J JL :b f 431 $ Digitized by the Internet Archive in 2019 with funding from Princeton Theological Seminary Library https://archive.org/details/studyofamericaniOObrig A STUDY OF AMERICAN INTELLIGENCE A STUDY OF AMERICAN INTELLIGENCE BY CARL C. BRIGHAM, Ph.D. ASSISTANT PROFESSOR OF PSYCHOLOGY IN PRINCETON UNIVERSITY A FOREWORD By Robert M. Yerkes, Ph.D. CHAIRMAN RESEARCH INFORMATION SERVICE: NATIONAL RESEARCH COUNCIL PRINCETON Princeton University Press LONDON: HUMPHREY MILFORD OXFORD UNIVERSITY PRESS 1923 COPYRIGHT 1922 BY CARL C. BRIGHAM PRINTED AT THE PRINCETON UNIVERSITY PRESS, PRINCETON, N. J. FOREWORD Two extraordinarily important tasks confront onr nation: the protection and improvement of the moral, mental and physical quality of its people and the re-shaping of its in¬ dustrial system so that it shall promote justice and encour¬ age creative and productive workmanship. I have been asked to write this Foreword because of my official con- nection, as chief of the Division of Psychology, Office of the Surgeon General of the Army, with psychological ex¬ amining during the war, but I have consented to write it because of my intense interest in the practical problems of immigration and my conviction that the psychological data obtained in the army have important bearing on some of them. When in April, 1917,1 visited Canada to learn what use our neighbors were making of psychological principles and methods in their military activities, I found Mr. Carl C. Brigham attached as psychologist to the Military Hospitals Commission. With him as my guide, I spent several hours in interviewing military and civil officers and in discussing our mutual problems and needs. The valuable information which Mr. Brigham helped me to secure and his advice contributed substantially to the report which I later pre¬ sented to my professional colleagues at home, and to rep¬ resentatives of the United States army. In October, 1917, our friend, eager for larger opportuni¬ ties for professional service than the Canadian army prom¬ ised, accepted appointment in the Sanitary Corps of the United States army for psychological service. He aided v VI AMERICAN INTELLIGENCE efficiently in the trials of methods of examining at Camp Dix, New Jersey, and he was then ordered to the office of the Surgeon General in Washington to help with the revis¬ ion of tests and the preparation of new methods. Thus he became thoroughly familiar with the procedures and results of psychological examining in the army, while at the same time contributing generously of ideas, labor and enthu¬ siasm. With deep satisfaction I use this opportunity to men¬ tion Mr. Brigham’s national service and his exceptional fitness to study and to discuss the relations of army meas¬ urements of intelligence to nativity and residence. It appears that Mr. Charles W. Gould, a clear, vigorous, fearless thinker on problems of race characteristics, amal¬ gamation of peoples and immigration, raised perplexing questions which drove Mr. Brigham to his careful and critical re-examination, analysis, and discussion of army data concerning the relations of intelligence to nativity and length of residence in the United States. In a recently published book, America , A Family Matter , to which this little book is a companion volume, Mr. Gould has pointed the lessons of history for our nation and has argued strongly for pure-bred races. For the observational data which Mr. Brigham used in preparing this book we are indebted to the competent and devoted company of psychologists which during the war labored in camp and laboratory on the preparation of meth¬ ods, the conduct of examinations, and the application of re¬ sults. But the fruits of the labors of these many psychol¬ ogists might have been lost to the world had it not been for the insight, zeal, and industry of Carl R. Brown, Mark A. May and Edwin G. Boring, who evolved methods of statis¬ tical treatment, applied them and prepared the resulting materials for publication. Mr. Brigham has rendered a notable service to psychol- FOREWORD Vll ogy, to sociology, and above all to our law-makers by carefully re-examining and re-presenting with illuminating discussion the data relative to intelligence and nativity first published in the official report of psychological exam¬ ining in the United States army. Far from belittling or casting doubt on the general reliability of the results con¬ tained in the report, he has essentially confirmed the major findings in the field of his special inquiry and has adduced new evidences of the trustworthiness and scientific value of the statistical methods used by military psychologists. His task has been arduous and difficult, involving an im¬ mense amount of tedious labor for mathematical calcula¬ tions and critical study of results. The volume which is the outcome of Mr. Brigham’s inquiry, and which I now have the responsibility and satisfaction of recommending, is sub¬ stantial as to fact and important in its practical implica¬ tions. It is not light or easy reading but it is better worth re-reading and reflective pondering than any explicit dis¬ cussion of immigration which I happen to know. The author presents not theories or opinions but facts. It be¬ hooves us to consider their reliability and their meaning, for no one of us as a citizen can afford to ignore the menace of race deterioration or the evident relations of immigra¬ tion to national progress and welfare. Robert M. Yerkes. Washington, D. C. June 1922 CONTENTS PAGE Foreword v Introduction xix Part I : The Army Tests Section 1 examination alpha 3 Section 2 examination beta 32 Section 3 the individual examinations 54 Section 4 reliability of the measures 59 Part II : Statistical Analysis of the Army Test Results Section 1 the principal sample 75 Section 2 analysis of the main groups of the PRINCIPAL SAMPLE 77 Section 3 analysis of the white draft into FOREIGN AND NATIVE BORN 84 Section 4 analysis of the foreign born white DRAFT INTO YEARS OF RESIDENCE GROUPS 88 Section 5 analysis of immigration to the UNITED STATES 112 Section 6 analysis of the foreign born white DRAFT BY COUNTRY OF BIRTH 118 Section 7 reliability of the results 154 Section 8 the race hypothesis 157 Section 9 re-examination of previous con¬ clusions IN THE LIGHT OF THE RACE HYPOTHESIS 177 Section 10 comparison of our results with the CONCLUSIONS OF OTHER WRITERS ON THE SUBJECT 182 197 Conclusions ix PLATES PLATE I II III IV V VI VII VIII IX X XI XII XIII XIV XV Alpha test 1 Alpha test 2 Alpha test 3 Alpha test 4 Alpha test 5 Alpha test 6 Alpha test 7 Alpha test 8 Beta test 1 Beta test 2 Beta test 3 Beta test 4 Beta test 5 Beta test 6 Beta test 7 : ORAL DIRECTIONS I ARITHMETICAL REASONING : PRACTICAL JUDGMENT : SYNONYM-ANTONYM : DISARRANGED SENTENCES : NUMBER SERIES COMPLETION : ANALOGIES : INFORMATION : MAZE : CUBE ANALYSIS : X-O SERIES : DIGIT-SYMBOL : NUMBER CHECKING : PICTURE COMPLETION I GEOMETRICAL CONSTRUCTION PAGE 5 9 13 18 21 24 26 29 35 38 41 44 47 50 53 x FIGURES FIGURE PAGE 1. Distribution of scores of the oral directions test 6 2. The Gaussian normal distribution 7 3. Distribution of scores of the arithmetical reason¬ ing test 10 4. Distribution of scores of the practical judgment test 14 5. Distribution of scores of the synonym-antonym test 17 6. Distribution of scores of the disarranged sentence test 20 7. Distribution of scores of the number series com¬ pletion test 23 8. Distribution of scores of the analogies test 25 9. Distribution of scores of the information test 28 10. Black-board chart for demonstrating the maze test 33 11. Distribution of scores of the maze test 34 12. Black-board chart for demonstrating the cube analysis test 36 13. Distribution of scores of the cube analysis test 37 14. Black-board chart for demonstrating the X-Q series test 39 15. Distribution of scores of the X-0 series test 40 16. Black-board chart for demonstrating the digit- o o symbol test 42 xi xii FIGURES FIGURE PAGE 17. Distribution of scores of the digit-symbol test 43 18. Black-board chart for demonstrating the number checking test 45 19. Distribution of scores of the number checking test 46 20. Black-board chart for demonstrating the picture completion test 48 21. Distribution of scores of the picture completion test 49 22. Black-board chart for demonstrating the geomet¬ rical construction test 51 23. Distribution of scores of the geometrical con¬ struction test 52 24. The normal distribution curve 59 25. A skewed distribution curve 60 26. Examination alpha as independent of education 65 27. Distribution of intelligence scores according to rank 66 28. Success in Officers’ Training Camps as predicted by examination alpha 67 29. Comparison of army test records with various independent criteria 69 30. Success in civil occupations compared with army test records 70 31. Distributions of scores of the white officers, white draft, and negro draft on the combined scale 81 32. Distributions of scores of the native born and for¬ eign born white draft on the combined scale 87 33. Apparently increasing average intelligence with increasing years of residence 94 34. Distributions of the alpha scores of three groups 108 35. Analysis of immigration by countries 114 36. Relative standing of the nativity groups accord¬ ing to their average intelligence 124 FIGURES Xlll FIGURE PAGE 37. Relative standing of the nativity groups in the proportions of A and B men, and D, D— and E men 146 38. The proportion of each nativity group obtaining scores at or above the average of the white officers 149 39. The proportion of each nativity group at or below the average of the negro draft 151 40. The proportion of each nativity group testing be¬ low the approximate “mental age” of eight 153 41. Ai nalysis of immigration to the United States ac- cording to the estimated amount of Nordic, Mediterranean and Alpine blood 164 42. Volume of immigration by decades 166 43. The distributions of the intelligence scores of the Nordic, Mediterranean and Alpine groups 170 44. The distributions of the intelligence scores of the English speaking Nordic and the non-English speaking Nordic groups 173 45. The distributions of the intelligence scores of the non-English speaking Nordic group and the combined Mediterranean and Alpine groups 175 46. The decline of intelligence with each succeeding period of immigration 198 47. The constituent elements of American intelligence 200 TABLES PAGE 1. Distribution of the intelligence scores of the main groups of the principal sample on the combined scale 80 2. Analysis of the white draft into foreign born and native born groups 86 3. Analysis of the foreign born white draft by years of residence in the United States 90 4. Comparison of the average scores on the combined scale of the five years of residence groups of the foreign born white draft 91 5. Comparison of the average scores on the combined scale of the native born white draft with the five years of residence groups of the foreign born white draft 92 6. Per cent, that emigration was of immigration for 15 countries since 1908 98-99 7. Distribution of alpha scores of five groups 106 8. Per cent, of total immigration coming from var¬ ious countries during periods roughly correspond¬ ing to the five years of residence groups 113 9. Analysis of the foreign born white draft by coun¬ try of birth: actual distributions 120-121 10. Analysis of the foreign born white draft by coun¬ try of birth: percentage distributions 122-123 11. Differences between England and other countries 126 12. Differences between Scotland and other countries 127 13. Differences between Holland and other countries 128 14. Differences between Germany and other countries 129 15. Differences between the United States and other countries 130 16. Differences between Denmark and other countries 131 17. Differences between Canada and other countries 132 xiv TABLES xv PAGE 18. Differences between Sweden and other countries 133 19. Differences between Norway and other countries 134 20. Differences between Belgium and other countries 135 21. Differences between Ireland and other countries 136 22. Differences between Austria and other countries 137 23. Differences between Turkey and other countries 138 24. Differences between Greece and other countries 139 25. Differences between Russia and other countries 140 26. Differences between Italy and other countries 141 27. Differences between Poland and other countries 142 28. Per cent, of each nativity group in the A and B groups 144 29. Per cent, of each nativity group in the D, D — and E groups 145 30. Per cent, of each nativity group at or above the average of the white officers 148 31. Per cent, of each nativity group at or below the average of the negro draft 150 32. Per cent, of each nativity group below the approxi¬ mate “mental age” of eight 152 33. Tentative estimates of the proportion of Nordic, Alpine and Mediterranean blood in each of the European countries 159 34. Arrivals of alien passengers and immigrants, 1820 to 1920 160-161 35. Estimate of the amount of Nordic, Mediterranean and Alpine blood coming to this country from Europe in each decade since 1840 163 36. Analysis of the foreign born white draft by races 169 37. Analysis of the Nordic sample into an English speaking Nordic group and a non-English speaking Nordic group 172 38. Population of the United States in 1920 203 ACKNOWLEDGMENTS This study is a continuation of the work of the small group of psychologists who carried on the difficult task of analyzing the data from the army psychological examina¬ tions in the office of the Surgeon General of the Army. My presentation contains nothing new in methodology and is merely an extension of lines of investigation suggested by this group of workers. It rests on the foundations which they built. I wish to make especial acknowledgment to Colonel Robert M. Yerkes, who has read the manuscript several times in its various stages of preparation and has given many helpful suggestions. Professor Carl R. Brown of the University of Michigan, formerly of the Surgeon General’s staff, assisted me when I first began to use the combined scale, and subsequently read Sections 1 to 7 of Part II in manuscript. Professor E. G. Boring of Harvard University read Sections 1 to 7 of Part II, and gave me invaluable assistance, especially in the treatment of Section 4. Pro¬ fessor Mark A. May of Syracuse University, Professors Edwin G. Conklin and Howard C. Warren of Princeton University read Sections 1 to 7 of Part II, and suggested many important changes. Without the assistance of all of these gentlemen I could not have carried through the task. Mr. Charles W. Gould suggested this continuation of the army investigations in the first instance, has sponsored the work throughout, has read and re-read all of the manu¬ script at every stage of its preparation, and is mainly responsible for the whole work. In my treatment of the race hypothesis I have relied on his judgment and on two books, Mr. Madison Grant’s Passing of the Great Race , and XVII xviii AMERICAN INTELLIGENCE Professor William Z. Ripley’s Races of Europe. These three gentlemen cannot be held responsible however for my percentage analysis of the present racial constitution of European countries, an analysis which I made as a novice in the field of anthropology, and for which I offer further apologies in the text of Section 8. Mr. David M. Maynard and Mr. Charles H. Helliwell, two of my undergraduate students, assisted me faithfully in carrying through the laborious statistical calculations involved in using the combined scale. Carl C. Brigham Princeton, N. J. September 1922 INTRODUCTION The question of the differences that may exist between the various races of man, or between various sub-species of the same race, or between political aggregations of men in nationality groups may easily become the subject of the most acrimonious discussion. The anthropologists of France and Germany, shortly after the close of the Franco- Prussian war, fought another national war on a small scale. It is difficult to keep racial hatreds and antipathies out of the most scholarly investigations in this field. The debate becomes especially bitter when mental traits are discussed. No one can become very indignant on finding his race classified by its skull dimensions, stature, or hair color, but let a person discover the statement that his race is unintelligent or emotionally unstable, and he is immedi¬ ately ready to do battle. Until recent years we have had no methods available for measuring mental traits scientifically, so that the lit¬ erature on race differences consists largely of opinions of students who are very apt to become biased, when, leaving the solid realm of physical measurements, they enter the more intangible field of estimating mental capacity. Gradually, however, various investigators using more or less refined psychological measurements commenced to as¬ semble a body of data that will some day reach respectable proportions. The status of the psychological investigations of race differences up to 1910 has been admirably sum¬ marized by Woodworth. 1 Since 1910, we have witnessed 1 R. S. ’Woodworth. Racial Differences in Mental Traits, Science, New Series, Vol. 31, pp. 171-1S6. XIX XX AMERICAN INTELLIGENCE in this country a remarkable development in methods of intelligence testing, and these methods have been applied to the study of race differences. Scattered investigations report and compare the intelligence scores of children of white, negro, or Indian parentage, and sometimes the scores of various nationality or nativity groups. The re¬ sults of these investigations are, however, almost impos¬ sible to correlate, for they have been made by different methods, by different measuring scales, on children of a wide variety of chronological ages, and above all, on com¬ paratively small groups of subjects, so that conclusions based on the studies have no high degree of reliability. For our purposes in this country, the army mental tests give us an opportunity for a national inventory of our own mental capacity, and the mental capacity of those we have invited to live with us. We find reported in Memoir XV of the National Academy of Sciences 1 the intelligence scores of about 81,000 native born Americans, 12,000 foreign born individuals, and 23,000 negroes. From the standpoint of the numbers examined, we have here an investigation which, of course, surpasses in reliability all preceding in¬ vestigations, assembled and correlated, a hundred fold. These army data constitute the first really significant con¬ tribution to the study of race differences in mental traits. They give us a scientific basis for our conclusions. When we consider the history of man during the half million years which have probably elapsed since the time of the erect primate, Pithecanthropus , the temporary polit¬ ical organizations, such as Greece, Rome, and our modern national groups, become of minor importance compared with the movements of races and peoples that have oc¬ curred. The tremendous expansion of the Alpine race at the end of the Neolithic and the beginning of the Bronze 1 Psychological Examining in the United States Army. Edited by Robert M. Yerkes. Washington: Government Printing Office, 1921, pp. 890. INTRODUCTION xxi Period, the submergence of this race by the Nordics in the 2000 years preceding the Christian era, and the subsequent peaceful re-conquest of Eastern Europe by the Alpine Slavs from the Dark Ages on, represent an historical movement in comparison with which the Great World War of 1914 resembles a petty family squabble. If the history of the United States could be written in terms of the movements of European peoples to this con¬ tinent, the first stage represents a Nordic immigration, for New England in Colonial times was populated by an almost pure Nordic type. There followed then a period of Nordic expansion. The next great movement consisted of the mi¬ grations of Western European Mediterraneans and Alpines from Ireland and Germany, a movement which started t/ about 1840, and which had practically stopped by 1890. Since there is a considerable proportion of Nordic blood in Ireland and Germanv, we should not regard the original %j 7 O O Nordic immigration as a movement which stopped sud¬ denly, but merely as having dwindled to two-fifths or one- half of the total racial stock coming here between 1840 and 1890. The third and last great movement consisted of mi¬ grations of the Alpine Slav and the Southern European Mediterraneans to this continent, a movement that started about 1890, and which has not yet ceased. Running parallel with the movements of these European peoples, we have the most sinister development in the history of this con¬ tinent, the importation of the negro. The army mental tests enable us to analyze the elements entering into American intelligence. The intelligence test records of the native born, the foreign born, and the negro are at our disposal. The records deserve the most serious study. But before considering the results of the army tests, a person should be well informed concerning the nature of the tests, and the manner in which they were constructed. AMERICAN INTELLIGENCE xxii The army psychological tests included three types of examination: (1) Group examination alpha, which included eight dif¬ ferent sorts of tests, most of which involved the ability to read English. (2) Group examination beta, which included seven dif¬ ferent sorts of tests, none of which involved the ability either to read English or to understand spoken English, the tests consisting of pictures, designs, etc., and being given by instructions in pantomime. (3) Individual examinations of two types: (a) Those involving the use of English, the Stanford revision of the Binet-Simon scale and the point scale, and ( b ) Those involving no English, consisting of con¬ struction puzzles, etc., the instructions being given by gestures,—the “performance scale.” When a detachment reported for psychological examina¬ tion, the first step was that of separating the English speaking and literate from the non-English speaking or illiterate. Those who were both English speaking and liter¬ ate were given examination alpha. All others were sent to beta. At the close of examination alpha, all men who had made low scores were sent to beta. After examination beta had been given, the examiners tried to recall for individual examinations all men who had made a low score in beta. In the rush of examining it was impossible to recall all men for individual examinations who should have been given special examinations, and some men were graded on alpha who should have been graded on beta, and vice versa, but INTRODUCTION xxm most men were properly graded by the rough methods in use. In each one of the examinations the range of scores was so great that most men had an opportunity to score. The great contribution of the committee that first de¬ vised the army examining methods and of the men who subsequently developed additional methods in the army consisted of creating and standardizing group examinations alpha and beta. The methods of individual examining were already in existence, the Stanford-Binet scale being an elab¬ oration of Binet’s “mental age” scale, and the tests of the performance scale having been more or less completely worked out by other investigators. The task of examining men in large groups was first carried through successfully in the army. Before the war, many psychologists would have scoffed at the notion of examining two or three hun¬ dred men at once by giving them booklets containing dif¬ ferent sorts of tests, but the large group examinations be¬ came matters of daily routine. Group tests have subsequent- lv been tried out in schools and industries with excellent */ results from the standpoint of test administration. Indeed, when the army alpha examination was given at Ohio State University in October, 1919, practically the entire student body, 6000 in number, was tested by five examiners in eight hours. In the service, it