i : ! [last 2 wks. [] before | | | | | 2 wks.-3mo. 12 | | | | | [J 312 months months ' Tr V ! ' last 2 wks. [] before \ | \ \ \ 0) 2wks.-3 mo. 12 | | | | | [CO ¥12 months months ] ' [ y | [J last 2 wke. [] before | ! | | | | ! ! | 1 38 TABLE P Name of Sample Person P-1. Have you ever been advised by a doctor to limit the amount or to avoid entirel certain kinds of food or beverages? y [C] Yes CJ No If ‘Yes,’ ask: o. For whotirensonoreonditton? = kim meme ion vio "ro; nt b. Are you still following this advice? [J Yes CJ Ne P-2. Ade presen time are you regularly taking ony medicine or treatment for any C1] Yes C1] Neo If “Yes, ask: a. For what condition? P-3. Do you have any condition which often causes you pain or discomfort? [1] Yes. CI Ne If "Yes," ask: a. What is the condition? P-4. pot fomitd health problem which Is a source of worry to you or other members of [) Yes CJ Ne f “Yes, ask: a. What is the problem? P-5. (For males): Are you limited in any way in the amount or kind of work you can do because of your health? 2 [3 Yes OI Ne b (For females): Are you limited in any way In the amount or kind of housework you can do because of your haclthh If “Yes, ask: a. What condition couses this? [CJ Excellent [] Good . , 0 | , P-6. In general, would you say your health is excellent, good, fair, or poor? [C] Fair [J Poor HAND RESPONDENT CARD TO P-7 (FORM NHS-S-13-6) P-7. List this card | conditions. PI | Syn I h condition which indicates how free! think most oth . Po o would talk about sach condition In an Imerviow iTke this Ms Tor if hey or ren other Eo of thelr family hod the condition, P-8. a. Did you work at any time during the past 2 weeks? [mR If *'No,"’ ask P-8-b and P-8-c: RT ~ b. Even though you did not work during that time do you have a ob or business? [J Yes CI No c. Were you looking for work or on layoff from a job? [C] Yes CJ No | Name and address P-9. What Is the nome and address of the doctor or clinic you usually go to for Youz own medical advice or treatment? a. During the past 12 3 aboyt how many times did you see or visit (doctor ’ Number of times or clinic named)? 4 b. Besides (the doctor or clinic nomed above) did you see or visit any other doctor during the past 12 months? } y [J Yes [J No (@e to P-10) Name and address If “Yes,” ask: Who was this? (Enter name and address) c. How many times did you see him during the past 12 months? Wugberioltiues d. Did you see any other doctors during the past 12 months? [CO] Yes [J No (Qo to P-10) Name and address If "Yes," ask: Who was this? (Enter name and address) eo. How many times did you see him during the past 12 months? - P-10. In conjunction with this survey we sometimes need to obtain additional information MEDICAL AUTHORIZATION FORM from medical and hospital records. In case you are selected as one of these persons for whom we wish to obtain additional information will you please sign this form [C] Signed (present release - Form NHS-S-13-7) which allows us to consult your health records to obtain this Information. [C] Refused: (Enter reason) NOTE TO INTERVIEWER: If interview not yet completed for non-sample persons, go back to Question 9 (on inside of questionnaire) and ask Q ions 9-11 for nple p Otherwise, go to front of questionnaire. FORM NHS-5-13-2 (9-4-02) 34 Version Two of Questionnaire Form Approved: Budget Bureau No. 68-R620-F9, 1 The National Health Survey is authorized by Public Law 652 of the 84th Congress (70 Stat. 489; 42 U.S.C. 303). All information which would permit identification of the individual will be held strictly confidential, will be used only by persons engaged in and for the purposes of the survey aad will not be disclosed ot released to others for any other purposes (22 FR 1687). omy lusdaag U.S. DEPARTMENT OF COMMERCE I. Questioansire BUREAU OF THE CENSUS ACTING AS COLLECTING AGENT FOR THE mam U.S. PUBLIC HEALTH SERVICE . of NATIONAL HEALTH SURVEY —— 2 ADDRESS 8. Address or description of location 3. Assignment No. b. Mailing address if not shown in (a): Include city and Scate 4. Serial No. 5. Which of these Inco represents your total family income for the past 12 months, that is, 's, ..'s, Group otc? {Show Card 1. Include Income from all sources, wd as wages, re from pdm ibd brig * help relatives, ete. hone No. 6. Whot Is the telephone number here? Telephone a [J Nove 7. If sample person has not been i viewed but i iew has been pleted for other related b ask: As | mentioned earlier, in each household we ask some special questions about one person for himself only. In vhis case, it is (Sample person). What is the earliest time | would be able to see him (or her)? Date Time (Enter best time to call). . . . . 8. RECORD Item 1 Com. 2 Com. 3 Com. 4 Com. S Com. or SALE Entire Date HOUSE. household - ——b————— fr mmm =| mm —— - |mmm————] | ——_——— HOLDS Time Record of Col.No. Callbacks SP for mm | Deen] Jee tied ew ae Sample Person 9. REASON TYPE A TYPE B TYPE Z FOR NON. [CJ Refusal [(Specity, e.q., Sample tamily [C] Interview not obtained /for ANTES: [C] No one at home - repeated calls WROYL 10 commit #434) Sample Person (SP) : [C] Temporarily absent (Specity reason) [C] Other (Specity) 10. Signature of interviewer 11. Code FOOTNOTES USCOMM-DC 1. o. What is the name of the head of this household? (Enter name in appropriate Last name Last name 1 column) b. What are the names of all other persons who live here? (List all persons who c. eee on else who livesihere who is now First name and initial Firat name and initial - temporarily In a hospital? ..cooueieencennns CI No [Yes (Lier) d. Away on business?...... [CJ No [] Yes (List) e. Onavisit?oooiaeinens [JNo [] Yes (List) f. Is there anyone else staying here now? ...... [JNo [7] Yes (List) 2. How are you related to the head of the household? (Enter relationship to Relationship Relationship head, for example: head, wife, daughter, grandson, mother-in-law, etc.) 3. How old were you on your last birthday? Age [] Under 1 year Ase [J Under 1 year [C] White [] Negro [] Other [C] White [] Negro [] Other b. Did you finish the -« grade (year)? 4. Race (Check one box for each person) 5. Sex (Check one box for each person) = ue | Mate a : Under 17 years Under 17 years If 17 years old or over, ask: oO aid [] Separated oO Marked ‘ (Separated 6. Are you now married, widowed, divorced, separated, or never married? CC] Widowed CJ Never [C) vidowed [C] Never (Check one box for each person) CC) Divorced Married [CJ Divorced Married If 17 years old or over, ask: [] Under 17 years ® [CJ Under 17 years oO 7. o. What is the highest grade you attended in school? Elem: 12345678 Elem: 12345678 (Citcle highest grade attended or check ‘‘None'’) High: 1234 High: 1234 College: 1234 5+ College: 1234 35+ [CJ None ClYes ~~ [ON ~~~ [C] Noae Elves "~~" "TON ~~ °° If 17 years old or over, ask: 8. a. Whot were you doing most of the past 12 months — (For male werling or doing something else? (For females): keeping house, working, or doing something else? If “Something else’’ checked, and person is 45 years old or over, ask: b. Are you retired? [C] Under 17 years [C1] Working Keeping house Something else [1 Yes [CT] Under 17 years CJ Vorking [C] Keeping house [C) Something else [J Yes NOTE: Beginnii ith Question 9, st interview the sample person for him- lt Chock the ora’ oa nus follow the Indicated order of asking the questions. [J Sample Person home and availa [C] Sample Person not at home or not available — continue interview for ble —- ask SAMPLE PERSON Q. 9-19 « 9. Wore you sick of any time LAST WEEK OR THE WEEK BEFORE? (That is, the 2-week period which ended this past Sunday night.) a. What was the matter? b. Anything else? [1 Yes CI Ne 1] Yes J No 10. Lost week or the week before did you take any medicine or treatment for any condition (besides . . . which you told me about)? a. For what conditions? b. Anything else? [2 Yes [CJ No C1 Yes [CJ Ne 11. Last week or the week before did you have any accidents or injuries? a. What were they? b. Anything else? [2 Yes 12. DURING THE PAST 12 MONTHS, have you seen or talked to a doctor about yourself? If “Yes, ask: a. For what conditions? b. Any other conditions? [Yes [ Yes CI No 13. Have you ever had to change