was found that one examiner could control a group of 200 men with ease. The alpha in¬ structions were read by the examiner, and the men ordered to start and stop at the proper time. Examination beta was more difficult to administer, and was given to smaller groups. The statistical methods of treating the results of the army tests used in this study are rather intricate, but the principles involved are easily understood. At the outset we must frankly admit that there were minor errors in the three types of examinations given. We can not correct the XXIV AMERICAN INTELLIGENCE type of tests that were used, but we can correct the method of scoring them. Most of the difficulties of scoring arise from the fact that different types of measuring scales were used. During the war, the different scales were converted into one general scale of letter grades (A, B, C + , C, C—, D and D—). This method was rough, and although it an¬ swered the purposes of the army at the time, it can not be used in any scientific interpretation of the results. Examination alpha was scored by finding the score on each of the eight tests, adding to get a total, and then con¬ verting the total into a letter grade. Beta was similarly scored. It is apparent that some tests in alpha might be more difficult than others, that some tests in beta might be easier than any test in alpha, and that variations might have occurred which it was impossible to predict at the time the examinations were made. Recognizing these facts, then, the army statisticians worked out another method of scoring the results, which eliminates all of these sources of error. This method is known as the combined scale , a theoretical intelligence scale running from 0 to 25, into which the alpha, beta and individual examination scores may be converted, so that we finally have one measure¬ ment instead of three. Psychological measurements involve much more than creating tests and giving tests. After all the results are in, we still have the problem of interpreting the results, and this interpretation is largely a statistical problem. Too much credit can not be given to the staff of the Psycholog¬ ical Division of the Surgeon General’s Office, who con¬ tinued in the service long after the war was over, patiently studying and analyzing the results. The combined scale was very largely the work of two young psychologists, Carl R. Brown and Mark A. May, and their work on this prob¬ lem, reported in Chapter 2, Part 3 of Memoir XV, is with- INTRODUCTION XXV out doubt the greatest contribution that has yet been made to the statistical phases of the science of mental measure¬ ment. The theory underlying the combined scale is simply that of regarding each test of alpha and beta as a separate measuring scale. One group of individuals including 1047 men born in English speaking countries, was examined on alpha, re-examined on beta, and if possible, examined again on the Stanford-Binet scale. This group of 1047 cases con¬ stituted the basis on which a method of combining the sep¬ arate tests into a combined scale was empirically evolved. From now on in the course of our study of the army test records, we must regard alpha and beta as two booklets containing, in all, fifteen different measuring scales of in¬ telligence. The first step in the study consists of under¬ standing the nature of each of the fifteen scales. In Part I, the fifteen tests have been reproduced (Plates I to XV), and the actual records of the 1047 men shown in each in¬ stance, so that the reader may see exactly how the tests worked. PART I THE ARMY TESTS SECTION I EXAMINATION ALPHA Alpha Test 1. Oral Directions The first test in alpha consisted of a series of commands or directions which were to be executed quickly. The in¬ structions, with the incidental commands about stopping and starting eliminated, are reproduced below. One may read the instructions for each item to himself slowly and turn the page to Plate I to test his own ability to execute the commands. Item 1. Item 2. Item 3. Item 4. Item 5. Instructions : Oral Directions (Form 8) Time limit : 5 seconds. 4 ‘Make a figure 2 in the second circle and also a cross in the third circle.” Time limit : 5 seconds. “Draw a line from circle 1 to circle 4 that will pass below circle 2 and above circle 3.” Time limit : 10 seconds. “Make a figure 1 in the space which is in the square but not in the triangle, and also make a cross in the space which is in the triangle and in the square.” Time limit : 10 seconds. “Make a figure 2 in the space which is in the circle but not in the triangle or square, and also make a figure 3 in the space which is in the tri¬ angle and circle, but not in the square.” Time limit : 10 seconds. “If taps sound in the evening, then put a cross 3 AMERICAN INTELLIGENCE 4 Item 6. Item 7. Item 8. Item 9. in the first circle; if not, draw a line under the word NO.” Time limit : 10 seconds. “Put in the first circle the right answer to the question: ‘How many months has a year?’ In the second circle do nothing, but in the fifth cir¬ cle put any number that is a wrong answer to the question that you have just answered cor¬ rectly.” Time limit : 10 seconds. “Cross old the letter just after F and also draw a line under the second letter after I.” Time limit : 10 seconds. “Make in the first circle the last letter of the first word; in the second circle the middle letter of the second word and in the third circle the^r^ letter of the third word.” Time limit : 15 seconds. “Cross out each number that is more than 50 but less than GO.” Item 10. Time limit : 15 seconds. “Put a 4 or a 5 in each of the two largest parts and any number between 6 and 9 in the part next in size to the smallest part.” Item 11. Time limit : 25 seconds. “Draw a line through every odd number that is not in a square, and also through every odd number that is in a square with a letter.” Item 12. Time limit : 10 seconds. “If 4 is more than 2, then cross out the number 3 unless 3 is more than 5, in which case draw a line under the number 4.” J-.Tl.Jl 1ooooo 5 ooo Yes No e OOOOO 7 ABGDEFGH1JKLMNOP 8 OOO MILITARY GUN CAMP 123456789 Plate I. Alpha Test 1 : Oral Directions. 5 6 AMERICAN INTELLIGENCE In scoring the papers one point was given for each cor¬ rect response. The group of 1047 individuals born in Eng¬ lish speaking countries obtained the following scores: Total score of test No. 0 1 2 3 4 5 6 7 8 9 10 11 12 Number who made each score ... 73 78 93 116 100 121 131 94 82 67 52 28 lg These scores are shown graphically in Figure 1, the horizontal direction indicating the total score from the lowest possible (0) to the highest possible (12), while Figure 1 . Distribution of scores of the Oral Directions test. (From p. 624, Memoir XV.) the vertical scale represents the number of cases getting each score, 72 at 0, 78 at 1, etc. For our purposes, we do not want a test that everyone can pass, for if everyone passed, no one would be graded. An ideal test would be one in which practically everyone could obtain some score and which very few could finish. AMERICAN INTELLIGENCE 7 Then all people would be measured. An ideal test would also show a distribution of responses grouped symmetric¬ ally about the average, for, as a general rule, all measures of individual differences in mental traits show a distribution similar to the normal probability or chance distribution. An ideal test would give the type of distribution shown in Figure 2. Figure 2. The Gaussian normal distribution. Examining Figure 1, one may see that in general the oral directions test gives a type of distribution which is approx¬ imately similar to the Gaussian normal distribution shown in Figure 2. Our distribution is limited slightly at the lower end, and it is easy to imagine that the introduction of two 8 AMERICAN INTELLIGENCE or three items easier than any in the present test would give us a step-down at the zero end of the scale similar to the one at the upper end. The oral directions test really gives an excellent score distribution. It is not a “speed” test in the popular sense, for the time limits for each item, while short, give ample time for following the directions. If a person understands the directions he can execute them easilv in the time al- «/ lowed. If the directions are not understood, an hour to execute them is no more generous than five seconds. In practice this test was useful in “acclimating” the recruit to the conditions of the examination. It was probably one of the poorest tests in alpha as a genuine test of intelligence, but it served its purpose as a “warming up” test. It is an adaptation of a type of test that has been used in psycho¬ logical laboratories for many years with rather mediocre results. Alpha Test 2.—Arithmetical Problems Time limit : 5 minutes Test 2 is more of a reasoning test than a measure of proficiency in the fundamental arithmetical operations. The first items really constitute a literacy test, for if a person can read, he can answer the questions correctly. The distribution of scores in this test is shown in Figure 3. The zero scores (66 in number) are probably due to the inclusion of illiterates in the group of 1047 cases. Disregard¬ ing the zero scores, the distribution is regular. This test illustrates admirably the principle of fixing a time limit such that very few people can answer all the items cor¬ rectly. The approximate rule adopted in fixing the time limit in the first instance was that this limit should be such that not more than five per cent, of an unselected group Get the answers to these examples as quickly as you can. Use the side of this page to figure on if you need to. SAMPLES 1 How many are 5 men and 10 men?.Answer ( 2 If you walk 4 miles an hour for 3 hours, how far do you walk?.Answer ( 1 How many are 60 guns and 5 guns?.Answer ( 2 If you save $9 a month for 3 months, how much will you save?..Answer ( 3 If 48 men are divided into squads of 8, how many squads will there be?.Answer ( 4 Mike had 11 cigars. He bought 2 more and then smoked 7. How many cigars did he have left?.Answer ( 5 A company advanced 8 miles and retreated 2 miles. How far was it then from its first position?.Answer ( 6 How many hours will it take a truck to go 42 miles at the rate of 3 miles an hour?...Answer ( 7 How many pencils can you buy for 60 cents at the rate of 2 for 5 cents?.Answer ( 8 A regiment marched 40 miles in five days. The first day they marched 9 miles, the second day 6 miles, the third 10 miles, the fourth 6 miles. How many miles did they march the last day?...Answer ( 9 If you buy 2 packages of tobacco at 8 cents each and a pipe for 65 cents, how much change should you get from a two-dollar bill?..........Answer ( 10 If it takes 4 men 3 days to dig a 120-foot drain, how many men are needed to dig it in half a day?..Answer ( 11 A dealer bought some mules for 82,000. He sold them for $2,400, making $50 on each mule. How many mules were there?.. ,.Answer ( 12 A rectangular bin holds £00 cubic feet of lime. If the bin is 10 feet long and 5 feet wide, how deep is it?.Answer ( 13 A recruit spent one-eighth of his spare change for post cards and twice as mugh for a box of letter paper, and then had $1.00 left. How much money did he have at first.Answer ( 14 If 2> l /2 tons of clover cost $14, what will OYi tons cost?. .Answer ( 15 A ship has provisions to last her crew of 700 men 2 months. How long would it last 400 men?.... . .Answer ( 16 If an aeroplane goes 250 yards in 10 seconds, how many feet does it go in a fifth of a second?...Answer ( .17 A U-boat makes 8 miles an hour under water and 20 miles on the surface. How long will it take to cross a 100-mile channel, if it has to go two-fifths of the way under water?..Answer ( 18 If 134 squads of men are to dig 3,618 yards of trench, how many yards must be dug by each squad?.Answer ( 19 A certain division contains 5,000 artillery, 15,000 infantry, and 1,000 cavalry. If each branch is expanded proportionately until there are in all 23,100 men, how many will be added to the artillery?.Answer ( 20 A commission house which had already supplied 1,897 barrels of apples to a cantonment delivered the remainder of its stock to 37 mess halls. Of this remainder each mess hall received 54 barrels t^Wfiat was th e tota l number of barrels supplied jh. Answer C 15 12 Plate II. Alpha Test 2: Arithmetical Reasoning (Form 8). 9 ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) 10 AMERICAN INTELLIGENCE would complete all the items in a test. In our group of 1047 cases, 5 persons answered 18 problems correctly in the five minutes allowed, but no one answered more than 18 prob¬ lems correctly. Of course no one was expected to answer them all. If a person passed 65% of each test in alpha he was graded “A”; perfection was not required. Figure 3 . Distribution of scores of the Arithmetical Reasoning test. (From p. 624, Memoir XV.) One often hears the statement that the army tests were “speed’' tests, and penalized the slow but accurate indi¬ vidual. Experiments were made to determine how the re¬ sults would change with extended time. A group of 475 men examined showed in Test 2 an improvement from an average of 8.00 to 9.16 with double time. In five minutes they solved on an average 8 problems correctly, in ten min¬ utes 9.16. The relationship between single time and double time scores may be measured by the statistical value known as the coefficient of correlation. Two measures that stand in a perfect one to one correspondence have a coefficient AMERICAN INTELLIGENCE 11 of correlation of 1.00. Two measures that stand in a perfect chance relationship have a correlation coefficient of 0. In practice it is found that a correlation of 0.90 is so high that one might substitute one series of measures for the other without seriously changing the results. The correlation be¬ tween the single time and double time scores was 0.937, a value so high that it indicates that there were very few changes in the relative position of the members of the group, and that such changes as occurred were small. The experiments that were conducted on time limits with the various tests all pointed to the conclusion that the re¬ sults would not be changed with the more extended time limits. Of course the absolute scores would be higher with the extended time, but the relative position of the mem¬ bers of the group would be about the same. In the experi¬ ment on double time referred to above, all the tests from 2 to 8 in alpha showed coefficients of correlation between single time and double time above 0.90 except Test 3 (0.879), and the correlation of the two total scores ob¬ tained under single and double time was 0.967. The army experimenters after considering all the evidence concluded that ‘'doubling the time does not result in any demonstr¬ able improvement in alpha as a whole.” (p. 417). It is prob¬ ably true that very high scores depend on "speed,” but inasmuch as a person only needed to answer correctly 65% of the items to be rated "A” and 50% of the items to be rated "B,” it can not be considered that "speed” is a factor that would affect the results seriously. The army findings of a correlation of 0.967 between the single time and double time trials of alpha, and the general conclusion that the results would not have been changed appreciably with more liberal time allowances, definitely controvert the popular belief that anything which is per¬ formed with a time limit handicaps the "slow but sure” 12 AMERICAN INTELLIGENCE individual. Popular judgment classifies the population into two groups, the “slow but sure” and the “quick and inac¬ curate,” and would have us believe that the quick type must of necessity be inaccurate, and that the sluggish indi¬ vidual is infallible. Science shows us that we really rate individuals on two scales, a scale of speed and a scale of accuracy, and that we find people who are both quick and accurate as well as people who are slow and inaccurate. Science would elaborate the popular classification by add¬ ing these two types. The popular “slow but sure” charac¬ terization is more apt to be an apology for dullness than a scientific diagnosis. At least in our consideration of the army test results we may definitely discard the opinion that we are testing “speed” rather than intelligence. The arithmetical reasoning test in alpha actually proved to be one of the best tests in the series. Alpha Test 3.— Practical Judgment Time limit : V /2 minutes The practical judgment test is one of the most interest¬ ing tests in alpha from many standpoints. There is no other test in alpha which contains, in all of the five forms used, so many individual items that may be criticised by a person who actually inquires into the logical validity of the an¬ swers accepted as correct. Item 12, for instance, might profitably be taken as the subject of an intercollegiate de¬ bate, as it has been the subject of many debates in the history of penology. The critics of the army tests are all too apt to consider the whole scale invalid if they can dis¬ cover a single incorrect item, for they fail to realize that a person could fall down on 35% of the individual items and still be rated “A.” This is a test of common sense. Below are sixteen questions. Three answers are given to each question. You are to look at the answers carefully; then make a cross in the square before the best answer to each question, as in the sample: SAMPLE < Why do we use stoves? Because □ they look well @ they keep us warm □ they are black Here the second answer is the best one and is marked with a cross. Begin with No. 1 and keep on until time is called. 1 It is wiser to put some money aside and not spend it all, so that you may O prepare for old age or sickness 0 collect all the different kinds of money 0 gamble when you wish 2 Shoes are made of leather, because 0 it is tanned 0 it is tough, pliable and warm 0 it cad be blackened 3 Why do soldiers wear wrist watches rather than pocket watches? Because 0 they keep better time 0 they are harder to break 0 they are handier 4 The mam reason why .-stone is used for building purposes is because 0 it makes a good appearance 0 it is strong and lasting 0 it is heavy 5 Why is beet better food than cabbage? Because 0 it tastes better 0 it is more nourishing 0 it is harder to obtain 6 If some one does you a favor, what should you do? 0 try to forget it 0 steal for him if he asks you to 0 return the favor 7 If you do not get a letter from home, which you know was written, it may be because 0 it was lost in the mails 0 you forgot to tell your people to write 0 the postal sendee has been discontinued 8 The main t hing the,fanners do is to 0 supply luxuries 0 make work for the unemployed 0 feed the nation Go to No. 9 above 9 If a man who can’t swim should fall into a river, he should 0 yell for help and try to scramble out 0 dive to the bottom and crawl out 0 lie on his back and float 10 Glass insulators are used to fasten telegraph wires because 0 the glass keeps the pole from being burned 0 the glass keeps the current from escaping 0 the glass is cheap and attractive 11 If your load of coal gets stuck in the mud, what should you do? 0 leave it there 0 get more horses or men to pull it out 0 throw off the load 12 Why are criminals locked up? 0 to protect society 0 to get even with them 0 to make them work 13 Why should a married man have his life in¬ sured? Because 0 death may come at any time 0 insurance companies are usually honest 0 his family will not then suffer if he dies 14 In Leap Year February has 29 days because 0 February is a short month 0 some people are born on February 29th 0 otherwise the calendar would not come out right 15 If you are held up and robbed in a strange city, you should 0 apply to the police for help 0 ask the first man you meet for money to get home 0 borrow some money at a bank 16 Why should we have Congressmen? Because 0 the people must be ruled 0 it insures truly representative government 0 the people are too many to meet and make their laws Plate III. Alpha Test 3: Practical Judgment (Form 8). 13 14 AMERICAN INTELLIGENCE The distribution of the scores made in Test 3 is shown in Figure 4. Figure 4. Distribution of scores of the Practical Judgment test. (From p. 624, Memoir XV .) Disregarding the large number of zero scores (163), which are probably the result of illiteracy plus failure to under¬ stand instructions, and also recognizing the fact that a few low positive scores may be due to chance, we may look at the distribution as entirely satisfactory. Many persons object to the short time limit (1 ]/o min¬ utes), but the test was undoubtedly more effective with this short limit than it would have been with the longer time. AMERICAN INTELLIGENCE -i ^ 10 The average score improves from G.32 to 9.85 with double time, and the correlation between and 3 minutes work on the test is 0.879. There are decided indications that double time would not be useful in improving the record of those whose score was high in the first \ x /i minutes. A very excellent criterion of the efficiency of a test is its value in differentiating between officers and men. In general, a sample of officers would contain a larger per¬ centage of men of high intelligence than a sample of en¬ listed men. The amount that a test differentiates the groups would indicate the value of the test. This test of practical judgment was the worst test in the whole series in differ¬ entiating officers from men. If we used this criterion alone there would be no possible excuse for retaining the test in the series. In differentiating officers from men, it was about twice as bad as the next to the poorest test (oral direc¬ tions). On the other hand, we need tests in alpha which are effective at the lower end of the scale, and we can set up as our criterion here the value of the test in differentiating between feeble-minded individuals and enlisted men. The alpha tests were given to the high grade feeble-minded population of two custodial institutions, and the results compared with a group of 300 English speaking enlisted men. Test 3 proved to be very much superior to any other test in the series in differentiating between feeble-minded individuals and enlisted men. This fact more than justifies the inclusion of Test 3 in the scale. All of these facts are difficult to interpret. My own inter¬ pretation is that the sixteen items did not measure or grade “practical judgment” in any sense, but that the inclusion of at least one very obviously false and really quite silly alternative in each item acted as an effective pitfall for the 16 AMERICAN INTELLIGENCE feeble-minded. At least we are sure that the actual experi¬ mental results are conclusive enough to dispose of any and all arm-chair criticisms. Alpha Test 4.— Synonym-Antonym Time limit : l3dz minutes If one will review the experimental literature of intelli¬ gence testing, he will find the synonym-antonym or “op¬ posites” test, used sometimes as a test of controlled asso¬ ciation, sometimes as a test of vocabulary, sometimes as a test of intelligence, but uniformly with excellent results. Given a group with a knowledge of English and sufficient intelligence to understand the nature of the problem, the synonym-antonym test will give as good a differentia¬ tion between the bright and dull members of the group, rated by an outside criterion, as any other standard test now available. The distribution of the scores in Test 4 is shown in Figure 5. The most striking feature of the distribution is the large number of scores that were either zero or one (393). This large number of zero scores is due to three causes. First, the illiterate group could not attempt it. Second, the stupid and literate could not understand the instructions and could not make the kind of judgment demanded. Third, in the long run chance or random responses would give scores around zero, for in scoring all tests that were a 50-50 guess, the total score was the number of right responses minus the number of wrong responses. If a person under¬ lined “same” for every item, his score would be 20 right minus 20 wrong, or zero. If he merely guessed, he would, Figure 5. Distribution of scores of the Synonym-Antonym Test. (From p. 625, Memoir XV.) IT If the two words of a pair mean the same or nearly the same, draw a line under same. If they mean the opposite, or nearly the opposite, draw a line under opposite. If you cannot be sure, .guess. The two samples are already marked as they should be. SAMPLES $ S o0< ^—bad.same— opposite 1 little—small. same —opposite 1 no—yes.same—opposite 1 2 day—night.same—opposite 2 3 go—leave.same—opposite 3 4 begin—commence...same—opposite 4 5 bitter—sweet.same—opposite 5 6 assume—suppose.same—opposite 6 7 command—obey.same—opposite 7 8 tease—plague.same—opposite 8 9 diligent—industrious..,.same—opposite 9 10 corrupt—honest..same—opposite 10 11 toward—from.same—opposite 11 12 masculine—feminine.same—opposite 12 13 complex—simple.same—opposite 13 14 sacred—hallowed...same—opposite 14 15 often—seldom.same—opposite 15 16 ancient—modern.same—opposite 16 17 enormous—gigantic.same—opposite 17 18 confer—grant.same—opposite 18 19 acquire—lose.same—opposite 19 20 compute—calculate.same—opposite 20 21 defile—purify..-.same—opposite 21 22 apprehensive—fearful.same—opposite 22 23 sterile—fertile.same—opposite 23 24 chasm—abyss.same—opposite 24 25 somber—gloomy.same—opposite 25 26 vestige—trace...same—opposite 26 27 vilify—praise..same—opposite 27 28 finite—limited.same—opposite 28 29 contradict—corroborate.same—opposite 29 30 immune—susceptible.same—opposite 30 31 credit—debit.same—opposite 31 32 assiduous—diligent.same—opposite 32 33 transient—permanent.same—opposite 33 34 palliate—mitigate.same—opposite 34 35 execrate—revile.. same—opposite 35 36 extinct—extant.same—opposite 36 37 pertinent—relevant.same—opposite 37 38 synchronous—simultaneous... .same—opposite 38 39 supercilious—disdainful.same—opposite 39 40 abstruse—recondite.same—opposite 40 Plate IV. Alpha Test 4: Synonym-Antonym (Form 8). 18 AMERICAN INTELLIGENCE 19 in the long run, guess half of the responses right and half wrong. Chance scores would then be zero (which includes all minus scores) and 1 or 2 on the positive side. In general we may interpret Test 4 as a “high grade” test. It is too difficult to give any differentiation between low grade individuals, but it effectively grades the higher orders of intelligence. The time limit is not too short, for doubling the time only raises the average score from 10.50 to 12.60, and the correlation between regular and extended time is 0.940. It is one of the most effective tests in the scale for differentiating officers from enlisted men, and for dif¬ ferentiating feeble-minded from enlisted men. The only criticism is that it was too hard for a large number of people examined. Figure 5 really gives only about half of the normal distribution. If the test were so easy that the lower end of the scale could be extended to about —20, the distribution would become normal. Alpha Test 5.—Disarranged Sentences Time limit : 2 minutes This test is an adaptation of a type of test which gives excellent results in the Binet-Simon scale. As it stands in alpha it is not a particularly good test. The distribution of scores shown in Figure 6 indicates a pile-up of zero scores due probably to the same three causes described as op¬ erating in Test 4. The test is fairly good in differentiating between officers and enlisted men, but for some reason or other it is the very worst test in the whole series in dif¬ ferentiating between feeble-minded and enlisted men. On the whole it is one of the poorest tests in our measuring scale. Figure 6. Distribution of scores of the Disarranged Sentence Test. (From p, 626, Memoir XV.) 20 The words A EATS COW GRASS in that order are mixed up and don’t make a sentence; but they would make a sentence if put in the right order: A COW EATS GRASS, and this statement is true. Again, the words HORSES FEATHERS HAVE ALL would make a sentence if put in the order ALL HORSES HAVE FEATHERS, but this statement is false. Below are twenty-four mixed-up sentences. Some of them are true and some are false. When I say “go,” take these sentences one at a time. Think what each would say if the words were straightened out, but don’t write them yourself. Then, if what it would say is true, draw a line under the word “true”; if what it would say is false, draw a line under the word “false.” If you can not be sure, guess. The two samples are already marked as they should be. Begin with No. 1 and work right down the page until time is called. ( a eats cow grass.true, .false SAMPLES , . , , „ — , , t horses feathers have all.true, . false 1 oranges yellow are.true.. false 1 2 hear are with to ears...true.. false 2 3 noise cannon never make a.true., false 3 4 trees in nests build birds.true.. false 4 5 nil water not and will mix....true. .false 5 6 bad are shots soldiers all.true., false 6 7 fuel wood are coal and for used..true. .false 7 8 moon earth the only from feet twenty the is.true.. false 8 9 to life water is necessary..true.. false 9 10 are clothes all made cotton of.true,..false 10 11 horses automobile an are than slower.true, .false 11 12 tropics is in the produced rubber.true..false 12 13 leaves the trees in lose their fall... .. .true, .false 13 14 place pole is north comfortable a the. true. .fake 14 15 sand of made bread powder and is..true., fake 15 16 saik k steamboat usually by propelled a.true., false 16 17 k the salty in water all lakes.true., fake 17 18 usually judge can we actions man his by a.true, .false 18 19 men misfortune have good never...true, .false 19 20 took valuable k for sharp making steel.true, .fake 20 21 due sometimes calamities are accident to.true..fake 21 22 forget trifling friends grievances never.true, .fake 22 23 feeling is of painful exaltation the.true.. fake 23 24 begin a and apple acorn ant words with the.true, .false 24 Plate V. Alpha Test 5: Disarranged Sentences (Form 8). 21 22 AMERICAN INTELLIGENCE Alpha Test 6.—Number Series Completion Time limit : 3 minutes This test is the only one in alpha demanding a high order of intelligence almost entirely independently of the use of language. The greatest difficulty was experienced with the instructions for this test, when the first trial was made at the army camps. The preliminary forms contained only two rows of samples, and the instructions included the rather involved statement: ‘ In the fines below, each number is gotten in a certain way from the numbers corning before it. Study out what this wav is in each fine and then write in the space left for it the number that should come next. The first two fines are already filled in as they should bed’ In the final alpha revision, four samples were included, and the instructions were simplified verbally and read very slowlv. The instructions were ffiven as follows: “Look at o the first sample row of figures at the top of the page: 2, 4, 6, 8, 10, 12; the two numbers that should come next are, of course, 14, 16/’ etc., for each sample. Long pauses fol¬ lowed the reading of each sample. The distribution of responses given in Figure 7 shows that the simplified instructions gave very good results, for although there were many zero scores in our experimental group of 1047 cases (,244), there were probably no more zero scores than might have been expected when we con¬ sider that the mere understanding of what was wanted re¬ quired considerable intelligence. On the whole the number series completion test proved to be entirely satisfactory. \ 23 SAMPLES 8 7 2 3 7 2 6 5 3 4 7 3 4 3 2 4 5 5 7 4 7 Look at each row of numbers below, and on the two dotted lines write the two numbers that should come next. 3 4 5 6 7 8 8 7 6 5 4 3 10 15 20 25 30 35 9 9 7 7 5 5 3 6 9 12 15 18 8 1 6 1 4 1 5 9 13 17 21 25 8 9 12 13 16 17 27 27 23 23 19 19 1 2 4 8 16 32 19 16 14 11 9 6 11 13 12 14 13 15 2 3 5 8 12 17 18 14 17 13 16 12 29 28 26 23 19 14 20 17 15 14 11 9 81 27 9 3 1 Vs 1 4 9 16 25 36 16 17 15 18 14 19 3 6 8 16 18 36 Plate VI. Alpha Test 6: Number Series Completion (Form AMERICAN INTELLIGENCE Alpha Test 7.—Analogies 25 Time limit : 3 minutes The analogies test gave results on a par with the syn¬ onym-antonym test. The distribution of scores given in Figure 8 shows a large number of zero scores (284), but this number is not larger than might have been expected, con¬ sidering the amount of intelligence necessary even to understand the nature of the task to be performed. This test again shows only a partial distribution. It it had been easier, it would probably have shown a symmetrical dis¬ tribution. Figure 8. Distribution of scores of the Analogies Test. (From p. 625, Memoir XV.) ( sky— blue:: grass— table green warm big SAMPLES < fish—swims :: man— paper time walks girl ( day—night:: white— red black clear pure In each of the lines below, the first two words are related to each other in some way. What you are to do in each line is to see what the relation is between the first two words, and under¬ line the word in heavy type that is related in the same way to the third word. Begin with No. 1 and mark as many sets as you can before time is called. 1 shoe—foot:: hat—kitten head knife penny. 1 2 pup—dog :: lamb—red door sheep book. 2 3 spring—summer :: autumn— winter warm harvest rise.. 3 4 devil—angel:: bad—mean disobedient defamed good....._ 4 5 finger—hand :: toe—body foot skin nail... 5 6 legs—frog:: wings—eat swim bird nest. 6 7 chew—teeth :: smell—sweet stink odor nose. 7 8 lion—roar:: dog—drive pony bark harness. 8 9 cat—tiger :: dog— wolf bark bite snap. 9 10 good—bad :: long—tall big snake short.•. 10 11 giant—large :: dwarf—jungle small beard ugly.. 11 12 winter—season:: January—February day month Christmas.. 12 13 skating—winter :: swimming— diving floating hole summer_ 13 14 blonde—light:: brunette—dark hair brilliant blonde. 14 15 love—friend :: hate—malice saint enemy dislike. 15 16 egg—bird:: seed—grow plant crack germinate. 16 17 dig—trench :: build—run house spade bullet. 17 18 agree—quarrel i: friend— comrade need mother enemy. 18 19 palace—king:: hut—peasant cottage farm city. 19 20 cloud-burst—shower:: cyclone—bath breeze destroy West... 20 21 Washington—Adams:: first— president second last Bryan .. 21 22 parents—command:: children—men shall women obey. 22 23 diamond—rare:: iron—common silver ore steel.•. . 23 24 yes—affirmative:: no—think knowledge yes negative. 24 25 hour—day:: day—night week hour noon. 25 26 eye—head :: window—key floor room door. 26 27 clothes—man :: hair—horse comb beard hat.. 27 28 draw—picture:: make—destroy table break hard.. 28 29 automobile—wagon :: motorcycle— ride speed bicycle car. 29 30 granary—wheat:: library— read books paper chairs. 30 31 Caucasian—English :: Mongolian—Chinese Indian negro yellow.. 31 32 Indiana—United States :: part— hair China Ohio whole. 32 33 esteem—despise :: friends—Quakers enemies lovers men. 33 34 abide—stay :: depart— come hence leave late. 34 35 abundant—scarce :: cheap—buy costly bargain nasty. 35 36 whale—large :: thunder—loud rain lightning kill. 36 37 reward—hero :: punish— God everlasting pain traitor.. 37 38 music—soothing :: noise—hear distracting sound report.. 38 39 book—writer :: statue— sculptor liberty picture state. •.... 39 40 wound—pain :: health — sickness disease exhilaration doctor.. 40 Plate VII. Alpha Test 7: Analogies (Form 8). 26 AMERICAN INTELLIGENCE 27 In the construction of this test it was deliberately plan¬ ned to make many items very difficult by introducing as a wrong alternative a word that was frequently associated with the key word. For instance, we find in the last 20 items the following easily associated pairs which would be wrong: (21) first-last, (22) no-yes, (25) day-night, (26) window- door, (27) hair-comb, (28) make-break, (29) motor-cycle- ride, (SO) library-read, (32) part-hair, (35) cheap-buy, (36) thunder-lightning, (38) noise-hear, (39) statue-liberty, (40) health-sickness. The test therefore involves not only the selection of the right word, but the refusal to accept as the solution a word that is exceedingly attractive owing to fre¬ quent associations. The analogies test is the most effective test in the entire series in differentiating officers from men. Eor some reason, not understood, it does not rank high in differentiating feeble-minded from enlisted men. The scores show a con¬ siderable average improvement with extension of time limit (8.60 to 12.46) but a correlation of 0.920 between three and six minutes work. On the whole it is an excellent test. 28 AMERICAN INTELLIGENCE Alpha Test 8.—Information Time limit : 4 minutes The army information test has been criticised more than any other test, and it has undoubtedly received much more abuse than it deserves. From the standpoint of test con¬ struction it is satisfactory, for, aside from the zero scores probably due to illiteracy, the distribution as shown in Figure 9 is rather good. Figure 9. Distribution of scores of the Information Test. (From p. 626, Memoir XV.) The most frequent charge made against the test is that a person could fail in certain items and still be intelligent. This is certainly true, and the criticism would be valid if anyone were expected to answer all the items, or if he were considered unintelligent if he failed. The average person answered less than 15 items correctly. Notice the sample sentence: People hear with the eyes ears nose mouth The correct word is ears, because it makes the truest sentence. In each of the sentences below you have four choices for the last word. Only one of them is cor¬ rect. In each sentence draw a line under the one of these four words which makes the truest sentence. If you can not be sure, guess. The two samples are already marked as they should be. ( People hear with the eyes ears nose mouth saaiplesI " ( France is in Europe Asia Africa Australia 1 The apple grows on a shrub vine bush tree. 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 IS 19 20 21 22 23 24 25 Five hundred is played with rackets pins cards dice.... The Percheron is a kind of goat horse cow sheep..... The most prominent industry of Gloucester is fishing packing brewing automobiles.. Sapphires are usually blue red green yellow....•... The Rhode Island Red is a kind of horse granite cattle, .fowl. Christie Mathewson is famous as a writer artist baseball player comedian. Revolvers are made by Swift & Co. Smith & Wesson W. L. Douglas B. T. Babbitt. Carrie Nation is known as a singer temperance agitator suffragist “There’s a reason” is an “ad” for a drink revolver flour cleanser. nurse. 2 3 4 5 6 7 8 9 10 11 1'. Artichoke is a kind of hay com vegetable fodder.. Chard is a fish lizard vegetable snake... Cornell Ur iversity is at Ithaca Cambridge Annapolis New Haven.... 13 Buenos Aires is a city of Spain Brazil Portugal Argentina ... 14 Ivory is obtained from elephants mines oysters reefs....... 15 Alfred Noyes is famous as a painter poet musician sculptor. 16 The armadillo is a kind of ornamental shrub animal musical instrument dagger. 17 The tendon of Achilles is in the heel head shoulder abdomen. 18 Crisco is a patent medicine disinfectant tooth-paste food product... 19 An aspen is a machine fabric tree drink. 20 The sabre is a kind of musket sword cannon pistol..... The mimeograph is a kind of typewriter copying machine phonograph pencil. Maroon is a food fabric drink color.’.. ... The clarionet is used in music stenography book-binding lithography. Denim is a dance food fabric drink... 21 22 23 24 25 26 The author of “Huckleberry Finn” is Poe Mark Twain Stevenson Hawthorne. 26 27 Faraday was most famous in literature war religion science;. 27 28 Air and gasolene are mixed in the accelerator carburetor gear case differential.' 28 29 The Brooklyn Nationals are called the Giants Orioles Superbas Indians.29 30 Pasteur is most famous in politics literature war science.... ..... 30 31 • Becky Sharp appears in Vanity Fair Romola The Christmas Carol Henry IV. 31 32 The number of a Kaffir’s legs is- two four six' eight.... . 32 33 Habeas corpus is a term used in medicine law theology pedagogy.. 33 34 Ensilage is a term used in fishing athletics farming hunting. 34 35 The forward pass is used in tennis hockey football golf. 35 36 General Lee surrendered at Appomattox in 1812 1865 1886 1832. 36 37 The watt is used in measuring wind power rainfall water power electricity. 37 38 The Pierce Arrow car is made in Buffalo Detroit Toledo Flint. 38 39 Napoleon defeated the Austrians at Friedland Wagram Waterloo Leipzig. 39 40-- An irregular four-sided figure is called a .scholium triangle trapezium pentagon. 40 Plate VIII. Alpha Test 8: Information (Form 8) 29 30 AMERICAN INTELLIGENCE The test was devised to sample as many fields of infor¬ mation as it was possible to sample with 40 items. In gen¬ eral the five information tests in the five forms of alpha sampled similar fields. For instance, the advertising slogans which appear in the five forms are “Hasn’t scratched yet” (Form 5), “The makings of a nation” (Form 6), “Even¬ tually, why not now?” (Form 7), “There’s areason” (Form 8), and “The flavor lasts” (Form 9), while the Overland, Buick, Rolls-Royce, Pierce Arrow and Packard appear in each of the five forms. Information tests vary considerably in construction. There is a great difference between asking for the date of Lee’s surrender and asking a person to choose between the four dates, 1812, 1865, 1886 and 1832. And again, a person is merely asked to elect whether the bassoon, xylophone, cymbal, clarionet and piccolo, appearing in each of the five forms of alpha, should be used in music, stenography, book binding, or lithography. Approximately a third of the times test for vocabulary rather than information in the literal sense. If a person, for instance, knows what a Zulu, or a Korean, or a Hottentot, or a Kaffir or a Papuan is, he very obviously knows the number of his legs. As a rule women object to the information test more than men because the test samples rather heavily the fields of sport, mechanical interests, etc. The chances are that this test would penalize women rather heavily, but as a general rule the results of comparing the two sexes on alpha as a whole at various colleges show very slight differences in favor of the men. The sex differences found are not large enough to be significant. At Camp Lee a group of 164 captains and 200 enlisted men of the same general intelligence level (i. e. A and B) were examined. The test showing the greatest differentia¬ tion between the two groups was the information test. The AMERICAN INTELLIGENCE 31 only other test showing a difference in favor of the captains was Test 4 (synonym-antonym test), while the test for arith¬ metical reasoning (Test 2) and practical judgment (Test 3) showed differences in favor of the A and B enlisted men. The differences were, of course, small, but the greatest dif¬ ference was shown by the information test. In differentiat¬ ing between officers and the general run of enlisted men, the information test was fairly effective, and it was very nearly as good as the arithmetical reasoning and synonym- antonym test in differentiating between feeble-minded in¬ dividuals and enlisted men, in spite of the fact that the feeble-minded obtained a somewhat higher percentage of their total score from this test than did the enlisted men. After weighing all the evidence, it would seem that we are justified in ignoring most of the arm-chair criticisms of this test and in accepting the experimental evidence tend¬ ing to show that the test was a fairly good one. The assump¬ tion underlying the use of a test of this type is that the more intelligent person has a broader range of general in¬ formation than an unintelligent person. Our evidence shows that this assumption is, in the main, correct. SECTION II EXAMINATION BETA When we turn to examination beta, we meet an en¬ tirely different problem, that of testing the intelligence of wholly or partially illiterate persons who could not take alpha on account of their language handicap, of testing non- English speaking persons some of whom knew only the simplest commands in English, and low grade individuals who did not have sufficient intelligence to make a substan¬ tial score on alpha. At the time of the first try-out of the army tests in the fall of 1917 at four cantonments, Devens, Dix, Lee and Taylor, examination a , the fore-runner of alpha, was in use, and various types of individual examina¬ tions were being tried out, but there was no non-verbal group test. To meet this need, a preliminary try-out of fif¬ teen tests was made early in 1918, and a final examination composed of seven tests was subsequently published and widely used throughout the country. In the following pages the seven beta tests are reproduced in Plates 9 to 15, the method of administering them is described, and the results from the experimental group of 1047 cases are presented. Examination beta was given under the most rigid experi¬ mental conditions. The experimenter stood on a platform back of which was a large black-board on which small dup¬ licates of the seven tests could be shown one at a time. A demonstrator, whose duties were to act out the test prob¬ lems on the black-board, was an essential part of the ex¬ periment. The experimenter in pantomime showed the de¬ monstrator what to do on the black-board, then, after his 32 AMERICAN INTELLIGENCE 33 performance was completed, orderlies throughout the room explained to the men that they were to go ahead with the test and do what the demonstrator had done. The orderlies’ vocabulary was limited to “Yes,” “No,” “Sure,” “Good,” “Quick,” “Hurry up,” “How many?” “Same,” “Do it,” “Fix it.” The experimenter used just as few words as pos¬ sible, and acted out every spoken sentence by pointing, motioning, etc. The demonstrator never spoke. His duties consisted of doing before the group just what the group was expected to do with the examination blanks. Beta Test 1.—Maze Time limit : 2 minutes The black-board was turned so that two sample mazes as shown in Figure 10 appeared. The experimenter traced TEST 1 “Lri ti r_ j, L r 1 ->• J rr c — i^n F _1 L i 1 r "" 1 1 Figure 10. Black-board chart for demonstrating the Maze Test. 34 AMERICAN INTELLIGENCE through the first maze on the black-board, and then mo¬ tioned the demonstrator to go ahead. The demonstrator traced through the maze with crayon very slowly. The ex¬ perimenter then traced through the second maze and mo¬ tioned the demonstrator to go ahead. The demonstrator in tracing this maze made a mistake by crossing the line at the end of a blind alley, was corrected by the experi¬ menter with vigorous shakes of the head and “no-no,” and made to re-trace his path back to where he could start right again. The demonstrator then traced through the rest of Figure 11 . Distribution of scores of the Maze Test. (From p. 627, Memoir XV.) 1 2 3 4 5 Plate IX. Beta Test 1: Maze. 35 36 AMERICAN INTELLIGENCE the maze with great semblance of haste, stopping momen¬ tarily at each ambiguous point only. The experimenter then motioned to the group to do the same thing on their exam¬ ination blanks. The experimenter and the orderlies walked about the room, motioning to the men who were not work¬ ing, and saying, “Do it, do it, hurry up, quick.” The results of the maze test are shown in Figure 11. The difference between the distribution of scores of this test and the alpha tests is remarkable. In the first place, our large number of zero scores has disappeared—only 19 in our group of 1047 failed to make any score. In the second place, the test is entirely too easy, for it is apparent that the men in the upper end of the scale could have done more in the time allowed. The maze test was intentionally made easy in order to get everybody started. We have at last found a test in which practically everybody can do some¬ thing. Aside from the language involved, every test in alpha is harder than this beta maze test, for no alpha test has less than 7% zero scores. Beta Test 2.