your eating, drinking or smoking habits becouse of some health condition? If “Yes,” ask: a. What condition caused this change? Record ONLY if not previously recorded and ask: b. Do you still have this condition? “TY No (Delete) Ces = had to make any other change In of doing thi o health condition? Tr your way of colng Things QlYes EN fon caused this change? Record ONLY if not previously recorded and ask: Ce ee meme b. Do you still have this condition? [3 Yes [C] No (Delete) 15. Have you ever had any other illness or Injury which bothers you or affects you J Yes [J No in ony way? a. What are the present effects? Hand respondent gonditions card with **A’* side up and pencil, then say: CJ All No's 16. H Vv f th K I 1 ? Pl k 16. Hove you £YE% hed on of p eond tions listed on this card ease chec CJ Yes's (One ot more) Ask respondent to turn card ver (to *‘B'* side), then say: 17. Have you had ny of these conditions DURING THE PAST 12 MONTHS? CJ All No's Please check ‘‘Yes' or “'No'’ for each one listed. [J Yes's (One or more) 18. o. Have you been in a hospital at any time during the past 12 months? If “'Yes," ask: 0 Yes One b. How many times were you in the hospital during that period? No. of times 19. a. Nova you bees ° tien} J a nursing home, rest home, or ony similar ‘place [J Yes CINo- If Yes," ask: b. How many times were you in a nursing home or rest home during that period? No. of times For non-sample persons 17 years old or over, show who R responded for Q. 9-11. For persons under 17 show who responded for them. [J Responded for self Col. was respondent [C] Responded for self 36 Last name 0 Last name eo Last namq 4 ] Last name S First name and initial First name and initial First name and initial First name and initial Relationship Relationship Relationship Relationship | Age e Age ~ Age Ag [C] Under 1 year A [J Under 1 year [C] Under 1 year [C] Under 1 year | CJ White [] Negro [] Other |[] White [] Negro [] Other [C] White [] Negro [] Other |[] White [_] Negro [J Other Male Male Male Male Female Female Female Female Under 17 year Under 17 years Under 17 years Under 17 years Martie S Lie ed MartieT? S La d [CC] Masrie [] Separated 5 Mae = Separated [] Widowed [J Never [J] Widowed. [) Never [[] Vidowed [_] Never [[]) Vidowed [_] Never [Ji d Married [Di d Married [C] Divorced Married [C] Divorced Married [CT] Under 17 years Oo [CT] Under 17 years ® [0] Under 17 years 0 [J Under 17 years 0 Elem: 12345678 Elem: 12345678 Elem: 12343678 Elem: 12345678 High: 1234 High: 1234 High: 1234 High: 1234 College: 123 45+ College: 123 4 5+ College: 123 45+ College: 123 45+ [C] None _ [J None = _ _ LE] None _ a Noge BJYes ~~~ ("No ~ 7 CYes ~~ “(INo” "7 [ “J Yes (mh) Yes Ne “7 Under 17 years [C] Under 17 years [J] Under 17 years [CJ Under 17 years Working [] Working [CJ Vorking Working [C] Keeping house [CJ Keeping house [C] Keeping house Keeping house [] Something else [C] Something else [CJ Something else [C) Something else [ Yes CO No | [] Yes [J No| [] Yes [No | [] Yes CJ Ne and Tables C-1, H and P for himself. THEN ask Q. 9-11 and Table C-2 for non-sample persons. non-sample persons only. (mm Yes [CO No |[] Yes [CI Neo | [] Yes [COINo | [CO] Yes I] No [CO Yes [No | [CO] Yes CI No| [CO] Yes ' [CJ No| [] Yes Ne | Yes [No |] Yes [CJ No| [J Yes [CI No | [1] Yes [CJ Ne . Lily Looe [C] Responded for self [C] Responded for self . [CO] Responded for self [C] Responded for self 2 Col.—was respondent Col.—was respondent ® Col. was respondent ® Col.——was respondent oO 37 _ TABLE C-1 (For SP ONLY): Fill one line of Table C-1 for Col. ues- | Name of condition as | Did you| Ask for all illnesses and Ask if the entry in Col. (e-1) 4 only [Ask for any entry in No, ‘|tion | reported in Questions | EVER | present effects of old injuries: | is: — ek; ie 1) or Col. (e-2): hw Ne: 9-17 hany | (a) If doctor talked to: An SrPaiimeit os very ist fac Sdes the words son talk to | What did the doctor say £ ! asthind Allergy®* T adoctor| it was? . . . did he give it oF oy “Condition' wheut a medical nome? a Symptom Ry Cyst Disease’ tie (b) If doctor not talked to: of anykind Growth “Trouble’’ 3 Rozoed Stiginal satry and | What was the couse of . . . ? Eon Seroke § required, OM Sorrel | What kind of ... 1s 4? Ask for all injuries during enough | *For an allergy or 4 past 2 weeks: jo vod stroke ask: . he pars of the body was priate of How does the allergy What kind of injury was it? lasses? {stroke} effecs you Anything else? p (a) | (b) (c) d) (e-1) (e-2) (e-3) (e-4) [J Yes| IC) Yes * x 1 ® [J Neo [CJ No [2 Yes [CO Yes * x 2 [Ne [CJ Neo [2] Yes, *IC] Yes * x 3 ® [J Ne [CJ Ne [3 Yes, I) Yes X= x 4 CNe CJ No [] Yes x|[] Yes x x s ® CJ Ne [CJ Ne [2] Yes| x|[] Yes x x J® a Se [J Yes| |] Yes ¥ x 7 CJ No CJ No TABLE H (For SP ONLY): Fill one line of Table H for each hospitalization | Col. o | You said that you were in the hospital How many |Complete from entries in Columas (c) For what condition did you enter the No. [tion |(once, twice, etc.) during the past year = [nights were |and (d); of, show calendar and ask the hospital -- do you know the medical of No. ih io ba - in Se questions — nome? & |per- en did you enter spital (the ospital : 3 ha lost Hime)t Qf exact {How many of (How many of [Ware you ny 8 umber not | Hess < © 8e:0 = 30lLin gio (Entry must show ‘Cause,’ ‘Kind," (Enter month, day and year; if exact known nights were | nights were [hospital on _— Apart of body" or ane detail 3 date not known, obeain estimate.) aecept dene 12 hen he? Joe veaale # out Sumdey required in Table C- 1) De EAN em wong © _______|_ @ _|__o "0 | w__ (a) | (b) Month | Day Year Nights Nights Nights |None | Yes | No (b) T T T 1 i | I I ! | 1 1 1 | 1 | | I | 1 I I I | | © } | & | | | | 1 I 1 I I | | | © i] | | 1 | | | FORM NHS-8-13-2 (8-402) 38 each condition reported in Questions 9-17 for the Sample Person. Ask only for: i yu yrs. How many During that |Ask ONLY if | Ask When did you first When did you | Ask only if Impairments and injuries d ask: deys did 2 eur oe | only notice ... la see or Socios sees veo keep period how [checked in if "Yes s talk to uring the es loflummation = ay you from many days |Cols. (g) and | in Col. (Sherk she first box doctor about | past 12 Ti Y koop your job oe (h): (i): . “es months: i Ne s vo from To business kon LAST WEEK | Did you Enter month | During the ot Pai schpol | ’ BU ’ k: 3 oll or most OR THE have to and year if | past 12 Sores week or rl o wav of the day? WE cut down) durirg months about Soreness week b ore? ore Y® |BEFORE did | for as 12 months; how many 'umor Enter number| If any «+. cause you | much as otherwise times have 3 Ulcers of days, or | ‘‘days’’ to cut down on| a doy? check you seen or € Infection Weakness Fheck co in [the hinge PY ee 12 Jed toa 8 " “None"’ ols. usually do months'’ or loctor What part of the body Is affected?| and ask or (h) ©, “never” about ... ? & Show detail for: Col. (h) ol. ky ox wl Ear or eye - (one or both) Head - (Skull, scalp, face) Back - - (Upper, middle, lower) Arm - | houlder, upper, elbow, oer and; one or bot i 3 Leg - Hip, upper, knee, lower, (f) (h) (i) 1 G0) niki, ODT; ORE OF BOtRY Lf" = =m momo wt wt Joon vm. os i oom. a “TC (e-9) Days ! None [Days None Days None] Yes ! No | Yes No (k) ) (m) x ) \ | 0 | [test 2 wks. (3 before [M/Y. | | | | 1 [J 2wks.-3mo. 12 (J B. 12mo. fa 1 | I I I | (C2) 312 months months [(] Never Somes | % | i i | 1 [1st 2 wks. (7 before |M/Y. _ | | I I I (J 2 wks.-3 mo. 12 (CJ B. 12mo. . 2 | | | | I [] 312 months months |] Never No. of times t 1 ye t t 2 1 1 | | I {J last 2 wks. [7] before [M/ You Does ! NOT | | ! I |Da2wks3mo. 12 [CIB 12m0. | orimen | 3 | | | ! | (0) 312 months months |] Never x ) \ \ | | [Cast 2 wks. [) before |M/Y. APPLY | | I 1 |] 2wks.-3mo. 12 () B. 12mo. 4 | | | | | (J 312 months months |] Never No. of times x \ \ \ | ) (J last 2 wks. (J before |M/Y. | | | | | (J 2 wks.-3mo. 12 (J B. 12mo. | | | I | (J 3-12 months , months((] Never No. of times 5 o ! ; : y | [last 2 wks. (before [MY | | | | | [] 2wks.-3mo. 12 (1B. 12mo._ 6 | 1 | | I |) %12 months * months|[] Never No. of times x : ! y ] [test 2 wks. 3 vetore [MY ) | | | | (2) 2 wks.-3 mo. 12 (J B. 12mo. - 7 1 | | | | (] 312 months months|(] Never No. of times reported in Questions 18 or 19. (If no hospitalization reported go to Table P) Nore any spsatians periomed on you What is the name and address of the hospital you were in? “w ”" NOT If Yes,’ ask: {Ene full name of hospital, street or highway on which it is ore a. What was the nome of the ocated, city and State; if city not known, enter county.) To operation? INTER- b. Any other operations? VIEWER ee __W a Tn. om. 8 FT 500 mt me. ms ___ a on i LI After Yes | If “Yes,” name of operation, etc. [No Name of hospital I Address . i Completing \ linens Tym SET 4 Table nt | ! City and ' | State go to ’ Jsetest Table P yt mmm emo | City and | | State : J 4 | City and TT 3 | State USCOMM-DC 39 TABLE C-2 FOR NON-SAMPLE PERSONS ONLY: Fill one line of Table C-2 y Ask It Ia Col. (e=1) [Ask only Ift|Ask f Sa us. Name of condition ss Ba kt Shue fold fa 3 fie tines " Sysars 0d Gel, al ea of No, | reportedls Questions (8) If doctor talked toi As Impalrment, fiona aad. ma ant i“ me or I lon, 4 ‘ " pe ae Musi |0 op o s medics! nome! a ! |dokee “Frowie" (b) § doctor Raita adin) Wheat wes the seuse of , . . ? Con you Whet kind of. + . Is It? i se (3) a8 oh | *For an allergy ot stroke i a hn log ordinary yy, dogs the ollergy port of the body wes prime wii (stroke) atfoet you glosses? Tibi ol tiury wos. 0 (a) | (b) (e) @ (e=1) (e=2) (e3) (ed) 2 Yes - CI Yes * : 1 CJ Ne CI Ne CE] Yes CC Yes x * 2 CI Ne CI Ne CC Yes CC Yeo 2 2 3 CI Ne CI Ne CJ Yes C1 Yes 2 x 4 Ne’ CI Ne CC Yes CC) Yes * 3 CI Ne CI Ne 2 Yeo CC) Yes * gl 6 Ne CaNe ’ CC Yeo CC) Yes ¥ * ? CI Ne CNe C1 Yes CC Yeo * 9 s CI Ne CNe CO Yee CO Yes ¥ ¥ 9 CI Ne CI Ne 2 Yeo Yes & 5 10 CJ Ne OI Ne 00 TO FRONT OF QUESTIONNAIRE FOOTNOTES 40 for each condition reported for each Noo-Sampi# Person Ask only for: If 6-16 year: 7 Ask ONLY if if [When did a aad injuries old ski * JL, During thet *Noae'* checked Aa yA first notice And foet How meny ok: period how [in columas (f) or |columa (i) |... Abscesses Inflammation days did How many |many doys |(8) sad (bX youhave | tire Actos News is oo hoop deys did a on LAST WEEK OR [to cut down hack th e Newr'tls you from ls + keep oop W THE WEEK for as much applies) Blood Clot sine school last [you from In olor BEFORE did as a day? Pp! Cader Sorsnans eer Ca Ty Cl § ne fo cut Cyst Tumor olin wn ™ " y ou Gro Ulcers Eater number |o, the If any ‘‘days’’| he fot ings J Infection Weakness of days or | before? entered in . " What part of the body Is affected? ape Enter oumbed (10 sire” | bok Cok. ei Show detail for: Col. (bh) of dave or en’ (k; "No" ski ok or Ea Jone or bot) None 1. (k) Upper, m oan lower) and ask Soa ulder, upper, elbow, Col. (b) ower, wrist, hand; one or Leg - et, knee, lower, 0 (Hip, oot; one or both) (3) 8) (b) Mm G4) le op come. Scr rg fi gl) rc (e=3) Days [ Noe Days) None Days | None | Yes No Yes i No (k) | 1 + 1 | | (last 2 wks. [) before | ' ' f ! CO) 2wks-3mo. 12 | i CJ) >12 months mouths | | | k ! [last 2 wks. [] before | | | | } () 2wks.-3mo. 12 ! | | i CO) >12 months months ! | } : i [C) last 2 whe. [2] before | | | | | [2 2wke.-3mo. 12 | = 1 | | [] 312 months moaths | \ 1 | last 2 wks. [] before 1 \ 1 1 DO 2wke-3mo. 12 ! | ! i C312 months months | ME | \ : [last 2 wha, [CO] befoee | ; \ | Olavke3mo. 12 | | \ | | [J 312 moaths months ! } | | | [last 2 wks. [] before ; | ! ' D2 wks-3mo. 12 ! | ! | [2 312 months moaths | y | | 1 (last 2 wks. [] before ! : 1 | OO 2wks.-3mo. 12 ) ] y \ (J ¥12 months moathe ! : ¥ ! ! [last 2 wka. [] before ] | I : O2vks3mo. 12 | 1 | | \ [2] 312 months months ! ] ! v ! (CD) last 2 wks. (] before : ! I J 2wkae3mo. 12 | 1, | | \ [C1 312 moaths months 1 ' 1 ! ! [last 2 wks. [1] before b ' : DO 2wks-dme. 12 | | | | = [CO] »12 mooths ~~ months 41 TABLE P [ Name of Sample Person Pl (Dees net apply) P-2, At the present ti | ’ Mh the pro) me are you regularly taking any medicine or treatment for ony If “Yes, ask: @. For whet condition? C1 Yes [J Ne De you have ony condition which often couses you pain or discomfort? If Yes," ask: . eo. Whet is the condition? [1] Yes CINe P-4. Do you have any health problem which Is @ source of worry to you or other members of our ly’ IV Yes, ask: a. What is the problem? [J Yes CI Ne P-S. (Does not apply) P-6. In general, would you say your health is excellent, good, foir, or poor? [J] Excellent [] Good [CO] Fai [CO] Poor HAND RESPONDENT CARD TO P-7 (FORM NHS-S-13-6) medical advice or treatment? P-7. List thi | ‘conditions. PI la exes I h condition which indicates how free! think t other roan o would tal ‘about soch condition In an Interview Tie this Ree hey or Shit other rn of thelr fam hod the condition. P-8. a. Did you work at any time during the post 2 weeks? [J Yes CC] No If “No,’” ask Pg-bond Pc: | FE=s=====-= a eer eg Sw a b. Even h you did not work during that time do you have a job or business? [J Yes CI Ne ¢. Were you looking for work or on layoff from a job? CJ Yes [CJ Ne Name and address P-9. What is the nome ond address of the docter or clinic you usually go to for your own cr Dorig Hive pouy 12 aviv abut hus tony fies 314 you swe or visht (doctor Nomber of times oT a During # post) 4 Sho many times you see or vist (i b. Besides (the doctor or clinic named above) did you see or visit any other doctor OC Yes [] No (Qe to P10) during the past 12 months? If “Yes,” ask: Who was this? (Enter name and address) c. How mony times did you see him during the past 12 months? Name and address Number of times d. Did you see any other doctors during the past 12 months? [ Yes [CJ No (Go to P-10) If “Yes," ask: Who was this? (Enter name and address) eo. How many times did you see him during the past 12 months? Name and address Number of times P-10. ju iqonjinstin with this survey we sometimes need to obtain additional information medical ond hospital records. In case you are selected as one of these persons we wish to obtain additional information will you please sign this form (ptesent release - Form NHS-S-13-7) which allows us to consult your health records to obtain this information. MEDICAL AUTHORIZATION FORM [0 Signed [CO] Refused: (Enter ) NOTE TO INTERVIEWER: If interview not yet completed for non-sample persons, go back fo Question 9 (on Inside of questionnaire) and ask Questions 9-11 for ncn-sample persons. Otherwise, go fo front of questionnaire. FORM NHS-5-13-2 (9-4-02) 42 Version Three of Questionnaire Form Approved: Budget Bureau No, 68-R620-F9. 1 The National Health Survey is authorized by Public Law 652 of the 84th Congress (70 Stat. 489; 42 U.S.C. 305). All information which would permit identification of the individual will be held strictly confidential, will be used only by persons engaged in and for the purposes of the survey and will not be disclosed or released to others for any other purposes (22 FR 1687). ony. NNs-5-13.3 U.S. DEPARTMENT OF COMMERCE 1.” Questicunsie BUREAU OF THE C SUS ACTING AS COLLECTING A T FOR THE U.S. PUBLIC HEALTH SERVICE of NATIONAL HEALTH SURVEY LT me 2 ADDRESS] a. Address or description of location 3. Assignment No. b. Mailing address if not shown in (a): Include city and State 4. Serial No. 5. Which of these income groups represents your total family income for the past 12 months, that is, your's, your--'s, Group etc? (Show Card H). Incl Income from all sources, such as wages, salaries, rents from property, pensions, help from relatives, etc. hone No. 6. What is the telephone number here? Telephone No [] None 7. If sample person has not been interviewed but interview has been completed for other related members, ask: As | mentioned earlier, in each household we ask some special questions about one person for himself only. In this case, it is (Sample person). What is the earliest time | would be able to see him (or her)? Date Time (Enter best time to call). . . . . 