—Cube Analysis Time limit : 2}^ minutes The black-board was turned to show a series of cubes like that in Figure 12. On a shelf was a real three cube TEST 2 m U U "□ □ Figure 12. Blackboard chart for demonstrating the Cube Analy¬ sis Test. AMERICAN INTELLIGENCE 37 model similar to the first one on the black-board. The ex¬ perimenter pointed to the three cube picture on the black¬ board, then to the model on the shelf, then to the picture on the black-board, and asked, “How much?” The experi¬ menter then counted aloud, putting up his fingers while counting, and encouraged the men to count with him. The experimenter then tapped each cube on the black-board and asked the demonstrator, “How much?” The demon¬ strator then went to the black-board, counted the cubes by pointing, and wrote the number 3 in the space below the illustration. A similar performance was enacted for the other three problems on the [black-board, the models being shown and elaborately counted. The distribution of scores of the cube analysis test is shown in Figure 13. Here we find a somewhat larger num- Figure 13. Distribution of scores of the Cube Analysis Test. (From p. 627, Memoir XV.) ber of zero scores (54) than in the maze test, but a fairly good distribution in general. On the whole the test is easy. Plate X. Beta Test 2: Cube Analysis. 38 AMERICAN INTELLIGENCE 39 Beta Test 3.—X-0 Series Time limit : 1% minutes The black-board was turned to show the chart repro¬ duced in Figure 14. The experimenter traced with a pointer TEST 3 0 oololol LIT] X XXX 0 o 2< o ox oc >x L IX XXO xjdxx 'XlOIXil 11II1 Figure 14. Black-board chart for demonstrating the X-0 Series Test. each “0” in the top chart, and then wrote (with his pointer) an imaginary “0” in the four remaining spaces. The dem¬ onstrator then filled in the four “O’s” with crayon. The experimenter then traced the first “X” by tracing a semi¬ circle above the chart and so on. The demonstrator then filled in every other space with an “X” following the ex¬ perimenter’s elaborate exercise. The demonstrator then worked out the remaining problems with the same ritual, following which the men were instructed to go ahead. The distribution of scores of the X-0 series test is shown in Figure 15. The army writers state that: “Beta 3 defies interpreta¬ tion.” (p. 638.) We know that the test was devised to dup¬ licate in pictorial form the number series completion test 40 AMERICAN INTELLIGENCE Figure 15. Distribution of scores of the X-0 Series Test. (From p. 627, Memoir XV.) in alpha. Aside from knowing the purpose of the test, we know very little about it. My own guess would be that the first five or six items were entirely too easy, and that if they had been disregarded in the scoring, as practice items, and six more items added, comparable in difficulty to the last six items, the distribution would have been satisfactory. 1 i* x I x ! x ___ 1 X X X 2 X X X ]— 1* X 3 *o|xU * 9 0 X o 4 XX X X X X 5 X|0 X 1* 0 X 0 X 0 6 7 8 * X 0 1 X i X 0 X X 0 X X c o 0 X, X 9 0 X X 0 9 X X XT *1 X 0 0 0 X X 0 O' 0 X X 0 9 0 9 X 0 X X c X X 0 X X 0 X 10 X 0 X X 0 X 0 x 1 X 0 X 0 11 X 0 X X o X X X 0 x!° X X X X o X 0 12 xix| xjx 9 '0 \ c 1 X X 0 XX X | X I o o o X x| o\ P XI* Beta Test S: X-O Series. 41 42 AMERICAN INTELLIGENCE Beta Test 4.—Digit-Symbol Time limit : 2 minutes The black-board was turned to the chart shown in Fig¬ ure 16. The experimenter pointed to each number and then TEST 4 1 2 [3] [4 5 6 7 a \s\ Ml □ L EJ 0 A X 3 1 Z 3 l 1 Z 1 3 4 7 5 4 1 6 Figure 16 . Black-board chart for demonstrating the Digit-Symbol Test. to the symbol under it. The experimenter then pointed to the number 3 in the sample, then to the space below it, then to the number 3 in the key above, then to the symbol for 3, and finally traced the outline of the symbol for 3 in the proper space in the sample. This procedure was then re¬ peated for the first five samples. The demonstrator then went to the black-board, and worked through the process of filling in the symbols under the figures, touching each figure and symbol in the key while he drew the proper sym¬ bol in the sample. The group then proceeded to fill in the symbols on the test blank. The distribution of scores of the digit-symbol test is shown in Figure 17. AMERICAN INTELLIGENCE 43 Figure 17. Distribution of scores of the Digit-Symbol Test. (From p. 628, Memoir XV.) This type of distribution is the same as that given by the eight alpha tests, the zero scores representing failure to understand instructions, and the distribution being fairly regular. The digit-symbol test is a standard test, and the results in beta are entirely satisfactory. 3 1 2 1 3 2 1 4 2 3 5 2 9 1 4 6 3 1 5 4 2 7 6 3 8 7 2 9 5 4 6 3 7 2 8 1 9 5 8 4 7 3 8 9 5 1 9 2 8 3 7 4 6 5 9 4 8 5 7 8 4 4'4 AMERICAN INTELLIGENCE 45 Beta Test 5.— Number Checking Time limit : 3 minutes The black-board was turned to the chart shown in Fig¬ ure 18. The experimenter pointed first to the 6 in the left TEST 62 . .... 62 59 . ....56 327 .... .. .327 249 .... .... 249 1536 ... ....1536 3745 ... ....3745 45010 .. ....45001 62019 .. ....62019 Figure 18 . Black-board chart for demonstrating the Checking Test. N umber hand column, then to the 6 in the right hand column, then to the 2 in the left hand column and to the 2 in the right hand column, nodded his head, said, kv Yes,” and made an imaginary cross on the dotted line. The demonstrator then O i / made an “X” on the line. The experimenter repeated the procedure for the second pair, but indicated clearly, by shaking his head and saying, “No,” that the 9 and the G were not alike. The experimenter then repeated the pro¬ cedure for three more sets, getting the men in the room to 46 AMERICAN INTELLIGENCE say,“Yes,” if the pairs were identical. The demonstrator then worked out the remaining items. The results of the number checking test shown in Fig¬ ure 19, give the same distribution characteristic of the TEST 5, BETA Figure 19. Distribution of scores of the Number Checking Test. (From p. 628, Memoir XV .) alpha tests. The instructions were clear and the test was en¬ tirely satisfactory. This test is an adaptation of a standard test in use for many years. 650 . . 650 10243586 ... 10243586 041 . . 044 659012534 . . 6590211354 2579 . . 2579 388172902 . 381872902 3281 . . 3281 631027594 . . 631027594 65190 . 55102 2499901354 . . 2499901534 39190 . . 39190 2261059310 . . 2261659310 658049 . 650849 2911038227 . . 2911038227 3295017 . . 3290517 313377752 . . 313377752 63015991 . . 63019991 1012938567 . . 1012938567 39007106 . . 39007106 7166220988 . . 7162220988 69931087 . . 69931087 3177628449 . 3177682449 251004818 . . 251004418 468672663 . . 468672663 299056013 . . 299056013 9104529003 . . 9194529003 36015992 . . 360155992 3484657120 „. 3484657210 3910066482 . . 391006482 8588172556 . 8581722556 8510273301 . .. . 8510273301 3120166671 . . 3120166671 263136990 . . 263136996 7611348879 . . 76111345879 451152903 . . 451152903 26557239164 . . 26557239164 3259016275 . . 3295016725 8819002341 . 8819002341 682039144 . 582039144 6571018034 . . 6571018034 61558529 . . 61588529 38779762514 . . 38779765214 211915883 . . 219915883 39008126557 . . 39008126657 670413822 . . 670143822 75658100398 . . 75658100398 17198591 . . 17198591 41181900726 . 41181900726 36482991 . . 36482991 6543920S17 . . 6543920871 Plate XIII. Beta Test 5: Number Checking. 47 48 AMERICAN INTELLIGENCE Beta Test 6.—Picture Completion Time limit : 3 minutes The black-board was turned to the chart shown in Fig¬ ure 20. The experimenter pointed to the hand and said, TEST 6 Figure 20. Black-board chart for demonstrating the Picture Com¬ pletion Test. “Fix it.” The demonstrator looked puzzled. The experi¬ menter pointed to the place where the finger was missing, and said, “Fix it; fix it.” The demonstrator then drew the finger. The experimenter then pointed to the fish, and the place for the eye, and said,“Fix it.” After the demonstrator had drawn in the eye, the experimenter pointed to each of the drawings and said, “Fix them all.” After the demon¬ strator had worked out all the remaining drawings the group proceeded to complete the drawings in the beta blank. The results of the pictorial completion test as given in Figure 21 show an excellent distribution of the same general type as the distribution of the eight alpha tests. The num¬ ber of zero scores (12) on this test is smaller than that of any other test in the entire alpha-beta series. This is an excellent test. Figure 21. Distribution of scores of the Picture Completion Test. (From p. 628, Memoir XV.) Plate XIV. Beta Test 6: Picture Completion. 50 AMERICAN INTELLIGENCE 51 Beta Test 7.—Geometrical Construction Time limit : 2j/2 minutes The black-board was turned to the chart shown in Fig¬ ure 22. The experimenter pointed to the square on the TEST 7 Figure 22. Black-board chart for demonstrating the Geometrical Construction Test. black-board, and taking two pieces of cardboard the same size as the drawings at the left of the square, fitted them on the two drawings. He then fitted the two pieces of card¬ board together on the square to show that they would fill it, and motioned to the demonstrator who drew, in the square, the lines indicating the manner in which the two pieces would fit. This procedure was repeated for the next two samples. The demonstrator then worked out the last sample. The responses given in Figure 23 show, aside from the zero scores, a peculiar distribution, the form of which may be interpreted by assuming that the test was too hard in its 52 AMERICAN INTELLIGENCE Figure 23. Distribution of scores of the Geometrical Construction Test. (From p. 628, Memoir XV .) beginning and too easy at the end. We can picture the dis¬ tribution of Test 7 as shown in Figure 23 as having come from the middle range of a more complete test such as Test 5, Figure 19. If we cut the distribution of Test 5 from 10 to 19, we can picture the range in which Test 7 was working. We may assume, then, that the inclusion of a few easier items and five or ten harder ones might have given a distribution similar to that of the alpha tests. The test is faulty because of the limitation of range at both ends. 1 r 2 /\ \ X - 3 - r z> 7 5 u 8 K N \l / > < > 8 r— ti V A 10 — Plate XV. Beta Test 7: Geometrical Construction. 53 SECTION III THE INDIVIDUAL EXAMINATIONS The greatest contribution of the army psychologists to the development of mental tests was that of creating the two group tests, alpha and beta, that have been discussed. Methods of individual examining had been in existence for several years. The basic series of tests of the individual ex¬ amination was the Stanford Revision of the Binet-Simon scale, which had become a standard measurement and needs no description here. Persons interested in this method should read Terman’s 1 book on the Stanford scale. A method of abbreviating the Stanford-Binet scale was worked out in the army, and proved satisfactory. The distribution of the scores in terms of the “mental ages” of the 653 men in the special experimental group of 1047 cases who took the Stanford-Binet examination was as follows: “Mental Ages”. 5 6 7 8 9 10 11 12 13 14 15 1G 17 18 19 No. of eases. 1 2 1 22 62 66 G9 81 69 77 63 54 47 34 5 A rough inspection of these figures shows that they give us the Gaussian normal distribution. The results obtained from the Stanford-Binet examination may be taken as en¬ tirely reliable without question. One difficulty in the popular interpretation of the results on the Stanford-Binet scale, and other scales constructed on the same principle, is the unfortunate use of the term “mental age,” a term first used by Binet and subsequently X L. M. Terman. The Measurement of Intelligence. Boston, 1916, pp. 362. 54 AMERICAN INTELLIGENCE 55 used in this country. The term “mental age” has no signi¬ ficance whatsoever aside from the particular scale from which it was derived. A person might have a “mental age” of 13 on the Stanford-Binet scale, of 11 on Goddard’s translation of Binet’s 1908 scale, of 12 on Goddard’s 1911 scale, and so forth for every scale in use. The term “mental age” really means a score on a particular series of tests. Through rather general usage, the Stanford-Binet scale is being adopted in this country as a standard. The Stanford-Binet scale was constructed out of some 90 different tests arranged for different age levels, six for each age level from 3 to 10, eight for 12, six for 14, six for 16 or “average adult,” six for 18 or “superior adult,” and sixteen alternative tests interspersed throughout the scale. A person obtains his total score or “mental age” by taking all the tests in perhaps four or five age level groups. For instance, if a person passed all the tests at the 9 year level, five out of the six at the 10 year level, four out of the eight at the 12 year level, two out of the six at the 14 year level, and failed all tests above 14, his total score or “mental age” would be ll^. In assigning a given test to any age level, all the tests were first tried out, and the positions of the tests juggled about so that the ten year old children tested had an average score of 10, the eleven year old children an average score of 11, etc. When we say that a person has a “mental age” of eight on the Stanford-Binet scale, we do not mean that he has the mentality of a child of eight, but that he made a total score on that scale equal to that of the average eight year child tested in the particular group on which the scale was standardized. In all, about 1000 children, approximately equally distributed in the chronological ages from 5 to 14, formed the basis of the Stanford standardization. This standardization is a very excellent method of measuring 56 AMERICAN INTELLIGENCE intelligence, and worked very well with our army adult group as shown by the score distribution of the 653 men in the special experimental group given above, but we should always regard the term “mental age” as a score , not as a diagnosis. By correlating the alpha test with the Stanford-Binet scale, we can find the approximately equivalent score, or “mental age” for each possible alpha score. The operation resembles that of expressing values sterling in dollars. One frequently hears the statement that the army tests proved that the average citizen of this country has a mental age equivalent to that of a child of thirteen. Nothing could be more ridiculous. It is true that the average score of a sample of 93,955 soldiers representing the entire white draft, when translated into the Stanford-Binet scale, is 13.14. This means that the approximately equivalent score on the Stanford-Binet is 13.14. To say that the average citizen has a mentality of the child of 13 is putting the cart before the horse, for we are grading 93,955 people, and by infer¬ ence the entire country, on a standard fixed by some 82 fourteen year old children who happened to be tested in California. In addition to the 1000 children on whom the Stanford- Binet scale was standardized, the tests at the 16 year “average adult” level, and the 18 year “superior adult” level were standardized on 30 business men, 150 “migrat¬ ing” unemployed men, 150 adolescent delinquents, and 50 high school students. It is thus seen that the Stanford- Binet standardization rests on a number of cases too small to upset the army standards based on 93,955 cases. The term “mental age” is bad scientific slang for a total score. Psychologists are gradually abandoning the age standardization of tests. At the same time, publicists in various fields, although novices in psychology, are drawing AMERICAN INTELLIGENCE 57 rather vicious conclusions from “mental age” findings. It is an unfortunate situation. The methods used in creating and standardizing psycho¬ logical tests are entirely empirical, and therefore rather hard to explain to the layman, who is familiar only with the “school teacher” type of examination. The school teacher writes an examination, and lets it stand as an ab¬ solute measure. The psychologist makes an examination, tries it out, and judges each individual member of the group as compared with the other members of the group. As more and more people are examined his standards of judgment become more reliable. In other words, his stand¬ ards are those that he gets, not those that he thinks he ought to get. Therefore, instead of deploring the fact that the average person has a “mental age” of thirteen, we can simply say that the conversion of the results of the army test into the Stanford-Binet scale shows an average score of 13, and that this is the score to be expected from the average adult. Another very common mis-statement prevalent con¬ cerning the army results is that they proved that 24.9% of the drafted men were illiterate. Among the men sent to examination beta would be found, first, English speaking illiterates, second, non-English speaking individuals, either literate or illiterate in their own tongue, third, defectives, and fourth, cases accidentally sent to the wrong examina¬ tion. The method of selecting men for beta varied from camp to camp, and sometimes from week to week in the same camp. There was no established criterion of literacy, and no uniform method of selecting illiterates. In a group of 1,552,256 men examined, 386,196 or 24.9% were, for some reason or other, sent to beta. The army definition of literacy as “ability to read and understand newspapers and write letters home ” can not he identified with the fact of having 58 AMERICAN INTELLIGENCE been sent to beta. The statistics of the army examinations give us no accurate figures on the percentage of illiteracy. The individual examination for illiterates and non- English speaking, the performance scale, was a composite scale, the tests of which were drawn from workers in vari¬ ous fields, particularly from H. A. Knox, who had worked on non-verbal performance tests at Ellis Island, R. Pintner and D. G. Paterson, who had developed a scale of perform¬ ance tests, William Healy, H. H. Goddard, and other in¬ vestigators. The performance examination was given some¬ times as the long scale (8 or 10 tests) and sometimes as the short scale (5 tests). The short scale showed a correlation of 0.97 with the long scale, so that the reduction of the number of tests to save time in examining was entirely jus¬ tified. A short-cut method was also used in giving the Stanford-Binet examination which was quite satisfactory (correlation 0.91). The short performance scale showed a correlation of 0.84 with the Stanford-Binet scale. The Yerkes-Bridges point scale which was sometimes used in¬ stead of the Stanford-Binet was also abbreviated, and the abbreviated point scale showed a correlation of 0.934 with the complete point scale. In general, the methods of indi¬ vidual examining in use were quite reliable, and so closely related to the Stanford-Binet scale that the results could be converted into Stanford-Binet scores without any ap¬ preciable source of error. In all calculations in this study, scores from the point scale and the performance scale ex¬ aminations have been treated by converting into Stanford- Binet scores. SECTION IV RELIABILITY OF THE MEASURES The reader, who has followed the discussion of the indi¬ vidual tests through the preceding pages, will be convinced that most of the tests used w r ere satisfactory. In general, the eight alpha tests, the Stanford-Binet scale, and tests 4,5,6 and 7 of beta gave complete or limited distributions which approximated the Gaussian normal curve, Figure 24. Figure 24. The normal distribution curve. A type of distribution given by all alpha tests, tests 4, 5, 6 and 7 of beta, and the Stan¬ ford-Binet scale. Beta test 3 gave a distribution which could not be inter¬ preted, while beta tests 1 and 2, being too easy, gave a skewed distribution of the approximate form of the curve shown in Figure 25. It is not necessary here to enter into any lengthy dis¬ cussion of the method of converting the results of these sixteen tests into the combined scale. The reader interested in the statistical methods used is referred to Chapter 2, Part 3 of Memoir XV (pp. 573-657). If the reader is satis- 59 60 AMERICAN INTELLIGENCE fled that 13 of the 16 tests give distributions conforming in a general way to that shown in Figure 24, that is all that is necessary. The tests can very obviously be equated, and a combined scale constructed. By treating each of the eight tests of alpha, each of the seven tests of beta, and the Stanford-Binet test as different measuring scales, the com¬ bined scale was evolved, based on the inter-relations of these sixteen scales as shown by refined methods of cor¬ relation. Figure 25. A skewed distribution curve. A type of distribution given by beta tests 1 and 2. The army results are reported in tables showing the number of men scoring in certain class intervals, i. e. be¬ tween 0 and 4, 5 and 9, 10 and 14, etc., up to the interval 205 to 212 on alpha; between 0 and 4, 5 and 9, 10 and 14, etc., up to 115-118 on beta. In the same way, the scores in other tests are reported in class intervals. The study of the 1047 cases showed how individuals falling in each of the class intervals were distributed on the theoretical com¬ bined scale, i. e., it was possible to find the combinations of tests from which individuals in each class interval would AMERICAN INTELLIGENCE 61 obtain their scores. The combined scale was therefore built empirically on the results of the 1047 cases. Tables were constructed on this basis showing how individuals falling in each class interval of each of the three examinations should be redistributed on the combined scale. It is then possible to take a group which had been examined partly by alpha, partly by beta, and partly by the Stanford- Binet examination, and determine how that group would have scored on the combined scale if all individuals in the group had been given all three examinations. The com¬ bined scale is the most accurate method available for treat¬ ing the data derived from the army examinations. In this study the data from the principal sample have been re-figured on the combined scale by the method de¬ scribed on page 652 of Memoir XV: “In each group the alpha distribution was distributed on the combined scale by the use of table 159, the beta dis¬ tributions by table 162, and the Stanford-Binet mental age distribution by table 163. The performance scale distribu¬ tions and the point scale distributions were handled in the following way: the performance distributions were first transformed into Stanford-Binet mental age distributions by the use of the regression formula: 0.50 Performance score + 72 12 This formula was derived from the correlation of a sample of 350 cases who had both Stanford-Binet mental age rat¬ ings and performance scale ratings. The point scale distri¬ butions were transformed into Stanford-Binet mental age distributions by the use of the table in the examiner’s guide, Part I, pages 195ff. These transformations only approxi¬ mate the truth, but owing to the fact that the performance Mental Age (in years) 62 AMERICAN INTELLIGENCE and point scale cases constitute less than 3 per cent of any group handled it would take a considerable error in trans¬ formation seriously to affect the whole.” The conversion of the data of the principal sample into the combined scale reported in Memoir XV contains some inaccuracies. The statistical labor involved in the evolution of the combined scale was so great that the method was not available until the report was practically completed. The calculations were made by different individuals working under pressure, and errors were unavoidable. It has there¬ fore been considered worth while to repeat these calcula¬ tions at leisure, checking each operation carefully and carrying the analysis of some of the groups of the principal sample further than that reported in Memoir XV. It is now necessary to review very briefly the results of checking the army mental tests against outside criteria. We might have a measuring scale, all elements of which gave perfect score distributions, and which were highly inter-correlated, but even