8. RECORD Item 1 Com. 2 Com. 3 Com. 4 Com. 5 Com. OF CALLS AT Entire Date HOUSE. household beepers] mmm |mm—_————] |=] mm HOLDS Time Record of Callbocks fr | Lo Eames. ll lemme] remem] Je————— Sample Person 9. REASON TYPE B TYPE Z FOR : NON- [C] Refusal [C] (Specity, o.4., Sample family [C] Interview not obtained for JT SR [CJ No one at home - repeated calls maved te. #4) Sample Person (SP) : [] Temporarily absent (Specify reason) - [C] Other (Specity) 10. Signature of interviewer MN. Code FOOTNOTES USCOMM-DC 43 1. a Wah is the. name of the head of this household? (Enter name in appropriate column Woah are the nomes of all other persons who live here? (List all persons who ive here) « Ja Shure saysng ales whe tives Te who 1s now [CONo [] Yes (Liev) [CJNo [CO] Yes (Liat) d. Away on business?..... e. On a visit?. sesesssseseses [JNo [] Yes (Liet) [No [] Yes (List) f. Is there nan. else staying here now? ...... First name and initial Last name 1 Firat name and initial 2. How ore you related to the head of the household? (Enter relationship to (For males): working or doing something else? (For females): keeping house, working, or doing g else? If “Something else’’ checked, and person is 45 years old or over, ask: b. Are you retired? oh Relationship Relationship bead, for example: head, wife, daughter, grandson, mother-in-law, etc.) 3. How old were you on your last birthday? Age [J Under 1 year Age [J Under 1year 4. Race (Check one box for each person) [0 White [] Negro [] Other [0] White [T] Negro [J Ocher Mal, 5. Sex (Check one box for each person) b= A h Mue | : Under 17 years Under 17 years U 17 'yeait od or over, Wek [] Married Separated 0 Marche? [Separated 6. Are you now married, widowed, divorced, separated, or never married? C3) Vidowed CJ) Never C5) Widowed CJ Never (Check one box for each person) C0) Divorced Married [) Divorced Married If 17 years old or over, ask: [C] Under 17 years ® [J Under 17 years Qo 7. ao. What is the highest grade you attended In school? Elem: 12345678 Elem: 123456768 (Citcle highest grade attended or check ‘‘None'’) High: 1234 High: 1234 College: 1234 5+ College: 123 4 5+ None [J None b. Did you finish the -- grade (year)? Yes “MN "°°" [Yes "" "THN ~~~" If 17 years old or over, ask: [J Under 17 years [C] Under 17 years 8. o. What were you doing most of the past 12 months — [J Working [] Working [C] Keeping house [C] Something else [[] Keeping house [C] Something else [C] Yes TE: Beginnii ith Question 9, st interview the somple on for him- No sell. Check the appropriate’ box and follow the indicated ra of asking the questions. [C] Sample Person bome and available — ask SAMPLE PERSON Q 919 [CJ Sample Person not at home or not available — continue interview for ou sick at any time LAST WEEK OR THE WEEK BEFORE? (That is, the period which ended this past Sunday night.) a. What was the matter? b. Anything else? 9. Were 2 [1 Yes CJ No [J Yes [J Neo 10. Last week or the week before did you take any medicine or treatment for any condition (besides . . . which you told me about)? a. For what conditions? b. Anything else? [] Yes [Ne [] Yes [J No 11. Last week or the week before did you have any accidents or injuries? a. What were they? b. Anything olge? [CO Yes [J Neo 12. Did you ever have an (any other) accident or injury that still bothers you or affects you In any way? a. In what way does it bother you? (Record present effects) b. Anything else? [J] Yes [CC] Yes [J Ne Hand respondent conditions card with ‘*A’* side up and pencil, then say: 13. Have you EVER had any of the conditions listed on this card? Please check “Yes or “No'’ for each one listed. [CJ All No's [J Yes's (One or more) Ask respondent to turn card over (to *'B’’ side), then say: 14. Have you had any of these conditions DURING THE PAST 12 MONTHS? Please check “Yes'' or No’ for each one listed. [CJ All No's [J Yes’s (One or more) 15. At the present time do you have any other ailments, conditions, or problems with your health = besides any you may have checked on the card or any that you told me about? a. What is the condition? (Record condition itself if still present; otherwise record present effects) b. Any other problems with your health? [3] Yes 18. a. Have you been In a hospital ot any time during the past 12 months? If “Yes,'* ask: b. How many times were you in the hospital during that period? [] Yes No. of times 19. a. Have you been during the past If *Yes," ask: b. How many times were you In a nursing home or rest home during thot period? a patient in a nursing home, rest home, or any similar place [J Yes CJ Ne No. of times Cre f For non-sample persons 17 years old or over, show who responded for Q. 9-11. For persons under 17 show who responded for them. [CT] Responded for self [C] Responded for self Col.—_was respondent 44 Last name 2)| Last name eo Last namg 4 ] Last name S First name and initial | First name and initial | First ame and inicial | First aame and inicial | Relationship Relationship Relationship Relationship ‘| Age Age Ase [CJ Under 1 year 4b [CJ Under 1 year [CJ Under 1 year AR [CT] Under 1 year [C] White [] Negro [] Other |[T] White [—] Negro [] Other [] White [] Negro [J] Other | [] White [] Negro [J Other Male Male * [J Male [CJ Male Female ] Female [C] Female [C] Female Under 17 years Under 17 years Under 17 years Under 17 years Marri H S I= d Masri i. S fe d [2] Mari [) Separated 0 Male [3] Separated Widowed [C] Never Widowed [_] Never [C] Vidowed [_] Never [[] Vidowed [_] Never [CJ Divorced Married [C] Divoeced Married [C] Divorced Married [CJ Divorced Marzied [C] Under 17 years QO [C] Under 17 years ® [CJ Under 17 years ©) [C] Under 17 years Ol Elem: 12345678 Elem: 12345678 Elem: 12345678 Elem: 12345678 High: 1234 High: 1234 High: 1234 High: 1234 College: 123 45+ College: 1234 5+ College: 1234 5+ College: 123 45+ [CJ None ~ | J None _ _. None Fos = =" FCT Yes Tro EIS“ rR" wn © Tab Under 17 years Under 17 years Under 17 years Under 17 years Vorking Working Vorking Working [CJ] Keeping house [C] Keeping house [C] Keeping house [CJ Keeping house [C] Something else [C] Something else [C] Something else [C] Something else tee emer cc —c—- -rme—a -———————— Hem ——— -———————— i 00 [] Yes CO Ne | [] Yes [CJ No| [] Yes [Ne | [J Yes CJ Ne and Tables C-1, H and P for himself. THEN ask Q. 9-11 and Table C-2 for non-sample persons. non-sample persons only. ' [1 Yes [No |[] Yes [CJ Ne| [J Yes ’ [CJ No | [J Yes J No Yes [No |[] Yes CO No| [J Yes [Ne | [] Yes Ne [J Yes [No |] Yes [J No| [J Yes [No | [J Yes CI Ne — [CJ Responded for self Col.——was respondent [CJ Responded for self Col. —was respondent [CJ Responded for self © Col. was respondent [J Responded for self Col.——was respondent 45 TABLE C-1 (For P ONLY): Fill one line of Table C-1 for each Col. Name of conditi Did you| Ask for all ill aad [Ask in Col. Askoaly [Ask for in fo er a J A dot seve omer ond fares eo: VATE If the eazy in Cale) bo [Co feb Col, (2% oe. 17 0 (a) f doctor talked to: No Segoe vit pte ek iecludes the Suess son telk to | What did the doctor sey A sadblind-| o adoctor| it wes? . . . did he give it or S008 pose Ashol Condition’ Shevh © medicel name? a Symptom eye trouble] Cyst Disease" (b) If doctor not talked to: any kind Grow ‘Trouble Resocd ortsinsl sutry and What wes the cove of... P[ Stroke : fequired. ra — Ye Whet kind of ... is ih? *F. aller, 1 JT TES, [fare = What pert of the body wes newspaper | Mow does the of hort? troke Whet kind of injury was it? oy (sabia) effesr you Anything else? (a) | (b) (c) (d) (e-1) (e-2) (e=3) (e=4) i [] Yes ; BIC) Yes * * © eh Sh CT] Yes) IC] Yes * ® 2 CI Ne CI Ne © [2 Yes “IC Yes * = 3 |@ Cie Owe | [J Yes EI] Yes x 4 ® CINe CINe [2] Yes) 2{[] Yes =x x s @®@ CINe CINe [2 Yes) 2] Yes = = 6 Ore Cine [2 Yes IC) Yes * . ? Cine CI Ne TABLE ” Fill one line of Table H for each hospitalization reported Col. You seid thet in the hospital |How many from entries ia Columas (c Por whet condition did you enter feo t "| (ome, twice, are.) during the post yoor — te aire ad Th er. whew Chantar and atk vie - rT he 1" awn digovomwaniomniinn [Forni - ' par i Soa loot Hemel (0 exact [How mony of {How many of [Wore you TIN at AL Rao, athr sumbet not [Whose - - these - - "oe ” (Eater month, day and year; if exact |kaown — |Sights wore nights ware (hospital en | (Eacry must show [Cases Kind, i date not knows, obtain eotimate.) accept bam in he eet ha or [loot Sundey Tequired ia Table C-1) a — J lw le fod ds (a) | Mosh | Day | Year Nights | Nights {Nose | Yes | ™ I I T T | | 1 ' 1 | 1 | | © | ' | | 1 1 T T | | | | © Po P| 1 1 | | T T T T | ' ' 1 3 1 1 1 1 ® 1 1 | ' | FORM MH80+10+D (90-00) 46 condition reported in’ Questions 9-17 for the Sample Person. Ask only for: ne If 6-16 yrs. flHow mony During thot |Ask ONLY if | Ask When did you first When did you | Ask only if Impairments and injuries Id ask:, doys did 2 woe, oman onl ) notice ... ? last see or doctor seen A v veo keep period how [checked in i “Yes 3 talk to o during the Roscenses Inflammation Cys Gd” you from many days |Cols. (g) and | in Col. {Glan ihe $113: box doctor about | past 12 Neuralgia keep your job did... (h): (i): PP ee months: Bleeding Neuritis yok from or businnss kop Jo LAST WEEK | Did you Enter month | During the Blood Clot Pains school lat Sa) wee k nn . ' OR THE have to and year if [past 12 Boils es w or fhe bef. . re . rid So WEEK cut down| during past | months about Cancer Soreness wee ore? ory oftheday’ |BEFORE did | for as 12 months; how many Cyst Tumor Enter number| If any . cause you | much as otherwise times have 3 Growth Ulcers of days, or | '‘days’’ to cut down on| a day? check Infection Weakness Sheek Loupe in |the things Ir “‘before 12 5 one’ ols. (g) [usual lo’ hs'’ What part of the body is affected? and ok or (h) tip very oe, © shout ..'