then we would need outside cri¬ teria to prove that we were measuring intelligence. Enough material is already on hand to prove that the army tests were reliable measures of intelligence. In the following dis¬ cussion we will cite several instances. The best proof of the validity of the test series comes from a study of the relation between the intelligence rat¬ ings and education. The correlation of the combined scale with reported school grade was 0.75 (based on 653 cases from the special experimental group of 1047 men). The correlation between alpha scores and schooling for this group was 0.75, the eight tests of alpha separately com¬ pared with schooling all showing correlations between 0.60 and 0.74. The correlation with beta total scores and school¬ ing for this same group was 0.67, and that between Stanford- Binet scores and schooling 0.65. These correlations show AMERICAN INTELLIGENCE 63 a positive relationship between intelligence as measured by the various methods and years of schooling. Very few people realize the severity of the elimination process that goes on from year to year in our schools and colleges. The study of the schooling of the native born white draft, as sampled by upward of 80,000 cases, showed the following startling facts: of every thousand native born recruits who entered the first grade, 970 remained in school till grade two, 940 till grade three, 905 till grade four, 830 till grade five, 735 till grade six, 630 till grade seven, and 490 till grade eight; 230 of them entered high school, 170 kept on till the end of the second year, 120 till the end of the third year, and 95 of the original thousand graduated from high school; 50 of these entered college, 40 kept on till the end of the second year of college, 20 till the end of the third year, and 10 graduated. It is, of course, impossible to determine how many of those that leave school leave on account of lack of pecuniary opportunity, or on account of lack of intelligence. It is ridiculous to assume that 1000 men in 1000 have sufficient intelligence to graduate from college, and equally absurd to assume that only 10 in 1000 have such a high intellectual endowment that they can graduate from college. Rut, inasmuch as we can not deny the intellectual elimination, we must expect a very high correlation between intelligence and schooling. The army tests uniformly show officers superior to en¬ listed men. This is to be expected, for officers were selected for ability. Nevertheless, one may perhaps contend that the high scores of the officers were due to superior educa¬ tion and not to greater intelligence. Very nearly half of the officers were college graduates, and another quarter had begun but not completed a college course. The objection that the superior scores of officers were due to education rather than intelligence may be effectively answered by a 64 AMERICAN INTELLIGENCE crucial test, which was made by the army investigators when they compared the alpha scores of 660 officers who had never gone beyond the eighth grade in school with the alpha scores of 13,943 native born recruits all of whom had gone beyond the eighth grade. The results of this comparison are reported (p. 779 of Memoir XV) as follows: “Every recruit in the recruit group has had more schooling than any officer in the officer group; the least educated recruit in the group has had a longer education than the best edu¬ cated officer included. And the group of officers neverthe¬ less makes a slightly higher record on examination alpha. It is evident then that the examination is measuring other qualities, in which officers stand above recruits, to a greater extent than it is measuring education.” The distri¬ butions of the alpha scores of these two groups are shown in Figure 26. In general, the comparison of the army test scores with education indicates that the tests are genuine measures of intelligence. The army investigators were, of course, called upon early in the war to prove that the tests they recommended were genuine tests of intelligence. For the assistance of army examiners and administrative officers having before them the problem of the assignment of men, a small pamphlet, Army Mental Tests (Washington, D. C., 1918, pp. 24) was prepared, presenting in graphic form the results of several different methods used in some of the camps for establishing the reliability of the tests by checking them against outside criteria. In the following pages, some of the charts from this booklet have been reproduced, and the method of interpreting the charts is described briefly. All of the methods reported use the letter grade classification (A, B, C, etc.), which is less accurate than the combined scale method used in this study, but they tend to show in OFFICERS Figure 26. Examination alpha as independent of education. Com¬ parison of alpha scores of officers of eighth grade schooling or less with alpha scores of native born white recruits of ninth grade schooling or more. “Although these groups overlap in schooling not at all; the officers make nevertheless slightly higher scores on alpha.” (Quotation from p. 779; and figure from p. 778; Memoir XV.) 65 66 AMERICAN INTELLIGENCE a general way the trend of the results, and that is all that is necessary in this case. In the long run, we should expect a small positive cor¬ relation between intelligence and rank. Intelligence is by no means the sole determiner of military success, but since it is one element in the complex of abilities required, we would expect to find a general tendency toward high scores with higher ranks. Figure 27, which is reproduced from the ENLISTED MEN (13792) —Relatively Illiterate ENLISTED MEN (82936) —Literate CORPORALS (4023) SERGEANTS (3393) O. T. C. (9240) Figure 27. Distribution of intelligence scores according to rank. The officers are above the candidates in the Officers’ Training Camps (O. T. C.), the candidate officers are above the sergeants, the sergeants above the corporals, and the corporals above the enlisted men. (From p. 8 of pamphlet, Army Mental Tests.) booklet referred to, shows the distribution of scores of vari¬ ous ranks on the rough A, B, C scale. The officers form a group quite distinct from the general run of enlisted men, and they are also above the candidates for commissions in AMERICAN INTELLIGENCE rr 67 the Officers’ Training Camp (O. T. C.) group. The sergeants are above the corporals, and the corporals above the en¬ listed men. The Officers’ Training Camps give an additional check on the intelligence tests. In the schools examined, the can¬ didates were recommended for a period of special training for commissions by the regimental organizations. The selec¬ tion of the candidates was very rigid, then, in the first in¬ stance. Figure 28 shows roughly the results of applying the A B C+ C C— D PERCENT SUCCESS PERCENT FAILURE O. T. C. 1375 MEN Figure 28. Success in Officers’ Training Camps as predicted by examination alpha. Each vertical bar represents all (100%) of the candidates who tested A, B, C-j-, etc. All men above the hori¬ zontal line eventually received commissions, and all men below failed. 91^3% of the men above C-f- received commissions. 58p3 of the men below C-f- failed. (From p. 9 of pamphlet, Army Menial Tests.) 68 AMERICAN INTELLIGENCE army tests to the training camp groups. Each solid vertical bar represents all the men of a given letter grade. There were, of course, more A and B men than C— and D men, but for purposes of comparison each letter group is treated as a whole (100%). All men above the horizontal line re¬ ceived commissions at the close of the Officers’ Training Camp, and all the men below the line failed to receive com¬ missions. Figure 28 shows clearly that about nine out of ten A and B men eventually received commissions, while for C— and D men the chances were very slight. Figure 29 shows in a general way the manner in which groups selected by various outside criteria contributed to the upper end of the intelligence scale (A and B), and to the lower end of the scale (C—, D and E). We are already familiar with the differences between ranks shown in this figure. Sixty company commanders were asked to designate their ten “best” and ten “poorest” privates. The results of this comparison of the ten “best” and ten “poorest” pri¬ vates also appears in Figure 29. The other two classifica¬ tions, “men of low military value,” and “unteachable men” represent a type of officers’ rating. In general the test re¬ sults check with officers’ ratings independently made. The results presented, showing the relation between rank and intelligence, and officers’ ratings and intelligence, indi¬ cate clearly a certain positive relationship between tests and military success. Recognizing the fact that intelligence is only one factor tending to produce military success, we accept the results of checking the tests against military cri¬ teria as additional proof that the tests are genuine measures of intelligence. A rough but rather interesting check of the army in¬ telligence tests may be made by glancing at the scores of men classified by occupations. Figure 30 gives the range of the intelligence scores of the middle 50% of various occu- A AND B D, D~, E C+, C, C— COMMISSIONED 8819 OFFICERS O. T I . S. STUDENTS 9240 SERGEANTS 3393 I CORPORALS 4093 TEN BEST PRIVATES GOG WHITE RECRUITS 77299 L ]! DISCIPLINARY CASES 491 Camp Dix czzzzzi i “TEN POOREST” PRIVATES 582 1 “MEN OF LOW MILITARY VALUE” 147 Camp Custer r i i UNTEACHABLE MEN 255 Camp Hancock CZD Figure 29. Comparison of army tests records with various inde¬ pendent criteria. The distributions of scores by ranks are shown in another way in Figure 27. The men were rated as “ten best,” “ten poorest,” “of low military value,” and “unteachable” by their officers. The chart shows a close correspondence between the brief psychological examinations and officers’ judgments made after weeks of observation. (From p. 10 of pamphlet, Army Mental Tests.) 69 L-P- 1 J 9— 1 1 c + I B C- 1 Laborer .. Gen. miner Teamster . Barber ... Horseshoer .... Bricklayer. Coolc . . . Baker. Painter. Gen. blacksmith. Gen. carpenter.. Butcher . Gen. machinist. Hand riveter .. \ Tel. and tel. lineman ... Gen. pipefitter. Plumber -. Tool and gauge maker. Gunsmith. Gen. mechanic. Gen. auto repairman .. *Auto engine mechanic . Auto assembler. Ship carpenter . Telephone operator _... C < Concrete const, foreman Stock-keeper .... Photographer. Telegrapher . R. R. clerk. Filing clerk. Gen. clerk. Army nurse Bookkeeper . B Dental officer. Mechanical draftsman Accountant. Civil engineer. Medical officer. ■^Engineer officer_..... ~ | 3 -| “ B Figure 30. Success in civil occupations compared with army test records. The figure shows in a general way the correlation be¬ tween intelligence as measured by the army tests and intelli¬ gence as indicated by position in civil life. (From p. 829, Memoir XV.) 70 AMERICAN INTELLIGENCE 71 pational groups, the position of the man half way up or down the scale being marked by a short vertical line. In some of the occupational groups, the number of cases is small, and the classification itself may be at fault in many instances, but the chart nevertheless shows a general ten¬ dency of the sort we should expect to find, for a process of intellectual selection occurs in industry which is just as rigid as that occurring in our public schools. We have briefly inspected the different sorts of evidence from independent fields which indicate that the army tests were genuine measures of intelligence. Further discussion of this point is unnecessary. The army tests were not in¬ fallible, and mistakes in classifying men were undoubtedly made, but the tests were satisfactory rough measures. When used in comparing groups as the tests are in this study, their reliability is increased, for errors in measure¬ ment would tend to equalize in each group. We should ex¬ pect the same percentage of error in classifying recruits born in Russia as we should recruits born in Sweden. Thus we use the tests as general measures of group tendencies, and as group measurements the tests have a sufficiently high degree of reliability to make positive conclusions possible. In the foregoing pages the army tests have been de¬ scribed briefly, the method of treating the results from vari¬ ous examinations bv the combined scale reviewed, certain misconceptions discussed, and a few T bits of supporting evidence assembled. Persons interested in a further study t/ of the tests should consult a little book by Yoakum and \erkes 1 , or, Memoir XV. We may now proceed to analyze American intelligence by treating the psychological exam¬ inations made in the army as a mental census of the population of this country. J C. S. Yoakum and R. M. Yerkes. Army Mental Tests. New York, 1920, pp. 203 PART II STATISTICAL ANALYSIS OF THE ARMY TEST RESULTS PART II. SECTION I THE PRINCIPAL SAMPLE All results from the psychological examinations in the camps were sent to Washington. It was impracticable as well as undesirable to tabulate the results in the case of every man examined. An intelligent selection or sampling of cases will give results more nearly typical of the country at large, than the entire group tested, which would be un¬ duly weighted for the more populous States, for camps giving the greatest number of examinations, particular draft quotas, etc. In order to obtain a sample for the white draft and the negro draft, cases were ‘'randomly” (or better, impartially) selected in accordance with certain definite principles. The groups were as follows: Group I: White draft, pro-rated, by States. . . . 41,278 Selected from 15 National Army camps, according to the State from which drafted, and according to the ratio of one recruit per thousand white male population. Group II: White draft, additional, by States. . . 14,684 Additional selection of cases in¬ tended to bring the total represen¬ tation from each State up to 1,000 cases. Group III: White draft, additional, by camps. . . 40,392 Additional selection of cases in¬ tended to bring the entire samp¬ ling of the white draft up to ap¬ proximately 100,000 cases. 75 AMERICAN INTELLIGENCE IV: Negro draft, pro-rated, by States. . . . 19,992 Selected in the same manner as Group I. V: Negro draft, additional, for Northern States. 5,400 Chosen to represent the negro draft of the north. groups were selected to meet other problems as VI: White officers.15,528 Selected proportionately to their occurrence in different arms of the service, with some additions to sup¬ plement the smaller arms, and the Medical Department. Group VIII: White established organizations. 24,205 Selected to provide comparison be¬ tween enlisted men of various arms of the service. Group X: Special experimental Group. 1,047 Randomly selected individuals of the white draft born in English speaking countries, who were given both alpha and beta, and, where possible, the Stanford-Binet ex¬ amination. These groups selected as representative of the country at large were analyzed by the Hollerith system of mechan¬ ical sorting. In this study we are concerned with: Groups I, II and III representing the white draft, Groups IV and V representing the negro draft, Group VI representing the white officers, and Group X, the special experimental group. 76 Group Group Other follows: Group SECTION II ANALYSIS OF THE MAIN GROUPS OF THE PRINCIPAL SAMPLE The tabulations in Memoir XV showing the distribution of scores on each type of examination of the white draft (Groups I, II, and III), the negro draft (Groups IV and V), and the white officers (Group VI) have been re-figured on the combined scale. The following tables were used: For the white draft: Alpha: Table 183 (p. 666) for men who took alpha only.67,254 Beta: Table 184 (p. 666) for men who took beta only, or alpha and beta.23,547 Individual: Table 185 (p. 667) for men who took Stanford- Binet examination only, or follow¬ ing alpha, following beta, or follow¬ ing alpha and beta. 1,246 Table 186 (p. 667) for men who took point scale examination only, or following alpha, following beta, or following alpha and beta. 689 Table 187 (p. 668) for men who took perform¬ ance scale examination only, or fol¬ lowing alpha, following beta, or fol¬ lowing alpha and beta. 1,237 Total Cases. 93,973 77 78 AMERICAN INTELLIGENCE For the negro draft: Alpha: Table 239 (p. 716) for men who took alpha only. 8,429 Beta: Table 241 (p. 717) for men who took beta only, or alpha and beta.14,350 Individual: Table 242 (p. 717) for men who took Stanford- Binet examination only, or follow- ing alpha, following beta, or follow¬ ing alpha and beta. 403 Table 229 (p. 711) for men who took point scale examination only, or following alpha, following beta, or following alpha and beta. 390 Table 228 (p. 710) for men who took perform¬ ance scale examination only, or fol¬ lowing alpha, following beta, or fol¬ lowing alpha and beta. 32 Total Cases. 23,604 For White officers: Alpha: Table 182 (p. 665) for all officers who took examination alpha only. 15,544 It will be remembered that examination beta was given to all men who had been selected as illiterate or non-Eng¬ lish speaking before examination alpha was given, and also to those who took alpha and failed to make a reliable score. In the same way individual examinations were given to the lowest scoring cases in beta. Consequently, in figuring the results, if a man has taken both alpha and beta, his alpha score is disregarded and his beta score taken.' In the same way, if he took both alpha and beta, and was then given an AMERICAN INTELLIGENCE 79 individual examination, his alpha and beta scores are dis¬ regarded, and his score on the individual examination is taken as expressing the best measure of his intelligence. In other words we use alpha as our measure if alpha only was given, beta as our measure if beta alone, or alpha and beta were given, and the individual examination as our measure if it was given at all, on the assumption that the most re¬ liable test of a man was the last one given. The distribution of the scores of the white officers, white draft and negro draft is shown in Table 1. The first three columns are read in this manner: six officers and one re¬ cruit measured between 24.0 and 24.9 on the combined scale, one hundred and six officers and eighteen recruits be¬ tween 23.0 and 23.9, etc. The first three columns show the actual distributions, i. e. six officers out of 15,543, one re¬ cruit out of 93,955, etc. The last three columns show each of these distributions arranged according to the number in each ten thousand who scored at each class interval of the combined scale. The last three columns read in this man¬ ner: four officers in ten thousand test between 24.0 and 24.9 on the combined scale; sixty-eight officers and two re¬ cruits in ten thousand score between 23.0 and 23.9 on the combined scale, etc. The “proportion in each ten thousand” may be read as a percentage by pointing off two decimal places. Table 1 also shows the average score of the white officers, white draft and colored draft on the combined scale to be 18.84, 13.54 and 10.41 respectively. The standard devia¬ tion (S. D.) is also shown. An average has little significance without reference to a measure of variability of the series of measurements on which it is based. The conventionally accepted measure of variability is the standard deviation, which is derived by taking the square root of the average of the squares of the individual deviations from the aver- Table No. 1 Distribution of the intelligence scores of the main groups of the principal sample on the combined scale. COMBINED SCALE INTERVALS ACTUAL DISTRIBUTION PROPORTION IN EACH TEN THOUSAND WHITE OFFICERS WHITE DRAFT NEGRO DRAFT WHITE OFFICERS WHITE DRAFT NEGRO DRAFT 24.0-24.9 6 1 4 23.0-23.9 106 18 68 2 • • • • 22.0-22.9 612 124 1 394 13 • • • • 21.0-21.9 1648 444 7 1060 48 3 20.0-20.9 2522 1006 16 1623 107 7 19.0-19.9 2836 1804 35 1824 192 15 18.0-18.9 2698 2996 81 1736 319 34 17.0-17.9 2155 4687 172 1387 499 73 16.0-16.9 1454 6847 330 935 729 140 15.0-15.9 837 9328 600 538 993 254 14.0-14.9 412 12019 1031 265 1279 437 13.0-13.9 179 14659 1793 115 1560 760 12.0-12.9 60 14002 2572 39 1490 1090 11.0-11.9 14 9481 2951 9 1009 1251 10.0-10.9 3 6227 3187 2 662 1351 9.0-9.9 1 4433 3319 1 472 1406 8.0-8.9 2876 2891 • • • • 306 1225 7.0-7.9 1683 2149 179 911 6.0-6.9 814 1315 87 557 5.0-5.9 334 684 36 290 4.0-4.9 122 302 13 128 3.0-3.9 37 112 4 48 2.0-2.9 11 38 1 16 1.0-1.9 2 10 .... .... 4 No. cases. 15543 93955 23596 Average. 18.84 13.54 10.41 S. D. 2.10 2.92 2.79 80 1800 -1 H fa < ffa Q O « O fa l H fa <: Q fa H fa C H O H C/3 fa fa CJ l-H fa fa o w H -H 3 £ 43 1 p-fa 43 -H f-t fa fa co o fa 43 H -M «+-l y fa Sh "fa 43 ""fa 43 O co be 43 fa ,"fa 43 fa • pH Ti •fa d fa a fa 0 43 -M CO 3 4-4 fa "fa 43 co 43 4-> *4 •pH rfa fan 43 £ f-t •\ ifa CO l-H c n fa • pH 43 G CO "fa • 43 CO fa • pH 43 rfa rH fa s 0 s C Crs a • n "3 s a o • rH > * > £ a) a rfa ?H Eh "fa . 43 aj ?h a • pH 2 co ^ § G ^fa ^ S g a -m fa ^ a ■H fa rfa (U biO _ O h 4j H-t H < J25 I I I I I jj i o H-l O H 03 Sh ' [ 'i 5-1 P_, ^ - a Sh G a o £ ■*-> 'Sc =4H «3 5h r 3 *C D 0 • pH -M r* pH c3 fee CO D • pH BO t—( 3 rn rH G CB H PH Vi CO Oj V D -t-J .O -w Vi c c po pH O V • CH O 0 P-H 0 P-H C !> (D -M -> ±0.1695 (4.5) 1 o bo o ±0.1592 (5.6) -1.45 ±0.1488 (9.8) -2.05 ±0.1691 (12.1) -1.78 ±0.1462 (12.2) 135 Table No. 21 Differences between IRELAND and other countries Number of cases 658 Average score 12.32 Standard deviation 2.60 England United States Canada Germany Holland Denmark Scotland Sweden Norway Belgium Austria Turkey Greece Russia Poland Italy +2.55 ±0.1094 (23.3) + 1.45 ±0.0688 (21.2) + 1.34 ±0.0896 (14.9) + 1.56 ±0.1157 (13.5) + 2.00 ±0.1523 (13.1) + 1.37 ±0.1079 (12.7) +2.02 ±0.1617 (12.5) +0.98 ±0.0918 (10.6) + 0.66 ±0.0961 (6.8) +0.47 ±0.1590(2.9) Difference unreliable. — 0.05 ±0.1269 (0.4) Difference unreliable. — 0.30 ±0.1132(2.7) Difference unreliable. -0.42 ±0.0972 (4.3) -0.98 ±0.0791 (12.4) -1.58 ±0.1126 (14.0) -1.31 ±0.0739 (17.7) 136 Table No. 22 Differences between AUSTRIA and other countries Number of cases 301 Average score 12.27 Standard deviation 2.75 England + 2.60 Enited States +1.50 Holland + 2.05 Canada + 1.39 Scotland + 2.07 Germany + 1.61 Denmark + 1.42 Sweden + 1.03 Norway + 0.71 Belgium + 0.52 Ireland + 0.05 Turkey -0.25 Greece -0.37 Russia -0.93 Poland -1.53 Italy -1.26 =*=0.1368 (10.0) ±0.1071 (14.0) ±0.1731 (11.8) ±0.1215 (11.4) ±0.1813 (11.4) ±0.1418 (11.3) ±0.1355 (10.5) ±0.1231 (8.4) ±0.1264 (5.6) ±0.1789 (2.9) Difference unreliable. ±0.1269 (0.4) Difference unreliable. ±0.1398 (1.8) Difference unreliable. ±0.1272 (2.9) Difference unreliable. ±0.1139 (8.2) ±0.1393 (10.9) ±0.1104 (11.4) 1ST Table No. 23 Differences between TURKEY and other countries Number of cases 423 Average score 12.02 Standard deviation 2.75 England +2.85 ±0.1241 (22.9) United States + 1.75 ±0.0904 (19.3) Canada + 1.64 ±0.1071 (15.3) Germany + 1.86 ±0.1297 (14.3) Holland +2.30 ±0.1632 (14.1) Denmark + 1.67 ±0.1228 (13.6) Scotland + 2.32 ±0.1720 (13.5) Sweden + 1.28 ±0.1089 (11.8) Norway + 0.96 ±0.1126 (8.5) Belgium + 0.77 ±0.1695 (4.5) Ireland +0.30 ±0.1132(2.7) Difference unreliable. Austria +0.25 ±0.1398 (1.8) Difference unreliable. Greece — 0.12 ±0.1135(1.0) Difference unreliable. Russia -0.68 ±0.0984 (6.9) Poland -1.28 ±0.1269 (10.1) Italy -1.01 ±0.0944 (10.7) 138 Table No. 24 Differences between GREECE and other countries Number of cases 572 Average score 11.90 Standard deviation 2.45 England +2.97 United States + 1.87 Canada + 1.76 Germany + 1.98 Denmark + 1.79 Holland +2.42 Sweden + 1.40 Scotland +2.44 Norway + 1.08 Belgium + 0.89 Ireland + 0.42 Austria + 0.37 Turkey + 0.12 Russia -0.56 Poland -1.16 Italy -0.89 ±0.1097 (27.0) ±0.0693 (26.9) ±0.0900 (19.6) ±0.1159 (17.1) ±0.1081 (16.5) ±0.1525 (15.9) ±0.0921 (15.2) ±0.1619 (15.0) ±0.0965 (11.2) ±0.1592 (5.6) ±0.0972 (4.3) ±0.1272 (2.9) Difference unreliable. ±0.1135 (1.0) Difference unreliable. ±0.0795 (7.1) ±0.1129 (10.3) ±0.0744 (11.9) 139 Table No. 25 Differences between RUSSIA and other countries Number of cases 2340 Average score 11.34 Standard deviation 2.83 United States + 2.43 England +3.53 Canada +2.32 Sweden + 1.96 Denmark + 2.35 Germany + 2.54 Holland +2.98 Norway + 1.64 Scotland + 3.00 Ireland + 0.98 Belgium + 1.45 Austria + 0.93 Greece + 0.56 Turkey + 0.68 Poland -0.60 Italy -0.33 ±0.0400 (60.7) ±0.0940 (37.5) ±0.0700 (33.1) ±0.0727 (26.9) ±0.0921 (25.5) ±0.1012 (25.1) ±0.1417 (21.0) ±0.0795 (20.6) ±0.1517 (19.8) ±0.0791 (12.4) ±0.1488 (9.8) ±0.1139 (8.2) ±0.0795 (7.1) ±0.0984 (6.9) ±0.0977 (6.1) ±0.0483 (6.9) 140 Table No. 26 Differences between ITALY and other countries Number of eases 4009 Average score 11.01 Standard deviation 2.60 LTnited States +2.76 ±0.0285 (96.8) England +3.86 ±0.0897 (43.0) Canada +2.65 ±0.0641 (41.3) Sweden + 2.29 ±0.0671 (34.1) Denmark +2.68 ±0.0877 (30.5) Germany + 2.87 ±0.0972 (29.5) Norway + 1.97 ±0.0729 (27.0) Holland +3.31 ±0.1388 (23.9) Scotland +3.33 ±0.1491 (22.3) Ireland + 1.31 ±0.0739 (17.7) Belgium + 1.78 ±0.1462 (12.2) Austria + 1.26 ±0.1104 (11.4) Turkey - + 1.01 ±0.0944 (10.7) Greece + 0.89 ±0.0744 (11.9) Russia ‘ + 0.33 ±0.0483 (6.9) Poland — 0.27 ±0.0936 (2.9) Difference unreliable. 