? § Show detail for: Col. (h) to Col.(k) x a Ear or eye - (one or both) ot « (Skull, scalp, face) - {Uppers middle, lower) Bank - {Shoul der, upper, elbow, Lower, ; Yen and; one Leg - yi BY tr, knee, lower, f) 8) (h) (i) i) ank! nkle, foot; one of both)’ —————t TT 1 -= =" (e=5) Days ! None\|Days !None | Dayd None| Yes ! No | Yes No (k) (1) (m) L | \ \ \ | [last 2 wks. [] before |M/Y. 1 I 1 | | (J 2wks.-3mo. 12 (J B. 12mo. © 1 | | | | y | |CJ3%12 months months |] Never 0. of times t + + + + % 1 1 1 | I [Cast 2 wks. [7] before |M/Y. I | | I I (J 2 wks.-3 mo. 12 (CZ) B. 12mo. 2 | | | | | |] 312 months months |) Never No. of times + t 1 t t = 1 | | | I |) last 2 wks. [7] before [M/Y—oux | I | I I - DOES ' NOT (J 2 wks.-3 mo. 12 (JB. 12mo. x 3 ! ! | ! } () 312 months ~~ months |[] Never No. of times ) ) | \ I [last 2 wks. [] before (M/Y________ AFPLY | | | I |) 2 wks.-3mo. 12 (1 B. 12mo. 4 | | | | | |) 312 months months|[] Never No. of times . p \ 1 \ 1. | CO tast 2 wks. ( before [M/Y. | | I | 1 [J 2 wks.-3 mo. 12 (2 B. 12 mo. I | | I I {Z] 3-12 months months|[] Never No. of times 5 x ' | | \ 1 [test 2 wks. before [WY 1 | | | I ( 2 wks.-3mo. 12 (JB. 12me. 6 | | | | | [C] 312 months moanths|(T] Never No. of times x ] : : y ! [J last 2 wks. [CO] before [M/Y — | | | | | [) 2 wks.-3 mo. 12 (J B. 12 mo. - 7 | ! |. | | | | 312 months months|(] Never No. of times in Questions 18 or 19. (If no hospitalization reported go to Table P) Were any o| tion s performed on Is th f the h in? doring a, ” “h hospital? on you What Is the name and address of the hospital you were in NOTE If “Yes, ask: Bees fol name of hospital, street or highway on which it is T0 a. Whot was the name of the ocated, city and State; if cley not known, enter county.) operation? INTER- b. Any other operations? VIEWER mm A ee eee After Yes J Yo” "name of operation, etc.| No Name of hospital i Address Combi | Street ” - z - - ——— em ————————— RK Table H ity a ! re go to : | Street Table P ! |Coyamd ~~~" "TTT" TTTmmmmT | | Stare T | | h USUCOMM-DC 47 TABLE C-2 FOR NON-SAMPLE PERSONS ONLY: Fill one line of Table C-2 Col. |Ques- Did yew Ask for all illnesses Ask if the entry in Col. (e-1) | Ask oaly if:|Ask for any ei in No. tion Name of condition as by R present effects of old +. is: $year old cal (eDor Col. (e- (e= of No. epserel 19 Questions ey (a) If doctor talked to: An Impairment, blindaces, at tae w per tolk to @ |Whet did the dector sey It of poor vision, | Allerg son doctor was? «- did he give ww a Symptom oe oir Anti Caiton’ Sbev} @ medical name’ os feof Soroke® “Trouble 8) Jf doctor dy talked o0: od Wheat wos the couse of . . . ? What kind of. . . Is I? i ask: (¢-2) = (¢-3) as seo well (op, on allergy or stroke 4 requieed, to reed ask: Ask for all injuries during ord) 4 past 2 weeks: A. How does the el Whet pert of the body wes print with |(stroke) affect you glosses? Worry kind of Inlvry was It? Anything else? (a) | (®) (c) (d) (e=1) (e-2) (e-3) (e-4) Yes [2 Yes * . 1 CJ No CJ No [] Yes ] Yes * 5 2 OI Ne CI Ne [C] Yes [Yes * = 3 [CJ No [No [J Yes [ Yes x r 4 [No [CJ Neo [ Yes [Yes * 8 5 CJ Ne Ne [] Yes [J Yes * : 6 CI No Ne [] Yes [Yes ¥ xX 7 [CJ Ne CI Ne [C] Yes [1 Yes 3 ® 8 CI Ne CI Ne [] Yes [Yes * 5 9 [CJ Ne [CJ Ne ] Yes [Yes © b 10 CI Ne CI Ne GO TO FRONT OF QUESTIONNAIRE FOOTNOTES 48 for each condition reported for each Non-Sample Person [last 2 wks. [] before Ask only for: If 6-16 years [If 17 years Ask ONLY if Ask only if | When did Impairments and injuries old ask: old ay Duris ther **Noae’’ checked |‘‘Yes’ Ts oo Lo And for: Hovwmony ask: period how in columns (f) or |column (i): oe Abscesses Inflammation days did How many |many days (8) and (hb): Did you have Aches Neuralgia «++ keep doys did did... LAST WEEK OR [to cut down | {Check the first Bleedin Neuritis you from ... keep [keep you THE WEEK for as much | box whic Blood Clot Pains school last [you from . [in bed all or |BEFORE did os aday? | epplies) Boils res week or the [your job most of the +. +. cause you haga Soreness week before? I business [doy? to cut down on Growth Ulcers Eater number ak If any ‘‘days’’ [ie things fou Infection Weakness of dave 2 before? Exteled in " REE chec! one’ Cols. or Yes’ checke: What port of the body Is affected? | JG 0 N0™ | Enter numbed ® LW i aay Gy If Show detail for: Col. (b) of days or 1. (k) **No*’ skip to Sas or A 4 « (one or both) check Col. (k) ull, scalp HN None Bod: Uppet, middle, lower) and ask Arm {Sie ulder, upper, elbow, Col. (h) ower, wrist, hand; one or both) Leg - Hip, er, knee, lower, ankle, foot; one or both) ) (e-5) Days | None |Days| None |Days None | Yes No Yes No (k) | | | 1 ! [last 2 wks. [] before , , ! | OO) 2wks.-3mo. 12 | 1 | | [J 312 months moaths i . | ; : ' [last 2 wks. [T] before | | | | [1] 2 wks.-3mo. 12 | ! \ \ | (C1) 312 months months 1 1 ! | : \ [last 2 wks. [CO] before | | 1 1 | [3] 2wks.-3mo. 12 | } | | | [] 312 months months | H | | y (J last 2 wks. [] before | | I I | [J 2 wks.-3 mo. 12 | ! : | [2 3-12 months months : \ y [last 2 wks. [] before \ I | | | (J 2wks.-3mo. 12 | | | | | [1 312 moaths months y ! | y (last 2 wks. [] before | \ | | [J 2 wks.-3mo. 12 | | | | | [CJ 3-12 months months ] ) | ! [last 2 wks. [(] before | [J 2wks.-3 mo. 12 I 1 ! [J 312 months months | y ' 1 1 | | | | | 1 | | | | | | | | | | ? | 1 | | | | | | | | | | | | | | | | | y | ! ! | ! I | ! | ! | ! | | | | | 1 [2wks.-3mo. 12 [0] 312 months months [last 2 wks. [] before J 2wks.-3mo. 12 [J 312 months months [last 2 wks. [] before [0 2 wks.-3mo. 12 [J] 312 mooths months 49 TABLE P . [Mame of Sample Person P.1. Neve you ever boon advised by doctor to limit the amount or te avold entirely certain n hinds of food or bavereges? HM **Yes," ask: @ Por whet reason or condition? b. Are you still following this edvice? [ Yes [C Yes P-2. At the present time ere you regularly toking any medicine or treatment for any condition? ¥ "Yes," ask: & For whet condition? [] Yes De you have any condition which often —v you pain or discomfort? ¥ “Yes, ask: o. Whet is the condition? P-3 [2] Yes CNe Pd. Du yee have oy haath problem wAIeh 14 seven of wary fo ow 1 for mambes of Eremit » hs o Whet ls the problem? [2 Yes CI Ne P.