141 Table No. 27 Differences between POLAND and other countries Number of cases 382 Average score 10.74 Standard deviation 2.59 United States +3.03 England +4.13 Canada + 2.92 Germany + 3.14 Denmark + 2.95 Sweden +2.56 Holland + 3.58 Scotland +3.60 Norway +2.24 Ireland + 1.58 Belgium + 2.05 Austria + 1.53 Greece + 1.16 Turkey + 1.28 Russia + 0.60 Italy + 0.27 ±0.0896 (33.8) ±0.1236 (33.4) ±0.1064 (27.4) ±0.1291 (24.3) ±0.1221 (24.1) ±0.1082 (23.6) ±0.1628 (21.9) ±0.1716 (20.9) ±0.1119 (20.1) ±0.1126 (14.0) ±0.1691 (12.1) ±0.1393 (10.9) ±0.1129 (10.3) ±0.1269 (10.1) ±0.0977 (6.1) ± 0.0936 (2.9) Difference unreliable. AMERICAN INTELLIGENCE 143 Tables 11 to 27 give the most accurate interpretation of the differences found between the various nativity groups that it is possible to derive from the army data. It is desir¬ able, however, to attempt to estimate the meaning of these differences in terms of standards which have some popular significance. For this reason, the combined scale distribu¬ tions in this study have been converted into estimates of «/ the per cent, of A and B men in each group, and the per cent, of D, D— and E men. It should be remembered that the army letter ratings are arbitrary ratings and have no real significance aside from the tests from which they were derived. The army rating “A” represents a certain score on the tests that should have been reached by 4% or 5% of the whole army group. In the same way the rating “B” was fixed so as to include the next 8% or 10%. It is of course absurd to deplore the fact that only 4% or 5% of the army were A men, when A was fixed so that only 4% or 5% could receive that rating. At the other end of the scale, the ratings D and D— were fixed so that they would include approximately 20% of the total group, and the E rating was reserved for those recom¬ mended for development battalions, special service organi¬ zations, rejection or discharge. The estimates made at the time the examinations were being standardized proved to be about right. The A and B groups which should have in¬ cluded 12% to 15% of the draft actually included 12%, and the D, D— and E groups, which should have included 20% to 25%, actually included 24%. Table 28 gives the per cent, of cases in each nativity group who would be classified A or B according to the criterion of the upper 12% of the total unselected white draft. Table 29 gives the per cent. w r ho would be classified as D, D — and E according to the criterion of 24% of the unselected group. The relations between the various nativ- 144 AMERICAN INTELLIGENCE ity groups given in Tables 28 and 29 are shown graphically in Figure 37. Another criterion that probably represents intellectual ability of a high order is the per cent, at or above the average of the white officers. The classification of the nativity groups according to this criterion is given in Table 30. At the other end of the scale, a criterion of inferiority that has a certain social significance is the per cent, at or below the average of the negro draft. Table 31 shows the nativity groups classified according to this criterion. The results given in Tables 30 and 31 are shown graphically in Figures 38 and 39. It is not possible to determine accurately the percentage Table No. 28 Per cent, of each nativity group in the A and B groups England. 19.0 Native Born White Draft. 13.2 Scotland. 13.1 Holland. 12.4 Total White Draft. 12.0 Canada. 11.1 Germany. 10.1 Denmark. 7.0 Sweden. 5.9 Norway. 5.3 Foreign Born White Draft. 4.6 Ireland. 4.3 Austria. 4.1 Turkey. 4.0 Russia. 3.3 Belgium. 2.9 Greece. 2.2 Italy. 1.5 Colored Draft. 1.4 Poland. 1.1 AMERICAN INTELLIGENCE 145 of feeble-mindedness in each group. The selection by the draft boards probably excluded all idiots and many im¬ beciles. The diagnosis of the border-line cases of feeble¬ mindedness is, in the last analysis, a social diagnosis, and can not be based on intelligence tests alone. It has been found, however, that a “mental age” of eight indicates an order of intelligence so low that the individual has diffi- culty in adjusting himself to his environment adequately. Table 32 gives a conservative estimate of the per cent, in each nativity group below T an approximate “mental age” of eight. The percentages in Table 32 are shown graphically in Figure 40. Table No. 29 Per cent, of each nativity group in the D, D —, and E groups England. 8.8 Holland. 12.0 Scotland. 13.5 Germany. 16.2 Denmark. 17.0 Native Born White Draft. 21.0 Canada. 21.6 Sweden. 23.2 Total White Draft., . 24.1 Norway. 29.0 Belgium. 29.3 Ireland. 38.1 Austria. 38.4 Turkey. 43.6 Greece. 44.6 Foreign Born White Draft. 44.6 Russia. 55.7 Italy. 60.5 Poland. 63.8 Colored Draft. 67.5 ENGLAND ////////777/7777777/77///77/7///77/77777//77///7777/m HOLLAND V/7//7/7/7A//777777Z7A7777//A7777/77777/777/77 SCOTLAND 777/777777^7777/777^777/7/7/7777777777777777/ GERMANY 77/77777777777//777777777777/7777//77/7777///7///A DENMARK NATIVE BORN WHITE DRAFT 77/Z7/7/7/77/////7///////7///7//7////7///77A CANADA 7/////////7777/////7///////////7////7//7///7/7777. SWEDEN 7///z7/////y///////////7//z/77/zzzA TOTAL WHITE DRAFT 777777/77////////////////////777/7//77777777_ NORWAY //////////////////////////////////////////////A BELGIUM m/////77///////77777777///////////////7 "~i IRELAND 7 777777777777///////7//777/7//m^. AUSTRIA W777/77////////////////////777777 TURKEY r////////////////77/77772Z^77Zm GREECE 17//7/Z/ZZZZ////7//7/7/777/7 FOREIGN BORN WHITE DRAFT /////////////////////////m. /////////////////////////AX /7Z777/77777777777\ POLAND NEGRO DRAFT '///y////7m/£Z7 D,D— & E //////////z c,ce?c+ 90 100 A.6B 146 Figure 37. The relative standing of the various nativity groups in the proportions of A and B men, and D, D—, and E men. In interpreting this chart, it should be remembered that A and B, and D, D—, and E do not represent absolute intelligence stand¬ ards, but rather standards arbitrarily fixed. In this case the standards were fixed by the 93,955 individuals making up the sample of the total white draft, A and B representing scores obtained by the upper 12% of this group, while D, D—, and E represent scores obtained by the lower 24%. The comparison is relative and not altogether reliable, for it fails to take into con¬ sideration the average, the number of cases, and the variability. Tables 28, 29, and this chart have been presented for the convenience of the reader, and to supplant Table 217, and Fig¬ ure 19, on pages 697 and 698 of Memoir XV, which are not based on combined scale results. 147 Table No. 30 Per cent, of each nativity group at or above the average of the white officers England. 6.2 Scotland. 4.8 Native Born White Draft. 4.6 Holland. 4.3 Total White Draft. 4.1 Canada. 3.5 Germanv. 3.4 Austria. 1.5 Sweden. 1.4 Foreign Born White Draft. 1.3 Ireland. 1.2 Denmark. 1.1 Norway. 1.0 Turkey. 0.8 Russia. 0.8 Belgium. 0.3 Italy. 0.3 Greece. 0.3 Poland. 0.1 Figure 38. The proportion of each nativity group obtaining intel¬ ligence scores at or above the average of the white officers (18.84). Reference to Figure 31 will show that this criterion indicates a relatively high order of intelligence. In comparing this Figure with Figures 37 and 39, it should be noted that each of the three figures has been drawn to a different scale. Our in¬ terpretation of these figures must be made with caution, for we are comparing extremes of the distribution curves without ref¬ erence to the position of the average, the variability about the average, or the number of cases in the various groups. 148 ENGLAND SCOTLAND NATIVE BORN WHITE DRAFT HOLLAND TOTAL WHITE DRAFT CANADA GERMANY AUSTRIA SWEDEN FOREIGN BORN WHITE DRAFT IRELAND DENMARK NORWAY TURKEY RUSSIA BELGIUM ITALY GREECE POLAND 0°/o 1% 2% 3 7o *% 5% 6% 7% . 0% 1% 2% 3% 4% 5% 6%' 7% 149 Table No. 31 Per cent, of each nativity group at or below the average of the negro draft England. 4.3 Holland. 4.9 Germany. 6.5 Scotland. 6.8 Denmark. 7.5 Native Born White Draft. 7.6 Sweden. 11.5 Canada. 11.6 Total White Draft. 13.7 Norway. 15.2 Belgium. 16.0 Ireland. 22.8 Austria. 24.5 Greece. 27.1 Turkey. 28.2 Foreign Born White Draft. 29.5 Russia. 39.0 Italy. 42.3 Poland. 46.0 Figure 39. The proportion of each nativity group at or below the average of the negro draft. Reference to Figure 31 will show that this criterion indicates a rather low order of intelligence. If 50% of any nativity group were at or below the negro aver- age, the two distributions would be approximately identical. Russia shows 39% below the negro average, Italy 42.3%, and Poland 46%. 150 20 % 30 7o 40% 50% ENGLAND HOLLAND GERMANY SCOTLAND DENMARK NATIVE BORN WHITE DRAFT SWEDEN CANADA TOTAL WHITE DRAFT NORWAY BELGIUM IRELAND AUSTRIA GREECE TURKEY FOREIGN BORN WHITE DRAFT RUSSIA ITALY POLAND 07c 10 % 0 % 10% 20% 30% 40% 50% > — , - ... , , .. — —---- . -i 151 Table No. 32 Per cent, of each nativity group below the approximate “mental age” of eight Holland. 0.1 Germany. 0.2 Denmark. 0.2 England. 0.3 Scotland. 0.4 Sweden. 0.4 Norway. 0.8 Canada. 0.9 Native Born White Draft. 1.1 Total White Draft.,. 1.4 Belgium. 1.6 Ireland. 1.9 Greece. 2.3 Austria. 2.7 Turkey. 3.1 Foreign Born White Draft. 3.2 Russia. 5.0 Italy. 5.2 Poland. 6.8 Colored Draft. 10.0 Figure 40. Proportion of each nativity group testing below the ap¬ proximate “mental age” of eight. This criterion indicates intelli¬ gence of a very low order. These individuals are probably capa¬ ble of adjusting themselves only to the simplest form of environ¬ ment, occupation and conditions of living. Few of them would be able to manage their affairs with ordinary prudence. Many of them should be in custodial institutions. 152 HOLLAND GERMANY DENMARK ENGLAND SCOTLAND SWEDEN NORWAY CANADA NATIVE BORN WHITE DRAFT TOTAL WHITE DRAFT BELGIUM IRELAND GREECE AUSTRIA TURKEY FOREIGN BORN WHITE DRAFT RUSSIA ITALY POLAND NEGRO DRAFT I ■ a 153 SECTION VII RELIABILITY OF THE RESULTS The results of the army psychological examination fig¬ ured by means of the combined scale give us the best avail¬ able measures of the intelligence of the various groups ex¬ amined. Do these results apply to the army as a whole? The logic underlying the answer to this question is the same as that underlying the judgment of the whole by one of its parts. The tea taster samples the tea to be graded. He does not need to brew a whole bale of tea to find its worth. In this experiment we have sampled the entire army by taking 15,543 white officers, 93,955 white recruits and 23,596 negro recruits. Our group of white recruits was sub¬ divided into 81,465 native born and 12,492 foreign born. No one would hesitate to accept the results of the 81,465 native born as typical of the army as a whole. If we continued sampling indefinitely, our results would increase in reliability only as the square root of the number of cases, and 81,465 cases constitute a sample that is a luxury from the point of view of size. In the same way, no one could seriously question the reliability of our sampling of 15,543 officers, 23,596 negroes and 12,492 foreign born as typical of the remainder of the officers, negroes, and foreign born whites in the army. The results from the 81,465 cases in the native born white draft may be taken as typical of white males between the ages of 21 and 31 and above the idiot or imbecile grade. In making our comparisons between other groups, we know that the Selective Service Act called all men to the colors impartially. The same regulations drew the Italians, the negroes, the native whites, the Polish, and all other 154 AMERICAN INTELLIGENCE 155 groups into the army. The method of sampling all the psychological records again drew these cases impartially. If our theory of sampling is correct, then we may accept the army results as very approximately typical of the male population as a whole. For instance, our figures in Table 3 show characteristic differences in the average score on the army tests of foreign born individuals in this country from 0 to 5 yrs., and those in this country from 6 to 10 yrs., etc. The same factors which determined the sampling of the 3,576 cases in this country 0 to 5 yrs. determined the sampling of the 4,287 cases in this country 6 to 10 yrs. As long as the principles of sampling are the same, we may take our small sample as typical of the group as a whole. The results of the psychological tests of foreign born individuals classified according to length of residence, taken as typical of our foreign born population as a whole, indicate definitely that the average intelligence of succeed¬ ing waves of immigration has become progressively lower. Immigrants coming to this country between 1913 and 1917 have a lower average intelligence than those coming to this country in the years 1908 to 1912. The group coming to this country in the years 1903 to 1907 had a higher average intelligence that the 1908 to 1912 group, and a lower average intelligence than immigrants coming to this country in the years 1898 to 1902. In drawing these conclusions we are taking the groups examined in the army as typical of the corresponding groups in the entire population. During the years 1913 to 1917, about 3 1/3 millons of immigrants came to this country. We are actually using 3,576 cases or about 0.1% as typical of the whole group. It may very properly be objected that this is too small a sample on which to base definite conclusions. We must therefore state our conclu- 156 AMERICAN INTELLIGENCE sions less dogmatically, and with the proviso that if the groups examined in the army are typical of the immigra¬ tion coming to this country in the same periods, then we know that our more recent periods of immigration give us an average intelligence which becomes progressively lower and lower. This tentative conclusion will be modified by any evidence which tends to support the hypothesis made. The same kind of argument from the sample to the group holds in our interpretation of the differences in the average intelligence scores of groups in our army born in different countries. For instance, in the period under consideration from 1887 to 1917 there have been about 3 7/8 millons of Italians, and over 3 million Russians who have come to this country. We are actually using 4,009 Italians and 2,340 Russians as typical samples of these groups. Of course no one would maintain that these 4,009 Italians are typical of the population of Italy. There are so many vari¬ able factors determining immigration that the immigrants can not themselves be taken as representative of the country as a whole. The question at issue is that of accept¬ ing 4,009 Italians as typical of the 3 7/8 millions who have come to this country since 1887. The chief claim to reli¬ ability of our sample from each country is the fact that the sample was drawn at random from the army group, and the fact that the Selective Service Act drew the men from each country impartially. SECTION VIII THE RACE HYPOTHESIS The results of the examination of the nativity groups suggest immediately that the race factor may underlie the large differences found. If we do find the common factor of race underlying the differences between the various nativ¬ ity groups, it will give our results much greater reliability, for the chance factors of sampling particularly inferior or superior groups in the small nativity samples would dis¬ appear in combination. Our figures are based on country of birth and no statistics are available for race. The race hypothesis must therefore be examined indirectly. Writers on immigration, for the most part, divide the countries of Europe into two groups (1) Northern and Western, and (2) Eastern and Southern, and usually as¬ sume that the immigration from Northern and Western Europe has been mostly Nordic. This traditional method is open to two very serious objections. In the first place, the classification fails to differentiate the Alpine and Medi¬ terranean race groups. In the second place the assumption that the immigration from Northern and Western Europe was mostly of a pure Nordic type is unwarranted, for this classification includes Germany and Ireland, two countries that have contributed very largely to our immigration in the past. The following figures show the size of the Irish and German immigration: 157 158 AMERICAN INTELLIGENCE DECADE TOTAL IMMIGRATION PER CENT. FROM IRELAND PER CENT. FROM GERMANY PER CENT. FROM IRELAND AND GERMANY 1820-1830 143,439 35 % 5 % 40 % 1831-1840 599,125 35 % 25 % 60 % 1841-1850 1 , 171,251 46 % 25 % 71 % 1851-1860 2 , 598,214 35 % 37 % 72 % 1861-1870 2 , 314,824 19 % 34 % 53 % 1871-1880 2 , 812,191 15 % 26 % 41 % 1881-1890 5 , 246,613 12 % 28 % 40 % 1891-1900 3 , 844,420 11 % 14 % 25 % 1901-1910 8 , 795,386 4 % 4 % 8 % 1911-1920 5 , 735,811 2}4% 2 ^% 5 % These figures show clearly the fallacy of assuming that the immigration from Northern and Western Europe has been predominately Nordic, for Ireland is largely Mediterranean and Germany largely Alpine. If we wish to obtain even approximate estimates of the contributions of each of the three European races to our importations, it is necessary to abandon the Northern and Western, and Eastern and Southern classification and try another method. If it were possible to make even approxi¬ mate estimates of the percentage of Nordic, Alpine and Mediterranean blood in each of the European nations send¬ ing immigrants to this country, such approximate estimates would be very much superior to the present method. In collaboration with students of this subject, I have constructed Table 33 which contains tentative estimates of the present blood constitution of the countries sending im¬ migrants to this country. This table is, of course, only an approximation to the truth, and many persons will dis¬ agree with the estimates. For this reason, I am re-publishing in Table 34, Table 68, page 100, of the Statistical Abstract for the United States for 1920, which shows the arrivals of alien passengers and immigrants by nationalities and by Table No. 33 Tentative estimates of the proportion of Nordic, Alpine and Mediterranean blood in each of the European coun¬ tries. PER CENT. PER CENT. PER CENT. NORDIC ALPINE MEDITERR Austria-Hungary. 10 90 0 Belgium. 60 40 0 Denmark. 85 15 0 France. 30 55 15 Germany. 40 60 0 Greece. 0 15 85 Italy. 5 25 70 Netherlands. 85 15 0 Norway. 90 10 0 Sweden. 100 0 0 Russia (including Poland). 5 95 0 Poland. 10 90 0 Spain. 10 5 85 Portugal. 5 0 95 Roumania. 0 100 0 Switzerland. 35 65 0 Turkev (unclassified). 0 20 80 Turkey (in Europe) (including Serbia, Montenegro and Bulgaria). 0 60 40 Turkey (in Asia). 0 10 90 England. 80 0 20 Ireland. 30 0 70 Scotland. 85 0 15 Wales. 40 0 60 British North America. 60 40 0 159 Table No. 34 No. 68 .—ARRIVALS OF ALIEN PASSENGERS AND IMMIGRANTS, 1820 TO 1920: By Nationali¬ ties and by Decades. [Sources: Records of the Bureau of Statistics prior to 1896; subsequently, reports of the Commissioner General of Immigration, Department of Labor. The figures represent “alien passengers” from Oct. 1 , 1820, to Dec. 31, 1867; “immigrants” from Jan. 1 , 1868, to date. COUNTRY OF LAST PERMANENT RESIDENCE oct. 1, 1820, to sept. 30, 1830 oct. 1, 1830, to dec. 31, 1840 JAN. 1, 1841, TO DEC. 31, 1850 JAN. 1, 1851, TO DEC. 31, 1860. Austria-Hungary. Belgium. 27 22 5,074 539 77,262 434,626 4,738 3,749 76,358 951,667 Denmark. 169 1,063 45,575 152,454 France. 8,497 6,761 Germany. Greece 1 . Italy. 408 2,253 1,412 1,870 8,251 9,231 10,789 Netherlands. 1,078 Norway. Sweden. } 91 1,201 13,903 20,931 Russia, including Russian Poland 2 .... Spain 3 . 91 ) 646 656 1,621 Portugal 4 . \ 2,622 2,954 2,759 10,353 Rumania 1 . Switzerland. 3,226 4,821 4,644 25,011 Turkey in Europe 5 . United Kingdom: England. 22,167 2,912 50,724 73,143 2,667 207,381 263,332 3,712 780,719 385,643 38,331 914,119 Scotland. Ireland. Wales 6 . Total United Kingdom. 75,803 283,191 1,047,763 1,338,093 Europe, not specified. 43 96 155 116 Total Europe. 98,816 495,688 1,597,502 2,452,657 British North America 7 . 2,277 4,817 105 13,624 6,599 44 41,723 3,271 368 59,309 3,078 449 Mexico 7 . Central America. West Indies: Bermuda and Miquelon. South America. 3,834 531 12,301 856 13,528 3,579 10,660 1,224 Total America 7 . 11,564 33,424 62,469 74,720 Islands of the Atlantic. 352 103 337 3,090 China. 2 8 35 41,397 India 8 . Japan 8 . Turkey in Asia 8 . Other Asia. 8 40 47 61 Total Asia. 10 48 82 41,458 Total Oceania. 2 9 29 158 Total Africa. 16 52 55 210 All other countries. 32,679 69,801 52,777 25,921 Grand total. 143,439 599,125 1,713,251 2,598,214 included in “Europe, not specified,” prior to 1891-1900. includes also Finland after 1872. includes Canary and Balearic Islands after 1900. 4 Figures include the Azores and Cape Verde Islands after 1879, they being classed with Portugal so far as that country is separately shown. 160 JAN. 1, 1861, to june 30, 1870 TEAKS ENDED JUNE 30- 1871 to 1880 1881 TO 1890 1891 TO 1900 1901 to 1910 1911 to 1920 7,800 72,969 353,719 597,047 2,145,266 896,342 6,734 7,221 20.177 20,062 41,635 33,746 17,094 31,771 88,132 52,670 65,285 41,983 35,984 72,206 50,464 36,006 73,379 61,897 787,468 718,182 1,452,970 543,922 341,498 143,945 15,996 167,519 184,201 11,728 55,759 307,309 655,694 2,045,877 1,109,524 9,102 16,541 53,701 31,816 48,262 43,718 f 95,265 190,505 66,395 109,298 211,245 568,362 \ 230,679 249,534 95,074 4,536 52,254 265,088 593,703 1,597,306 921,957 ( 6,723 27,935 68,611 8,493 9,893 6,535 \ 23,010 69,149 89,732 14,559 53,008 13,311 23,286 28,293 81,988 33,149 34,922 23,091 2,562 118,202 77,098 568,128 460,479 657,488 271,094 388,017 249,944 38,768 87,564 149,869 60,053 120,469 78,601 435,778 436,871 655,482 403,496 339,065 145,937 11,186 17,464 13,107 1,042,674 984,914 1,462,839 745,829 865,015 487,589 210 656 10,318 4,370 1,719 18,350 2,064,407 2,261,904 4,721,602 3,703,061 8,136,016 4,376,564 153,871 383,269 392,802 2,631 179,226 742,185 2,191 5,362 1,913 746 49,642 219,004 96 210 462 1,183 8,112 17,159 9,043 13,957 29,042 35,040 107,548 123,424 1,396 928 2,304 3,059 17,280 41,899 166,597 403,726 426,523 42,659 361,808 1,143,671 3,446 10,056 15,798 64,301 123,201 61,711 23,166 20,605 21,278 26 4,713 2,082 26,855 129,797 83*837 8,398 77,393 79*389 308 622 6,669 28^370 11,059 5,973 64,609 123,823 68,380 86,815 243,567 192,559 221 10,913 12,574 8,793 12,973 13,427 312 229 437 1,343 7,368 8,443 15,232 1,540 1,299 1,749 33,654 1,147 2,314,824 2,812,191 5,246,613 3,844,420 8,795,386 5,735,811 includes Serbia, Bulgaria, and Montenegro prior to 1920; included in “Europe, not specified,” prior to 1891- 1900; also, after 1919, Czechoslovakia, Poland, and the Kingdom of the Serbs, Croats, and Slovenes. 6 Not separately stated prior to 1891-1900. immigrants from British North America and Mexico were not reported from 1886 to 1895, inclusive. 8 Not separately enumerated prior to 1899. 