§, (For males): AS 1S Ning fe Sl way In the ameunt. ox Sil of wel. you son do becouse of your hes! For fe: aay dew limited In the amount or kind of hevsewerk (For fema! TR Inns wy or . you con — «. Whet condition couvses this? [1 Yes P-6. In general, would you sey your health Is excellent, goed, fair, or poor? [CJ] Excellent [] Good [CJ Fair [CO] Poor HAND RESPONDENT CARD TO PTIRORN NMS-S-13-6) P72. Lisi mil LS3rd ues tions, Pleese on XX" | this = thet wort Th h conditl Whey or a. Jin inthe Kdicutes ho how red think i” ise hod the ¢ ition In on Interview Ps. & Did you werk ot ony time during the pest 2 weeks? If “ No," ask P-8-b end P-B-c: . b. Bven Neb utih. dering It Vime dbyen Rave o lok wr businnss? e. Were you job? ng for werk or on layoff from © CI Ne PS. What Is the name end address of the decter or clinic you usually go te for YOUT OWN medical edvice or trostment? @ During the pest 12 months about how many times did you see or visit (doctor or slinic ? ourself Number of times b. Besides Sr chile hom sbove) did you see er visit any other decter during the past 12 menths? [J Yes [CJ No (Ge to P-10) U "Yes," ask: Whe wes this? (Enter name and address) a. How mony times did you see him during the pest 12 months? Name and address d. Did you see eny other doctors during the pest 12 menths? If “Yes,’’ ask: Whe wes this? (Enter neme and address) ©. How mony times did you see him during the pest 12 months? P10 Foie nn i 3 phon we sometimes need 94 it Ena Information W case you ere these persens for whem we wish te obtain odditionel information will you please s sign this — {eataet aia relossel Von Form N NHS-S 157 which allows vs te consult your reserds MEDICAL AUTHORIZATION FORM [Signed [OO Refused: (Enter tron! Questions 9-11 for non-somple persons. Otherwise, go to front of questionnaire. NOTE TO INTERVIEWER: If interview not yet completed for non-sample persens, Trane TT eT FORM MNES 19-8 (94-00) 50 Physician Visit Record ve. PHYSICIAN VISIT RECORD — NATIONAL HEALTH SURVEY Budget Barve To, 64-0420.88 ——— Meniter. If Possible, Complete Question A Before Patient Is Seen By Doctor A. Patient's Last Name First Nome Medico Racord Number Doctor's Name Clinic DOCTOR: Complete One Column of Questions 1 through 8 for: (Condition) or Impression (I.E., Diabetes, Hypertension, etc.) and —Each (Joint Pain, Skin Rash, etc.) Not a Part of Diagnosis (Condition) or Impression, Provided That the Diagnosis (Condition), Impression, or Symptom Was Considered, Noted in Record, or Mentioned Today By Either You or the Patient. If More Than 2 Col Are Needed, Use the Continuation Sheet !f There Is Ne Diegnesis (Condition), Impression, or Symplom for the Patient, Check [J and Print Reason for This Visit: Then, Skip to Question 8. Column 1 Column 2 Print Medical Term Print Medical Torm — (1). PRINT name of diagnosis (condition), impression) or symptom (medical terms if possible) (2). Was the diagnosis (condition), impression, or ves [wm] Noi] YES J NO cl symptom mentioned by you today? Mint term vied Print form vied If yes, PRINT term used (3). Was the diognesis (condition), Impression, or vas[l_] Noi] ves] wo J symplom mentioned by the patient today? Print term weed Print form vied If yes, PRINT jorm used Th (4). When do you think the patient first became ’ aware of the diagnosis (condition), mpression. Chock one box Check one ban or symplom specified in Question 1 a. Over 3 months ogo 9. | — o [CJ b. During past 3 months but before today ». b. ¢. Today « 3] < | Ex < A one "* Be « EB (5). How much emphasis did you give today to the LJ Made =. paint of - Made 'a pointe 1 diagnosis (condition), impression, or symptom EJ Poyed it down ET] Ployed it down $pacified in Question 1? [Ex Neither of these [Ea Neither of these (6). At some time during the past week wes this Chock one bex | diagnosis (condition), impression, Chak on bur. In such Vine San but/1a och fide ny oui. Impression; ot syomIem ves NO DON'T_KNOW No DON'T_KNOW. a. Marked or moderate pain a. cc | F— 63 a. oc | F— |} b. Morked or moderate emotional stress . [CJ | [om LC Ec EE o Gear dom dor 3 bed ca Ba Cm Ba = d. Other change in activity « [C3 = (a « [I [om «. Other trouble (PRINT) oe ’ immense | o (7). Action taken today related fo the diagnosis : (condition), impression, or symptom specified ca No: action: foken:todoy ET No sition taken; today in Question 1. (De net enter actions taken ONLY] for purposes of a routine physical examination.) Chock epplicable boxes Check applicable boxes Mentioned, but Mentioned, but Ordered or not erdorod Ordored or not ardrod 6. Medication or T=) . — 6 L— a £3 b. Laboratory fests 70 cc b. — le70 | oo] ». —] c. X-ray examination n cc - | E—] n Lo « =] d. Future visit to you 7 | E— d Fo 72 | L— d Eo o. Referral to other M.D. 73 | C— . — 73 cc . | Ex 1. Future hospitalization 74 oc f. [Ea] 74 cc [} [Ea 9. Future surgery 75 oc 9 | Fa 7s 3 9 | -— h. Change in diet or drink 76 3 n ) 76 cc nh 3 i. Change in smoking 7 © i | En) id cc k Ea i. Bed rest nn 03 ) cc] 7 ; si i J L. Other change in activity 7 — k | E— 7 E— k EE I. Other action (PRINT) ee | 80 20 (8). Doctor's Siy Date of Visit. Dey Yoor Date form completed if different fromabove_____. ~~ © Manth Dey Yoor Sample of Completed Physician Visits Record Summary NATIONAL HEALTH SURVEY M.D. A CODE H LOC. DATE RET. DIAG. 525 210 04 01 12-18-1 02-05-2 OF OF PHYSICIAN VISIT RECORD CARD NO. ANSWERS BORN /98' SEX M' 9 VISITS IN 1960 SURVEY REFERRED BY COLLEGE OF OPTOMETRY 9 NONE 9 NONE 9 > © w Qo oO © m= © oo] Q = La] < = ad 99999 ABCDE ANGINAL SYNDROME 210 210 01 01 03-30-2 05-03-2 OF OF 1 CHEST PAIN 1 CHEST PAIN 1 1 133 ABC 39 919 DE AB Qe © = oo] Q = -. « = ol ANGINAL SYNDROME 1 ANGINA 1 CHEST PAIN 1 3 132 ABC 29 911 DE AB Qo o = lo] Q o —- << = EY ANGINAL SYNDROME 210 210 210 210 01 03 03 03 06-14-2 07-30-2 07-30-2 08-29-2 OF OF OF OF 2 NONE 1 CHEST PAIN 1 2 2 2229 919 ABCDE AB Qo Oo = ol [~] Im © jy “© = = ANGINAL SYNDROME 1 ANGINA 1 CHEST PAIN 2 3 2 2229 91991999919°9 ABCDE ABCDEFGHIJKL ARTERIO SCLEROTIC HEART DISEASE ANGINAL SYNDROME 1 CHEST PAIN 1 ANGINA 1 3 1 1229 9 ABCDE © Q © © oS © 199 AB 4 OSTEOARTHRITIS LUMBAR SPINE 2 NONE . 2 NONE 22229 9999199999999 ABCDE ABCDEFGHIJKL ANGINA PECTORIS 1 CHEST PAIN 1 CHEST PAIN 1 3 21 229 9 ABCDE »> © w © Q © oH Bm © ny © Q © © — © © = © = © 52 — 000 —m Recode number 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 APPENDIX II DIAGNOSTIC RECODE? Title Tuberculosis (active) (inactive), all sites----. Other chronic infective and parasitic diseases------=---c-mmmmmem ooo Malignant neoplasms--=-==-==-=-===o=ccooouu Benign and unspecified neoplasms----------- Hay fever, without asthma-----=------c----- Asthma (with or without hay fever) (bronchial) (not otherwise specified)--=----==ccmomeuoon- Other allergic disorders not elsewhere classifiable-----------cmmmmmmmm eee Diseases of the thyroid gland--------------- Diabetes (mellitus)----=-==---------ccoco--- Anemia and other diseases of the blood and blood-forming organs, 3 mo.+----=-----=--- Vascular lesions of the central nervous SY SEEM = === m= mmm mmm meee eee oon Headache and migraine, chronic------------- Specified mental disorders, not elsewhere classifiable---==----=-ccccmmmme moo I11-defined mental and nervous trouble, not elsewhere classifiable, 3 mo, += ----==------ Diseases of the.heart, not elsewhere classifiable (chronic rheumatic) (arterio- sclerotic) (hypertensive)-----------ccoao-un Hypertension, not elsewhere classifiable, with- out heart involvement----------cceoocomoono International Classification of Diseases (Seventh Revision) inclusions as modified by NCHS 001-007, 008, 009-S, 010-012, 014-019 020-029, 031-034, 036-039; 040-056, 057 excl. 057.1; 058-064; 070-074; 080, 082, 083.0, 084-096.8; 096.X, 100-138 140-205 210-239 240 ‘245 (242-244, 246-S not used) 250-254 260 290-299 330-334 354, 791 083.1, 083.2, 300-324, excl, 318.3 327-S (318.3, 326.3, 326.4, 790.0, 790.2) 410-443 (782.1, 782.2, 782.4) 444-447 3 : 5 : The recode categories 1-46 are the same as those used in the Recode 3 for the Health Interview Survey. Recodes 48-50 were included in Re- code 47 in the original recode. 53 Recode number 17 18 19 20 21 22 23 24 25 26 27 28 20 30 31 32 33 34 35 36 37 38 39 40 41 42 VALICOBE VEIIG = msm ms uo sm mnt msm i om HETTIOLTINOIAG. 5 wooo sn im soso i oa Rheumatic fever; arteriosclerosis, not elsewhere classifiable; other chronic diseases of the circulatory system--------=-=------- Chronic sinusitig§==-=====cccmmmommmcaeeea Chronic bronchitis -=--===ceccmmmmmeeaaa Other chronic diseases of the respiratory ES EE tt EE EE Ulcer of stomach and duodenum-------=--=--- Hernia (abdominal cavity)------=-=c-cooo-o- Diseases of the gallbladder, chronic--------- Other chronic diseases of the digestive SYSLEM- == mmmmmmmmmc mmm mmm mmm mem mmm mem Disorders of menstruation-----------e-c--- Menopausal symptoms, except psychosis----- Urinary calculi; prostate disorders; other chronic genitourinary conditions------------ Chronic skin diseases-------==-coccecmauan Arthritis and chronic rheumatism----------- Other chronic musculoskeletal disorders----- Fractures, 3 mo.+, no residual specified----- it ry . sst Other injuries, 3mo.+, no residual specified -- Severe visual impairment--------ccccceoo-- Other visual impairment------==-=nomceeuuu Hearing impairments----------cececmmeuaa- Speech defectS--==-=--cmmommmomm eee Paralysis =------ccmmm meee Absence, fingers, toes, only---------------- Absence, major extremities--------c-oeou-- Impairments (except paralysis and absence), back or spine-------ce-meeemmmmee— em International Classification of Diseases (Seventh Revision) inclusions as modified by NCHS 460, 462 461 400-402, 403-S; 450-456, 463-468; 782.0, 782.3, 782.5-782.8, 782.X 513 502 510.0, 512, 514-517, 523-526; (480-493, 3 mo.+; 511, 518-522, 527, 783, if 3 mo.+) 540-542 560, 561 584-586 Any in 530-539, 543-545, 551-553, 570, 572-583, 587, 784.5-784.7, 785.0-785.3, 785.5, 785.7-785.X (784.0- 784.4, 784.8, 785.4, 785.6) 634 635 602, 604, 610-612; 620, 592, 594, 623; 591, 593, 600, 601, 603, 605-609, 613-617, 621, 624-633, 636, 637, 786, 789, if 3 mo.+ 690-716, - if 3 mo.+ except 694 725 (720-724 not used), 726.0, 726.1, 726.3, 727 730.1, 730.2, 744; - [731-733, 735, 738, 740-743, if 3 mo.+] 800.9-829.9 850.9-999.9* * Unspecified residuals, 3 mo.+, of dislocations, sprains, strains, are coded to X70.9-X79.9, by site. 54 Recode International Classification of Diseases number Title (Seventh Revision) inclusions as modified by NCHS 43 Impairments (except paralysis and absence), upper extremities and shoulders------------ 44 Impairments (except paralysis and absence), lower extremities and hips with any other site-- 45 Impairments (except paralysis and absence), multiple not elsewhere classifiable, and ill- defined, limbs, back, trunk----------------- 46 Other impairments----------=------------- 47 Other chronic conditions, not impairmentsand All other ICD code numbers which may be chronic not in recodes 48-50--------=-----commooo- conditions 48 Chronic diseases of eye, not impairments---- 370-388, if 3 mo.+; 753.0, pt. 753.1 49 Chronic diseases of ear, not impairments---- 390-396, if 3 mo.+ 50 Chronic organic nervous system conditions--- 340-350, 353, 355-369; pt. 753.1; pts. 780, 781, if 3mo.+ —000—— 55 APPENDIX lI SAMPLING DESIGN Introduction The sampling design consists of the selection of the sample of respondents, the allocation of the sample to interviewers, and the procedures used in calculating the estimates. Family Account Numbers and Medical Record Numbers at KFHP The main devices used in selecting the samples were the Family and Medical Record Numbers, which are now discussed. . On enrollment in KFHP, a new subscriber is as- signed a seven-digit number called the Family Account Number. There is one Family Account Number for the subscriber and the covered members of his family. For the subscriber the Family Account Number is also his Medical Record Number. Other members of his family are also assigned individual Medical Record Numbers which are in sequence after the Family Ac- count Number for all members covered when the sub- scriber joins and. which are the next higher numbers for those joining the covered membership—e.g., new- born infants at a later time. Thus, the Family Account Numbers are the Medical Record Numbers of the sub- scriber, and each member of KFHP, subscriber or not, has his own seven-digit Medical Record Number. The records for each person include both his Family Ac- count Number and his Medical Record Number, Population For purposes of this study the population consisted of all members of KFHP that met the following require- ments: (1) They were members during the 6-month period January through June 1960 and during the study itself. (2) They were at least 17 years of age at the date of interview. (3) They were not members of the Culinary Workers Union. 56 Selection and Assignment to Interviewers of the Interview Sample Introduction.—The two main samples in the study were the PVR Sample, for which medical records were prepared, and the Interview Sample, a subsample of the PVR Sample for which interviews and comparisons with the medical records were made. In this section the selection of these two samples, the weights of the elements of the Interview Sample, the interviewers' assignments, and the dates of beginning and terminating interviews are discussed. Preliminary Sample. —The population from which the Preliminary Sample was drawn consisted of all sub- scribers to KFHP and the covered members of their families 15 years of age and over who were members of KFHP during the 6 months January through June 1960 and who were not members of the Culinary Workers Union. The Preliminary Sample consisted of those with terminal digits 2, 5, or 7, and thus included approxi- mately 30 percent of the population. Physician Visit Recora (PVR) Sample—allocation to five waves or sequences.—Using the data on number of visits to SCPMG of each person in the Preliminary Sample for the 6 months January through June 1960, the Preliminary Sample was classified into two strata— those who had made 0, 1, 2, 3, or 4 visits to SCPMG during the 6-month period and those who had made 5 or more visits during that period. The PVR Sample consisted of an approximately 10-percent sample from the first stratum and an ap- proximately 20-percent sample chosen from the second stratum, selected as indicated in tables I and II. Table I. Sampling procedure for those making 0 through 4 visits during January-June 1960 Of those whose seventh digit (Medical Record Number) is---=-c-cmeca--- 0123456789 Include in the sample those whose fifth digit 18-ccccmcmccccccnccene 2468075913 Table II. Sampling procedure for those making 5 or more visits during January-June 1960 0f those whose seventh digit (Medical Record Number) ig--=c-ececc-c-- 0123456789 Include in the sample those whose fifth digit Loemmnneamnnnmansameans 2468075913 6035748291 tor convenience in initiating the PVR record keeping and in the interviewing, the sample was ran- domly allocated to five waves or sequences of approxi- mately equal sizes (see table III). Record keeping be- gan at 3-week intervals for the five waves. The PVR Sample thus selected consisted ot 4,922 names. These were allocated to five sequences or waves according to the sixth digits of the Medical Record Numbers as stated in table III. The staggered beginnings of the waves facilitated both the operations of record keeping at SCPMG and the interviewing by the Bureau of the Census later on. from the combination of the 1960 visit strata and the study year visit strata, Table IV. Sampling ratios and weights for in- terview sample Number of visits Approximately Sanpling Weight January-June first 11 1960 months of study year 0-4 0 1 in 10 20 0-4 1 lin 3 6 0-4 2-5 1 in 2 4 0-4 6 and over All 5 and over 0 1 in 10 10 5 and over 1 and over All 1 Table III. Allocation of sample to sequences or waves Consists of all persons The date on The sequence | i, the PVR which PVR's having iden- sample having began to be A fioation sixth digit filled out for number (Medical the sequence or Record wave was-— Number) 1 2 or 5 October 15, 1961 2 l or 8 November 5, 1961 3 6 or 9 November 26, 1961 4 0 or 4 December 17, 1961 5 3 or 7 January 7, 1962 Interview Sample— determination of weights. — Ap- proximately 11 months after the beginning of each wave, the number of visits of gach person on the PVR Sample was tallied from the PVR's for that person. Using those data on number of visits, the Interview Sample was se- lected from the PVR Sample in accordance with tabie IV. Also, in table IV are given the weights resulting Allocation of the Intevview Sample among areas ana interviewers—dates of interviewing.