161 162 AMERICAN INTELLIGENCE decades from 1820 to 1920. My own Tables 9 and 10 give the distribution of the intelligence scores on the combined scale for the nativity groups we are studying. Anybody who disagrees with the estimates given in Table 33 may take these tables and split them according to any other estimates he wishes to make. However, minor changes in the proportions given in Table 33 would make very little dif¬ ference in the final results. The figures which follow are merely estimates based on Table 33. I am not claiming that these figures are absolutely reliable, but merely that they represent very much closer approximations to the truth than would be obtained from the Northern and West¬ ern, and Southern and Eastern classification. To obtain an estimate of the proportion of Nordic, Al¬ pine, and Mediterranean blood in our immigration since 1840, the immigration figures by countries, given in Table 34, have been cut according to the proportions given in Table 33 and re-combined into percentage estimates which are given in Table 35. These estimates show in general an immigration prior to 1890 which ran 40% or 50% Nordic blood. Since 1890, the proportion of Nordic blood has dropped to 20% or 25%, the Alpine stock now constituting about 50% of the total and the Mediterranean 20% or 25%. The proportions given in Table 35 are shown graphically in Figure 41. The percentage estimates, given in Figure 35 and shown graphically in Figure 41, should be considered in connection with the total volume of immigration for each decade given in Table 34 and shown graphically in Figure 42. Table No. 35 Estimate of the amount of Nordic, Alpine and Mediter¬ ranean blood coming to this country from Europe in each decade since 1840. DECADE TOTAL IMMIGRATION PER CENT. NORDIC BLOOD PER CENT. ALPINE BLOOD PER CENT MEDITERRANEAN BLOOD PER CENT. OTHERS AND UNCLASSIFIED 1841-1850 1,713,251 40.5 19.0 36.2 4.3 1851-1860 2,598,214 42.3 25.5 28.9 3.3 1861-1870 2,314,824 50.6 26.0 19.2 4.2 1871-1880 2,812,191 48.8 28.5 16.7 6.0 1881-1890 5,246,613 46.1 35.2 16.0 2.7 1891-1900 3,844,420 30.2 43.8 22.5 3.5 1901-1910 8,795,386 19.8 51.3 24.3 4.6 1911-1920 5,735,811 22.6 44.0 23.7 9.7 100 1841-1850 1851-1800 1801-1870 1871-1880 1881-1890 1891-1900 1901-1910 1910-1920 G3 o £ O rG cc G G o o 03 G G i 03 GS —i G 5^S s ■» b G «+-< '"g G G3 G s-< be 03 be Jh G 03 22 03 Sh c 03 03 G 03 03 -l-> G -u C/2 G 22 2G 03 G W G3 O o 2? f-< Q ° ^ >o o cN »o o 03 03 22 G +-> g3 o CO « o r—H >r ^ r\ J* 3 v\) IQ H - " 03 - O G O 03 G G3 03 c n • ph >* f-i 03 > o CJ 03 Jh 03 • pH G P 03 03 G • pH < r 3 G G G 03 i> • pH be G3 G u o o 2? 03 G qj s-h be G 03 u G -£ G ^ § cr 03 03 22 G 03 a G G G G O be • fm £ G G 03 hc PH G s- ?H 03 £3 03 03 03 «+-i 03 w • pH 03 r S & O G fc «-M 03 3 b g co ph C5s «3 £ o Sh «-W G £ G o t-l o 03 G G G 03 J-i ?H 03 P "H • pH -M G 03 ^5 03 03 4-a G 03 rG 22 G 03 G o 03 03 w • pH c , «+-i J-2) pH o G rr> 03 03 03 03 o g3 G 03 -? >> JS 03 -T f-i 03 03 G ?h be «\ -H «3 be 03 g ^5 ?H c ^ £ be 33 G iO G O '-M G Jh be 03 Sh 03 •s a rn O '—i Sh _03 t+4 J: be G • pH a o © G O 03 33 -W -G G 03 r~* • H Gh < 03 Stt 03 a 03 G 03 03 r—i b£ G 9.000,000 © a S3 •rH © ci u tc B S3 * o CO CO CO tM o i—< a a) S3 g <3 g c3 -M ■—I S3 « <0 _ S3 <» 33 co o o « , n “ ^ gj >>■5 r- 4^ ^ o H -r $0,0 , ? s o g S3 be o' _2 O ^*0 8 8 | g ^ g si ^ »H m Lj ^ O frt J= ° C g g T3 ^ .2 r d s: « -£ « °3 g b Sh « .££.33 CO © S co ®0 c 00 S O S3 ^ •pH GJ C 3 Q «« D J O ^ GO O « 2 » 00 5 3 ,2 - 3 2^^ ® ° K* «N ^ » 3 ^ 4J oj h- i i CJ S 3 c 3 co S3 ci O .pH • ^ CO 23 « • *—< £3 S Ov' do *H c3 bJD 33 Iffl ^ 2 05 ‘fn ^H H-> CO S3 <1 o o Ci 1—1 1-1 Ci T™ 1 1 S3 Oi o Oi 44> £ 4 -J t—l S3 S3 ci GO 33 3 o o so —H CO -M CO S3 c3 r-C 03 4 -> H 4 . 168 AMERICAN INTELLIGENCE In order to obtain an estimate of the intelligence of the three European races in this country, the distributions of the intelligence scores on the combined scale given in Table 9 were cut according to the proportions given in Table 33, and re-combined into Nordic, Alpine, and Medi¬ terranean groups. The final distributions are, of course, neither purely Nordic, Alpine, nor Mediterranean, but the sample of individuals we have thus selected as Nordic is undoubtedly more typical of the Nordic race type than it is of the Alpine and Mediterranean types. In the same way, the Alpine and Mediterranean groups are more typi¬ cal of each of these race types than they are of either of the other two. With thus much of apology for the method, I w T ill, in the following pages, simply for brevity of expres¬ sion, call these groups Nordic, Alpine, and Mediterranean. The reader must bear in mind that the distributions are only approximate samplings. The actual distributions on the combined scale of the three race groups so selected are given in Table 36, togeth¬ er with the proportions in each thousand. The distribution curves of the three groups are shown in Figure 43, in which the horizontal direction represents scores on the com¬ bined scale, and the vertical direction proportions in each thousand making each intelligence score. The differences found are very marked. The difference between the Nordic and Alpine group is 1.61 =*=0.042, a difference which is 38.3 times the probable error of the difference. The difference between the Nordic and Medi¬ terranean group is 1.85 ±0.042, a difference which is 44 times the probable error of the difference. The Alpine and Medi¬ terranean groups are, on the other hand, very much closer together, the difference being 0.24 ± 0.04, a difference which is 6 times the probable error of the difference. The easiest and most obvious objection that can be made Table No. 36 Analysis of the foreign bom white draft by races. Distri¬ butions of the intelligence scores of the Nordic, Alpine and Mediterranean groups. COMBINED ACTUAL DISTRIBUTION PROPORTION IN EACH SCALE THOSUAND INTERVALS NORDIC ALPINE MEDITER- NORDIC ALPINE MEDITER RANEAN RANEAN 24.0-24.9 23.0-23.9 .... .... .... .... .... .... 22.0-22.9 3 1 1 1 .... .... 21.0-21.9 8 5 2 2 1 .... 20.0-20.9 19 11 5 5 2 2 19.0-19.9 37 22 11 11 5 3 IS.0-18.9 71 47 26 21 10 6 17.0-17.9 135 90 55 39 19 13 16.0-16.9 238 155 103 09 32 24 15.0-15.9 357 240 180 103 51 43 14.0-14.9 469 372 296 136 78 71 13.0-13.9 566 544 468 164 114 111 12.0-12.9 528 650 591 153 136 141 11.0-11.9 371 628 590 107 132 140 10.0-10.9 260 595 569 75 125 136 9.0- 9.9 184 546 523 53 115 125 8.0- 8.9 112 403 376 32 85 90 7.0- 7.9 59 248 223 17 52 53 6.0- 6.9 26 124 108 8 26 26 5.0- 5.9 9 52 47 3 11 11 4.0- 4.9 3 19 16 1 4 4 3.0- 3.9 1 6 5 .... 2 1 2.0- 2.9 • • • • 2 1 .... 1.0- 1.9 .... • • • • .... .... .... No. of cases. 3456 4766 4196 Average.... 13.28 11.67 11.43 S.D. 2.70 2.87 2.70 169 ic/ure 43 . The distributions of the intelligence scores of the Nordic, Mediterranean and Alpine groups. This chart indicates clearly the superiority of the Nordic group. 73 . 9 % of the Nordic group are above the average Alpine, and 76 . 5 % of the Nordic group are above the average Mediterranean. The Mediterranean and Alpine groups are very similar, 52 . 3 % of the Alpines exceeding the average Mediterranean. AMERICAN INTELLIGENCE 171 to these findings is that the superiority of the Nordic group is due to the fact that it contains so many English speaking persons, and that lack of facility in the use of English is a handicap to the non-English speaking foreign born in the army tests. We have previously examined this hypothesis in connection with the argument establishing the fact that each succeeding five year period since 1902 shows a gradual deterioration in the intelligence of the immigrants examined in the army, and have definitely shown that the language factor does not distort the scores of the years of residence groups. There is, however, a considerable amount of wish¬ ful thinking on the subject of race, and it is well to make assurance doubly sure by testing the hypothesis that the superiority of the Nordic group is caused by the presence in the group of English speaking populations. It is possible to split the Nordic distribution in such a way that one group will contain representatives from countries which are predominantly English speaking (Eng¬ land, Scotland, Ireland and Canada), while the other group will contain representatives from countries which are pre¬ dominantly non-English speaking (Holland, Denmark, Germany, Sweden, Norway, Belgium, Austria, Russia, Italy and Poland). This we have done, and the results are given in Table 37, the two distributions being shown in Figure 44. The distributions of the English speaking Nordic group and the non-English speaking Nordic group show a differ¬ ence of 0.87^0.065, a difference which is 13.4 times the probable error of the difference. There are, of course, cogent historical and sociological reasons accounting for the in¬ feriority of the non-English speaking Nordic group. On the other hand, if one washes to deny, in the teeth of the facts, the superiority of the Nordic race on the ground that the language factor mysteriously aids this group when tested, Table No. 37 Analysis of the total Nordic sample into an English speak- ing Nordic group and a non-English speaking Nordic group. COMBINED ACTUAL DISTRIBUTION PROPORTION IN EACH SCALE THOUSAND INTERVALS ENGLISH NON-ENGLISH ENGLISH NON-ENGLISH SPEAKING SPEAKING SPEAKING SPEAKING NORDIC NORDIC NORDIC NORDIC 24.0-24.9 23.0-23.9 i • • • • .... 22.0-22.9 2 .... 2 .... 21.0-21.9 7 2 6 1 20.0-20.9 12 6 10 3 19.0-19.9 21 16 17 7 18.0-18.9 39 32 32 14 17.0-17.9 67 67 54 30 1G.0-16.9 108 131 87 59 15.0-15.9 143 214 116 96 14.0-14.9 176 293 143 132 13.0-13.9 201 365 163 164 12.0-12.9 172 356 139 160 11.0-11.9 109 262 88 118 10.0-10.9 70 189 57 85 9.0- 9.9 49 135 40 61 8.0- 8.9 31 82 25 37 7.0- 7.9 16 43 13 19 G.0- 6.9 7 19 6 9 5.0- 5.9 2 7 2 3 4.0- 4.9 1 2 .... 2 3.0- 3.9 .... 1 • • • • • • • • 2.0- 2.9 .... . . . , • . . . • • • • 1.0- 1.9 .... .... . . . . • • • • No. of cases. .. 1234 2222 Average . 13.84 12.97 S. D . 2.79 2.60 200 u M Q 03 C £ G £ ►H f*3 2 CD CD HH HH c £ K i <3 O £ I I O Q PS G £ O fc HH < « ca c/) HH J G £ PS I I ° X • r-H n T3 o ?H o 52 ; be o •i-H Od c 3 D Oh CD pG ’£ CD p—I «5 w o d pO 4-1 i+H o CD D Sh o a CD 0 ) Jh D rO o * o pG CD D o o d ?H D • rH T 3 D pO cd L_i d t-i OJ ?H D £ CD C o D 1 be £ c 3 ?H D > Cj d pO 441 d d O D £h D 5 « bed G r d d pO H CD % ° • C3 *r-( gn O O O * d g-srs O^S b£ ^ ^ r 1 be a « _ a TJ H fH H< c3 D Oh CD 4-4 o G> G C3 Oh D d 5+4 o CD o O • f-H -M o p-O CD (D g; o % C D ?H -pH be 03 PH D O b /0 g be C O be K C ^pO "-H PJ CD -"3 Kjr P^H .-4 —! HQ /-K cs ns D ?H o o r1 CD ^ W 4 ( O ^ c3 i-O ■M 4-4 0 v-i D o 0 .5 GO D SS'^P CO • Tjd g »o . -id fee C3 C D *M Oh r3 cd O u 2 ^ 1-0 >4 Oh CD p Ec pO CD o Po T 3 O c3 CD t4H TpH D O ^ o S ^H be 0 c3 pO ,nb < D c be. o 0 D PH CD c 3 CD o D D D tH P~ o —H CO CO be 174 AMERICAN INTELLIGENCE he may cut out of the Nordic distribution the English speaking Nordics, and still find a marked superiority of the non-English speaking Nordics over the Alpine and Medi¬ terranean groups. The difference between the non-English speaking Nordic group and the Alpine group is 1.30 =*= 0.047, a difference which is 27.6 times the probable error of the difference. The difference between the non-English speak¬ ing Nordic group and the Mediterranean group is 1.54 =*= 0.047, a difference which is 31.3 times the probable error of the difference. The distributions are shown graphically in Figure 45. Discarding the English speaking Nordics entirely, we still find tremendous differences between the non-English speaking Nordic group and the Alpine and Mediterranean groups, a fact which clearly indicates that the underlying cause of the nativity differences we have shown is race, and not language. It may be convenient for some to interpret the differ¬ ences found between the representatives of the three Euro¬ pean races in this country in terms of the standards having popular significance which were used in Section VI. The cri¬ teria of the per cent. A and B, and the per cent. D, D — and E give the following results: PER CENT. PER CENT. A AND B D,D— AND E English speaking Nordic. 12.3 19.9 Total Nordic. 8.1 25.8 Non-English speaking Nordic. 5.7 29.1 Alpine. 3.8 50.3 Mediterranean. 2.5 53.6 The criteria of the per cent, at or above the average white officer, and at or below the average of the negro draft give the following results: 200 bC P co 0 U cu P au uh au & M P 0 U A CO • I—I UP ^ .2 au cu rP 4-> d 0) au cu K CU bC| a W i p o p cu uP -u «-M o co au u o a CO CU cu P au be O u d P d S

•cH ?-( o «-W o co P O •l-H H-> P P •p-H -M Cfc • rH n3 (V G • pH Ph r—H < _ ^ *0 p ° • rH rP cu au d rP £ H-H O CU £ ° bo p P • pH O H-> au P d P o P au rP d au au cu X au Pi P au Ph co UP co Ph P o *H be d P P © Ph H g f a ^co o g 6 *> i> d P P Sh o -c-> CU P Uh P O a cu rP »o au §,£ • (VI be cu •p-H d f-H o H-> P cc 3 CU op • rH P be • rH co K*~J U au > au U< p H au rP CU _, bo^ P ^ P be p ccS u 2 au u p p co co CU cu p au Ch -< au be au o cu ^ % be ^ P au rP H~H p P4 & g £ Ph au P cu UP 4J Ph 4 J ■—J P <3 up au be © 03 ^ uh au au pp >■ au P rP co co • rH d up . co a- a be p O Uh Ph •N m Ph P - O U be UP CO • rH r —* be p W 4 h o au co P au 176 AMERICAN INTELLIGENCE PER CENT. PER CENT. AT OR ABOVE AT OR BELOW AVERAGE AVERAGE OF WHITE THE NEGRO OFFICER DRAFT English speaking Nordic. 4.0 10.9 Total Nordic. 2.3 14.5 Non-English speaking Nordic. 1.3 16.5 Alpine. 1.0 34.5 Mediterranean. 0.5 36.5 The criterion of the per cent, below an approximate “mental age” of eight gives the following results: PER CENT. BELOW “mental age” 8 English speaking Nordic. 0.8 Total Nordic. 1.1 Non-English speaking Nordic. 1.3 Alpine. 4.2 Mediterranean. 4.2 SECTION IX RE-EXAMINATION OF PREVIOUS CONCLUSIONS IN THE LIGHT OF THE RACE HYPOTHESIS It is now necessary to retrace our steps for a moment to examine some of our previous conclusions in the light of this new hypothesis. The hypothesis that the differences between the nativity groups found in the army tests are due to the race factor may be used to re-test our previous conclusions that each succeeding five year period of immi¬ gration since 1902 has given us an increasingly inferior selection of individuals (Section IV). The periods which we sample by means of the army data, and the average score on the combined scale of each sample are as follows: PERIOD NUMBER OF CASES COMBINED SCALE AVERAGE 1887-1897 764 13.82 1898-1902 771 13.55 1903-1907 1897 12.47 1908-1912 4287 11.74 1913-1917 3576 11.41 Table 35, which gives our estimates of the per cent, of Nordic, Alpine and Mediterranean blood coming to this country, shows that the big change in immigration came between the decades 1881-1890 and 1891-1900, the per¬ centage of Nordic blood which formerly ran from 40% to 50% having dropped to 30% in the decade 1891-1900, and to approximately 20% or 25% in the two subsequent dec¬ ades. On the other hand, the big drop in the intelligence of immigrants arriving came after 1902. The change in 177 178 AMERICAN INTELLIGENCE character of the immigration would account for part of the decline in the average intelligence of succeeding periods of immigration, but not for all of it. The decline in intelligence is due to two factors, the change in the races migrating to this country, and to the additional factor of the sending of lower and lower representatives of each race. The only tendency which would relieve this deplorable situation would be a current of emigration strong enough to counteract the current of immigration. Table 6 preced¬ ing shows the ratio between emigration and immigration for each of the nativity groups involved in this study, and we find in general between 1908 and 1917 a return current approximately one third of the arriving current. Unfortunately, no emigration statistics are available prior to 1908, and the figures after 1912 are distorted by the Balkan and European wars. The only sample that we can take that is comparatively free from outside influences is the sample 1908-1912. Taking the figures of arrivals and departures for this period, and dividing them into Nordic, Alpine and Mediterranean groups according to the method previously outlined, we obtain the following percentage estimates: Per cent, of Nordic ALIEN IMMIGRANTS ADMITTED ALIEN EMIGRANTS DEPARTED NET IMMIGRATION blood. Per cent, of Alpine 21.2 16.0 23.9 blood. Per cent, of Mediter- 50.4 50.6 50.2 ranean blood. Per cent, others and 23.2 28.6 20.5 unclassified. 5.2 4.8 5.4 AMERICAN INTELLIGENCE 179 The sample from this five year period shows a slight change (approximately 3%) in favor of the Nordic type and against the Mediterranean type, the Alpine immigra¬ tion holding its own. There is therefore no relief from our receding curve of intelligence from emigration, if this five year period be taken as typical of the outward alien pas¬ senger movement in other years. It will be remembered that the army authors tentatively offered the hypothesis that the more intelligent immigrants remained in this country, while the more stupid ones went home, as a possible method of accounting for the increase of intelligence scores with increasing years of residence. The gain of 3% in favor of the Nordic immigration would produce a very slight tendency in this direction, but not enough to account for the actual increase of intelligence scores found with increasing years of residence, 11.41 (1913-1917) to 13.82 (1887-1897). It will also be remembered that the army writers offered «/ the hypothesis of the better adaptation of the more thor¬ oughly Americanized group to the situation of the examin¬ ation to account for the increases shown. The factor of the adaptation to the situation of the examination cannot be dissected out of the total scores of the test. If such a factor were present, it would fall equally heavily on Nordic, Alpine and Mediterranean alike, unless the change in the character of immigration were so complete that the groups sampled at the two extremes of the residence groups (1887-1897 and 1913-1917) represented different race groups. But the difference between these two years of residence groups (2.41 ±0.0735) is so marked that it would be neces¬ sary to assume (if our Nordic group were the more thor¬ oughly Americanized) that the 1887-1897 group was com¬ posed entirely of English speaking Nordics or their equiva¬ lent in intelligence, and that our 1913-1917 group was 180 AMERICAN INTELLIGENCE composed entirely of Mediterraneans or their equivalent in intelligence, assumptions quite unwarranted in view of the fact that in the two years of residence groups 1887- 1897 and 1898-1902 we sampled 1545 individuals, while our Nordic group includes 3456 cases, and also in view of the fact that the Nordic immigration has dropped, in the period observed, at the outside from 45% to 20%. We may therefore conclude that the intangible factor of 4 ‘the more thoroughly Americanized group” can not be used to ex¬ plain the high test record of the Nordic group. There is only one other possible escape from the con¬ clusion that our test results indicate a genuine intellectual superiority of the Nordic group over the Alpine and Medi¬ terranean groups, and that is the assumption that the situation of the examination involved a situation that was “typically Nordic.” This assumption of course lands us in a perfect circle of reasoning. It would leave us with the conclusion that there was something mysteriously Nordic about alpha and beta that favored this race. We should have to assume that the Nordic, no matter where he may be, in the Canadian Northwest, in the Highlands of Scot¬ land, or on the shores of the Baltic, is always ready for an intelligence test. Perhaps it would be easier to say that the Nordic is intelligent. A situation “typically Nordic” could not be used, however, to account for the slight but real dif¬ ference between the English speaking Nordic and the non- English speaking Nordic groups. It is therefore best to abandon the attempt to account for the differences by the more or less feeble hypotheses that would make these dif¬ ferences an artifact of the method of examining, and recog¬ nize the fact that we are dealing with real differences in the intelligence of immigrants coming to our shores. We have previously noted the fact that the foreign born in the army sampled as representative of the immigrants AMERICAN INTELLIGENCE 181 coming to this country between 1887 and 1897 were statis¬ tically identical with the native born white draft. The change in the character of our immigration came between 1890 and 1900. The real drop in the curve of intelligence, however, started about 1900. We, therefore,cannot account for the drop in the intelligence of the immigrants sampled as representatives of those coming to this country in each five year period since 1902 by the race hypothesis entirely. SECTION X COMPARISON OF OUR RESULTS WITH THE CONCLUSIONS OF OTHER WRITERS ON THE SUBJECT In a very definite way, the results which we obtain by interpreting the army data by means of the race hypothesis support Mr. Madison Grant’s 1 thesis of the superiority of the Nordic type: “The Nordics are, all over the world, a race of soldiers, sailors, adventurers, and explorers, but above all, of rulers, organizers, and aristocrats in sharp contrast to the essentially peasant and democratic char¬ acter of the Alpines. The Nordic race is domineering, in¬ dividualistic, self-reliant, and jealous of their personal freedom both in political and religious systems, and as a result they are usually Protestants. Chivalry and knight¬ hood and their still surviving but greatly impaired counter¬ parts are peculiarly Nordic traits, and feudalism, class distinctions, and race pride among Europeans are traceable for the most part to the north.” (p.228.) “The pure Nordic peoples are characterized by a greater stability and steadi¬ ness than are mixed peoples such as the Irish, the ancient Gauls, and the Athenians, among all of whom the lack of these qualities was balanced by a correspondingly greater versatility.” (pp. 228-229.) Our results based on the army data also support Mr. Grant’s estimates of the Alpine race: “The Alpine race is always and everywhere a race of peasants, an agricultural and never a maritime race. In fact they only extend to salt water at the head of the Adriatic and, like all purely iMadison Grant. The Passing of the Great Race. New York, 1922, p. 476. 182 AMERICAN INTELLIGENCE 183 agricultural communities throughout Europe, tend toward democracy, although they are submissive to authority both political and religious, being usually Roman Catholics in western Europe. This race is essentially of the soil, and in towns the type is mediocre and bourgeois.” (p. 227.) Our results also support de Lapouge 1 in his contention that the Nordic type is superior to the Alpine. He says con¬ cerning the Alpine: “II est le parfait esclave, le serf ideal, le sujet modele, et dans les republiques comme la notre, le citoyen le mieux vu, ear il tolere tous les abus.” (p. 233.) “Les etats brachycephales, France, Autriche, Turquie, sans parler de la Pologne qui n’est plus, sont loin d’offrir la vitalite des Etats-Unis ou de l’Angleterre. Cependant la mediocrite nieme du brachycephale est une force. Ce neutre echappe a toutes les causes de destruction. Noiraud, courtaud, lourdaud, le brachycephale regne aujour d’hui del’Atlantique a la Mer Noire. Comme la mauvaise monnaie chasse Y autre, sa race a supplante la race meilleure. II est inerte, il est mediocre, mais se multiplie. Sa patience est au-dessus des epreuves; il est sujet sournis, soldat passif, fonctionnaire obeissant. Il ne porte pas ombrage, il ne se revolte point.” (p. 481.) It must, however, be frankly admitted that our results, which show the Mediterranean race inferior to the Alpine, are in contradiction with those of most writers who have inferred the intellectual level of a race from its historical achievements. Mr. Grant, for instance, says: “The mental characteristics of the Mediterranean race are well known, and this race, while inferior in bodily stamina to both the Nordic and the Alpine, is probably the superior of both, certainly of the Alpines, in intellectual attainments. In the field of art its superiority to both the other European races Georges Yacher de Lapouge. L’Aryen, son role social. Paris, 1899, p. 563. 