— With minor modifications, the service area of the Kaiser Foun- dation Health Plan was divided into four areas, three of ‘which were in Los Angeles and the fourth which contained Fontana and nearby areas. The four areas are those of the present study. After the Interview Sample was selected for a given wave, the addresses of its members were located and the sample was thus distributed among the four areas. For each of the four areas, the Interview Sample was allocated at random among the three questionnaires. Because of problems of cost and administration, how- ever, interpenetrating samples were not used for inter- viewer assignments within all areas. In the three Los Angeles areas, the interviewers shifted from area to area in different waves. In the Fontana area, the in- terviewers were the same in all waves, One year after the beginning of the PVR record keeping for a wave, the PVR record keeping terminated. Interviewing of that wave then began and continued for 2 to 3 weeks afterwards. The only change from the original plans occurred in Waves 4 and S in order to avoid the possibly higher noninterview rates between Christmas and New Year. The dates are given in table V. 57 Table V. Scheduled and actual interviewing dates, by wave Scheduled dates Actual dates Wave Beginning Ending Beginning Ending! 1 October 22, 1962 November 10, 1962 As scheduled 2 November 5, 1962 November 24, 1962 As scheduled 3 November 26, 1962 December 15, 1962 As scheduled 4 December 17, 1962 January 5, 1963 December 12, 1962 December 22, 19622 5 January 7, 1963 January 26, 1963 January 3, 1963 January 16, 1963 In some cases, interviewing occurred after the stated ending date, but these were few in num- ber. 2The change in dates for Wave 4 was primarily to Christmas season. Final changes in the sample.—During data proc- essing, two changes were made in the sample to be tabulated. These were as follows: (1) (2) All persons under 17 years of age on the date of interview were eliminated. It had been decided earlier that only one per- son would be interviewed in any household. Consequently if any household had two members or more selected for the sample, all but one were eliminated from the Interview Sample, but the information for the sample person not eliminated was duplicated and in one instance triplicated. Interview Sample for Which PVR’s Were Not Used In any record-check study for which special records such as the PVR's are being prepared, there are always the possibilities that these special records are incom- plete or inaccurate or that the respondent has become aware of the study sufficiently to influence his reporting. Consequently a further sample, called Wave 6, was se- lected as follows: ‘ 58 1) 2 The Wage 6 Sample was selected from persons in the Preliminary Sample who had not been selected for the PVR Sample but who had as a sixth digit of their Medical Record Numbers either 0, 3, 4, or 7—i.e., the sixth digits corresponding to Waves 4 or S. A 10-percent sample was selected from those with 0, 3, 4, or 7 as the sixth digit of their Medical Record Numbers in accordance with table VII, Table VII. those reduce the amount of interviewing during the First-stage 10-percent sample from having sixth digits identifying se- quences or Waves 4 and 5 Of those whose seventh digit (Medical Record Number) ig-=--cccccccaa Include in the first stage sample those whose fifth digit is--- 0123456789 45903217638 @) (4) 000 The resulting sample, called the PC Sample, then consisted of a subsample of one in six of those selected in item 2 who had made 0 to 4 visits to SCPMG during January-June 1960 and a sample of one in three of those who had made 5 visits or more to SCPMG during January-June 1960. For the PC, or Wave 6 Sample, medical rec- ords (PC) were obtained by using the patient charts (PC) the study year. The persons in the PC Sample were not in the PVR Sample, and no indication of their being in the PC Sample could have reached the physicians and, through them, the patients, because physicians were not involved in the preparation of the medical records (PC). The medical records (PC) were then used to select an Interview Sample that consisted of all persons in the PC Sample who had made sat least one visit to SCPMG during the study year, and a sample of 1 in 10 of those was selected. Thus the weights for Wave 6 are 3, 6, 15, and 30. + I1 8 COAUVRRBNMENT DRINTING NARRIOE . 1072 €16_914 7a) Series 1. Series 2, Series 3, Series 4. Series 10. Series 11. Series 12. Series 13, Series 14. Series 20. Series 21, Series 22, VITAL AND HEALTH STATISTICS PUBLICATION SERIES Formerly Public Health Service Publication No. 1000 Programs and collection procedures.— Reports which describe the general programs of the National Center for Health Statistics and its offices and divisions, data collection methods used, definitions, and other material necessary for understanding the data. Data evaluation and methods research.— Studies of new statistical methodology including: experi- mental tests of new survey methods, studies of vital statistics collection methods, new analytical techniques, objective evaluations of reliability of collected data, contributions to statistical theory. Analytical studies.—Reports presenting analytical or interpretive studies based on vital and health statistics, carrying the analysis further than the expository types of reports in the other series, Documents and committee reports.—Final reports of major committees concerned with vital and health statistics, and documents such as recommended model vital registration laws and revised birth and death certificates, Data from the Health Interview Survev.—Statistics on illness, accidental injuries, disability, use of hospital, medicah, dental, and other services, and other health-related topics, based on data collected in a continuing national household interview survey, Data from the Health Examination Survey.—Data from direct examination, testing, and measure- ment of national samples of the civilian, noninstitutional population provide the basis for two types of reports: (1) estimates of the medically defined prevalence of specific diseases in the United States and the distributions of the population with respect to physical, physiological, and psycho- logical characteristics; and (2) analysis of relationships among the various measurements without reference to an explicit finite universe of persons. Data from the Institutional Population Surveys — Statistics relating to the health characteristics of persons in institutions, and their medical, nursing, and personal care received, based on national samples of establishments providing these services and samples of the residents or patients, Data from the Hospital Discharge Survey.—Statistics relating to discharged patients in short-stay hospitals, based on a sample of patient records in a national sample of hospitals. Data on health resources: manpower and facilities. —Statistics on the numbers, geographic distri- bution, and characteristics of health resources including physicians, dentists, nurses, other health occupations, hospitals, nursing homes, and outpatient facilities. Data on mortality.—Various statistics on mortality other than as included in regular annual or monthly reports—special analyses by cause of death, age, and other demographic variables, also geographic and time series analyses. Data on natality, marriage, and divorce.,—Various statistics on natality, marriage, and divorce other than as included in regular annual or monthly reports—special analyses by demographic variables, also geographic and time series analyses, studies of fertility. Data from the National Natality and Mortality Surveys,— Statistics on characteristics of births and deaths not available from the vital records, based on sample surveys stemming from these records, including such topics as mortality by socioeconomic class, hospital experience in the last year of life, medical care during pregnancy, health insurance coverage, etc. For a list of titles of reports published in these series, write to: Office of Information National Center for Health Statistics Public Health Service, HSMHA Rockville, Md, 20852 DHEW Publication No. (HSM) 73-1331 3 Series 2-No. 57 U.S. DEPARTMENT OF HEALTH, EDUCATION, AND WELFARE > 2 POSTAGE AND FEES PAID Public Health Service U.S. DEPARTMENT OF HEW a—— HEALTH SERVICES AND MENTAL HEALTH ADMINISTRATION SSA HEW 396 5600 Fishers Lane Rockville, Maryland 20852 OFFICIAL BUSINESS THIRD CLASS Penalty for Private Use, $300 BLK. RT. U.C. BERKELEY LIBRARIES LE (021206070