184 AMERICAN INTELLIGENCE is unquestioned, although in literature and in scientific re¬ search and discovery the Nordics far excel it.” (P. 229). 1 The apparent contradiction between our results and the estimates of other observers has a very obvious solution, viz., that those who draw their conclusions from historical data are studying the Mediterranean race as it was at the period of its greatest development, when it produced the civilizations of Egypt, Mesopotamia, Phoenicia, and Crete, and, with a Nordic predominance, gave the civilizations of Greece and Rome, while our data sample this race group as it is at the present time. The sample we have taken as representative of the Medi¬ terranean race, as it is now constituted, is drawn from im¬ migrants in our army born, for the most part, in Greece, Ireland, Italy, and Turkey, and inasmuch as the number from Italy (4009) is so large, our Mediterranean sample is heavily weighted (approximately 2/3) by this nativity group. In regard to the Irish, Mr. Madison Grant says: “In spite of the fact that Paleoliths have not been found there, some indications of Paleolithic man appear in Ireland, both as single characters and as individuals. Being, like Brittany situated on the extreme western outposts of Eurasia, it has more than its share of generalized and low types surviving in the living populations, and these types, the Firbolgs, have imparted a distinct and very undesirable aspect to a large portion of the inhabitants of the west and south and have greatly lowered the intellectual status of the popula¬ tion as a whole. The cross between these elements and the x The quotations I have chosen from Mr. Madison Grant’s chapter on Racial Aptitudes most certainly do not do justice to that author, but they seemed to me to summarize his general position briefly. The entire book should be read to appreciate the soundness of Mr. Grant’s position and the compelling force of his arguments. AMERICAN INTELLIGENCE 185 Nordics appears to be a bad one, and the mental and cul¬ tural traits of the aborigines have proved to be exceedingly persistent and appear especially in the unstable tempera¬ ment and the lack of coordinating and reasoning power, so often found among the Irish. To the dominance of the Mediterraneans mixed with Pre-Neolithic survivals in the south and west are to be attributed the aloofness of the is¬ land from the general trend of European civilization and its long adherence to ancient forms of religion and even to Pre-Christian superstitions.” (pp. 202-203.) The immigrants in this country from Italy come mostly from Southern Italy and Sicily. The following quotation from Ripley 1 concerning Sicily is significant: “Commanding both straits at the waist of the Mediter¬ ranean, it has been, as Freeman in his masterly description puts it, ‘the meeting place of the nations.’ Tempting, there¬ fore, and accessible, this island has been incessantly over¬ run by invaders from all over Europe—Sicani, Siculi, Feni- cii, Greeks, and Romans, followed by Albanians, Vandals, Goths, Saracens, Normans, and last by the French and Spaniards. Is it any wonder that its people are less pure in physical type than the Sardinians or even the Calabrians of the mainland near by? Especially is this noticeable on its southern coasts, always more open to colonization than the northern edge. Nor is it surprising, as Freeman rightly adds, that ‘for the very reason that Sicily has found dwell¬ ing places for so many nations, a Sicilian nation there never has been.’” (p. 271.) The secret of the whole dilemma is the intermingling of races around the Mediterranean littoral in the last 2500 years. It is beside the point to contrast our results obtained by the actual psychological measurements of living repre- ^illiam Z. Ripley. The Races of Europe. New York, 1899, p. G24. 186 AMERICAN INTELLIGENCE sentatives of this race with the attainments of the tem¬ porary civilizations that flared up in historical times. The whole question of the degeneration of these peoples has been discussed by Mr. Charles W. Gould, 1 and our results from the examinations of drafted men born in these regions support his position. It is rather difficult to compare our results from the race groups with the various hypotheses erected by Pro¬ fessor William McDougall, 2 who, while he does not claim for the Nordic race “any general innate superiority” (p. 29), analyzes the mental constitution of this race and the other European races in such a way that an examination of his theories will be interesting. Professor McDougall’s hypotheses, very briefly and inadequately stated are: that the Nordic is stronger in the instinct of curiosity, the root of wonder, than the Mediterranean; the herd instinct, the root of sociability, is stronger in the Mediterranean than in the Nordic; the Nordic is constitutionally introvert, the Mediterranean constitutionally extrovert; the instinct of self-assertion is strong in the Nordic; the Alpine is introvert but not so strongly introvert as the Nordic; the Alpine has a high degree of sociability, is perhaps relatively weak in curiosity, and strong in the instinct of submission. In discussing innate differences in instinctive endowment, psychologists are still more or less in a speculative realm, but the field is open to experimental attack, and a body of knowledge based on experimentation is gradually growing. At the present time we must rely on concensus of opinion rather then experiment. A census of text-books on psychol- ogy would show “curiosity” usually listed as an instinctive Charles TV. Gould. America, a Family Matter. New York, 1922, p. 196. 2 William McDougall. Is America Safe for Democracy . New York, 1921, p. 213. AMERICAN INTELLIGENCE 187 tendency. If we follow Professor Thorndike 1 in his analysis, and eliminate many tendencies that others include, we shall still have left the instincts of multiform mental and phy¬ sical activity as the potent movers of men’s economic and recreative activities, (p. 144.) If any instinctive tendency finds expression in the tasks assigned by the army tests, it is this instinct for multiform mental activity, more vaguely termed “curiosity.” Our tests, however, measure the end result of such a tendency and not the tendency itself, and it is only in this vague way that our results showing the definite intellectual superiority of the Nordic race can be taken as substantiating or contributing to Professor Mc- Dougall’s hypothesis. It is difficult to check our results from the analysis of the foreign born white draft by country of birth (reported in Section VI) with the results of other investigators, on ac¬ count of the different tests that were used, and the differ¬ ent methods of selecting subjects. Miss Murdoch 2 exam¬ ined, by means of the Pressey group point scale, 500 Jew¬ ish children and 500 Italian children at one school in New York City, and 500 American children and 230 negro children at another school. The American and Jewish chil¬ dren tested about the same. About 15% of the Italians equalled or exceeded the median of the Jews, and about 30% of the negroes equalled or exceeded the median of the Jews. The investigation equalizes the environmental factor by selecting, in one instance, Italians and Jews from the same school and consequently from the same general neighborhood (East 110th St. near 2nd Ave.), and, in the other instance, by selecting native white and negro chil- : E. L. Thorndike, Educational Psychology, Vol. I. The Original Nature of Man , New York, 1919, p. 327. 2 K. Murdoch. A Study of Race Differences in New York City. School and Society, 1920, pp. 11, 147-150. 188 AMERICAN INTELLIGENCE dren from the same general neighborhood (West side, 8th Ave., near 140th St.), but the American children living in this neighborhood can not be taken as typical of the country as a whole. Miss Arlitt 1 concludes from her examination of 343 chil¬ dren, (191 native born Americans, 87 Italians, and 71 ne¬ groes), by the Stanford-Binet scale, that “there is a marked difference in the distribution of intelligence in groups of the same race but different social status,” and states that “race norms which do not take the social status factor into account are apt to be to that extent invalid.” (p. 183.) This position seems to ignore the observation, repeatedly confirmed by experiment, that children from the profes¬ sional, semi-professional and higher business classes have, on the whole, an hereditary endowment superior to that of children from the semi-skilled and unskilled laboring classes. Ter man 2 states “It has in fact been found wherever comparisons have been made that children of superior social status yield a higher average mental age than chil¬ dren of the laboring classes. . . . However, the common opinion that the child from a cultured home does better in tests solely by reason of his superior home advantages is an entirely gratuitous assumption. Practically all of the in¬ vestigations which have been made of the influence of nature and nurture on mental performance agree in attrib¬ uting far more to original endowment than to environ¬ ment. Common observation would itself suggest that the social class to which the family belongs depends less on chance than on the parents’ native qualities of intellect and character.” (p. 115.) l A. H. Arlitt. On the Need for Caution in Establishing Race Norms. Journal of Applied Psychology, 1921, pp. 5, 179-183. 2 L. M. Terman. The Measurement of Intelligence. Boston, 1916, p. 362. AMERICAN INTELLIGENCE 189 One frequently hears the opinion expressed in scientific circles that differences found between racial groups can not be attributed to race unless the individuals examined are drawn from the same social milieu. Miss Arlitt finds native born white children of inferior and very inferior social status above the Italian and negro children in in¬ telligence, but attributes the larger differences found be¬ tween the entire native white group and the Italian and negro groups to the fact that three eighths of the native white children come from homes of superior and very superior social status. In the same way, Miss Murdoch finds Jews living near East 110th St. and 2nd Ave. in New York City not very inferior to native born whites living in the mixed white and negro section around 8th Ave. and 140th St. The equalization of the environmental factor is a necessary control in certain phases of scientific experi¬ ments on race differences, but conclusions as to the intelli¬ gence of racial groups must be drawn from samples taken at random from the entire country. These conditions are more nearly met by the army sampling of individuals in the draft. Our samples of 81,465 native born individuals in the white draft, of 12,492 foreign born individuals, and 23,596 negroes are drawn impartially from every section of the country. If we selected our native born Americans from those who live in the same squalid conditions in which we find most of our negro and foreign born population, we would not have a fair sample. It is unfortunate that our army data classify foreign born individuals only by country of origin, so that we have no separate intelligence distributions for the Jews. Accord¬ ing to the 1910 census, about 50% of the foreign born popu¬ lation reporting Russia as their country of origin spoke Hebrew or Yiddish, about 25% spoke Polish, less than 3% spoke Russian, and the rest spoke Lithuanian, Lettish, 190 AMERICAN INTELLIGENCE German, Finnish, Ruthenian and other tongues. From the immigration statistics showing aliens admitted classified according to race or people, we find about 10% (arriving between 1900 and 1920) reported as Hebrew. It is fair to assume that our army sample of immigrants from Russia is at least one half Jewish, and that the sample we have selected as Alpine 1 is from one fifth to one fourth Jewish. Our figures, then, would rather tend to disprove the popular belief that the Jew 7 is highly intelligent. Immi¬ grants examined in the army, w r ho report their birthplace as Russia, had an average intelligence below those from all other countries except Poland and Italy. It is perhaps significant to note, how r ever, that the sample from Russia has a higher standard deviation (2.83) than that of any other immigrant group sampled, and that the Alpine group has a higher standard deviation than the Nordic or Mediterranean groups (2.60). If we assume that the Jewish immigrants have a low 7 average intelligence, but a higher variability than other nativity groups, this would reconcile our figures w 7 ith popular belief, and, at the same time, with the fact that investigators searching for talent in New York City and California schools find a frequent occurence of talent among Jewish children, The able Jew is popularly recognized not only because of his ability, but because he is able and a Jew 7 . Our results showing the marked intellectual inferiority of the negro are corrobated by practically all of the in¬ vestigators w 7 ho have used psychological tests on white and negro groups. This inferiority holds even when a low in¬ tellectual sampling of whites is made by selecting only ^here is no serious objection, from the anthropological standpoint, to classi¬ fying the northern Jew as an Alpine, for he has the head form, stature, and color of his Slavic neighbors. He is an Alpine Slav. AMERICAN INTELLIGENCE 191 those who live in the same environment, and who have had the same educational opportunities. Professor Ferguson, 1 who has studied the problem most carefully, concludes that in general 25% of the negroes exceed the median white. Our figures show a greater difference than he estimates, less than 12% of the negroes exceeding the average of the native born white draft. Professor Ferguson also estimates that 20% of pure negroes, 25% of negroes three quarters pure, 30% of the true mulattoes, and 35% of the quad¬ roons equal or exceed the average score of comparable whites. The discrepancies between data from various investiga¬ tors as to the amount of difference between negroes and whites probably result from different methods of selecting whites. If we compare negroes only to those whites who live in the same neighborhood, and who have had the same educational opportunities, our differences are smaller than those obtained by comparing samples of the entire white and negro populations. Some writers would account for the differences found between white and negro by differences of educational opportunity alone. The army tests showed the northern negro superior to the southern negro, and this superiority is attributed to the superior educational opportunities in the North. The educational record of the negro sample we are studying shows that more than half of the negroes from the southern States did not go beyond the third grade, and only 7% finished the eighth grade, while about half of the northern negroes finished the fifth grade, and a quarter finished the eighth grade. That the difference between the northern and southern negro is not entirely due to school- X G. O. Ferguson. The Mental Status of the American Negro. Scientific Monthly, 1921, pp. 12, 533-543. 192 AMERICAN INTELLIGENCE ing, but partly to intelligence, is shown by the fact that groups of southern and northern negroes of equal schooling show striking differences in intelligence. The superior intelligence measurements of the northern negro are due to three factors: first, the greater amount of educational opportunity, which does affect, to some ex¬ tent, scores on our present intelligence tests; second, the greater amount of admixture of white blood; and, third, the operation of economic and social forces, such as higher wages, better living conditions, identical school privileges, and a less^complete social ostracism, tending to draw the more intelligent negro to the North. It is impossible to dissect out of this complex of forces the relative weight of each factor. No psychologist would maintain that the men¬ tal tests he is now using do not measure educational oppor¬ tunity to some extent. On the other hand, it is absurd to attribute all differences found between northern and south¬ ern negroes to superior educational opportunities in the North, for differences are found between groups of the same schooling, and differences are shown by beta as well as by alpha. At the present stage of development of psychological tests, we can not measure the actual amount of difference in intelligence due to race or nativity. We can only prove that differences do exist, and we can interpret these differ¬ ences in terms that have great social and economic signifi¬ cance. The intellectual superiority of our Nordic group over the Alpine, Mediterranean, and negro groups has been demonstrated. If a person is unwilling to accept the race hypothesis as developed here, he may go back to the original nativity groups, and he can not deny the fact that differences exist. When our methods of measuring intellectual capacity have been perfected, we will be in a position to determine AMERICAN INTELLIGENCE 193 quantitatively the amount of race differences. Rough group tests of the type we are now using will indicate the fact that differences exist. However, while scientists are perfect¬ ing their methods of examining, it would be well for them to perfect their logic at the same time. Particularly mis¬ leading and unsound is the theory that disregards all dif¬ ferences found between racial groups unless the groups have had the same educational and environmental oppor¬ tunities. This theory in its most extreme form is set forth by Garth 1 as follows: “The elements in a study of racial mental similarities or differences must be these: (1) Two so-called races R x and R 2 , ( 2 ) an equal amount of educational opportunity, E, which should include social pressure and racial patterns of thought, and (3) psychological tests, D, within the grasp of both racial groups. We should have as a result of our experiment R x E D equal to, greater than, or less than R 2 E D. In this experiment the only unknown elements should be Rx and R 2 . If E could be made equal the experiment could be worked. “This element of educational opportunity-nurture, is the one causing most of the trouble in racial psychology as an uncontrollable element. It does not offer quite so much difficulty in the study of sex differences, yet it is there only in smaller degree than in racial differences, and as it is controlled the ‘sex differences’ tend to disappear. Since this element of education, or nurture, cannot be eliminated it would be safer to take for comparison such racial groups as have had as nearly the same educational opportunity as is possible having any disparity of this sort well in mind when we interpret the results of the experiment. Having ! T. R. Garth. White, Indian and Negro Work Curves. Journal of Applied Psy¬ chology, 1921, 5, 14-25. 194 AMERICAN INTELLIGENCE done this, we first take the complete distributions on the scale of measurement for the groups as statements of the true facts of the case, race for race. We then combine these distributions into a total distribution of accomplishment of all the races taken together to see if we have multimodal effects. Should we find these effects we may conclude that we have evidence of types, or racial types, and there should in this case be one mode for each racial group. But should the combined distribution for the several racial groups reveal only one mode we may conclude that the test reveals no types—no real racial differences but rather similarities.” (p- 16.) If intelligence counts for anything in the competition among human beings, it is natural to expect that individ¬ uals of superior intelligence will adjust themselves more easily to their physical and social environment, and that they will endow their children not only with material goods, but with the ability to adjust themselves to the same or a more complex environment. To select individuals who have fallen behind in the struggle to adjust themselves to the civilization their race has built as typical of that race is an error, for their position itself shows that they are, for the most part, individuals with an inferior heredi¬ tary endowment. t/ In the same wav, our educational institutions are them- selves a part of our own race heritage. The average negro child can not advance through an educational curriculum adapted to the Anglo-Saxon child in step with that child. To select children of equal education, age for age, in the two groups, is to sample either superior negroes or inferior whites. The scientific problem is that of eliminating from the tests used as measuring instruments those particular tests which demonstrably measure nurture, and to measure, AMERICAN INTELLIGENCE 195 with genuine tests of native intelligence, random or im¬ partial samples from each race throughout the entire range of its geographical and institutional distribution. CONCLUSIONS Our study of the army tests of foreign born individuals has pointed at every step to the conclusion that the aver¬ age intelligence of our immigrants is declining. This deteri¬ oration in the intellectual level of immigrants has been found to be due to two causes. The migrations of the Alpine and Mediterranean races have increased to such an extent in the last thirty or forty years that this blood now consti- tutes 70% or 75% of the total immigration. The represen¬ tatives of the Alpine and Mediterranean races in our immigration are intellectually inferior to the representa¬ tives of the Nordic race which formerly made up about 50% of our immigration. In addition, we find that we are getting progressively lower and lower types from each nativity group or race. In the light of our findings in Sections IV and IX, it is pos¬ sible to re-draw our curve (Figure 33) representing in¬ crease of intelligence score with increasing years of residence and to represent it truly as in Figure 46, which shows the decline of intelligence with each succeeding period of im¬ migration. It is also possible to make a picture of the elements now entering into American intelligence. At one extreme we have the distribution of the Nordic race group. At the other extreme we have the American negro. Between the Nordic and the negro, but closer to the negro than to the Nordic, we find the Alpine and Mediterranean types. These distributions we have projected together in Figure 47. Throughout this study all measurements have been made in terms of averages and variability about the average. In interpreting averages, we must never forget that they stand 197 16-20 J1-15 6-10 0-5 1898 1903 1908 1913 TO TO TO TO 1902 1907 1912 1917 YEARS RESIDENCE IN UNITED STATES 198 1887 TO 1897 Figure 46. The decline of intelligence with each succeeding period of immigration. The apparent increase of intelligence with in¬ creasing length of residence, as shown in Figure 33, has been proved to be a progressive decrease in the intellectual level of immigrants coming to this country in each succeeding five-year period since 1902. The evidence indicates that the immigrants prior to 1902 were intellectually equal to the native born white draft. The army sample of “native born” includes, besides na¬ tive born of native parentage, the native born of foreign or mixed parentage. It is perhaps possible that the native born of native parentage might have tested higher than 13.77. The position of the white draft born in England is shown above. Although the true position of the native born American may be a matter of speculation, there is no doubt that the more recent immigrants are intellectually closer to the negro than to the native born white sample. 199 TOTAL NORDIC —.COMBINED ALPINE AND MEDITERRANEAN-NEGRO DRAFT a i-s ^ = = ^ 3 05 4H © C S Sh 05 3 © r* a u Sh © o’ >5 s- c " cl « 0) CO G ^ S3 c 3 © © u X 3 G2 • p-i j_> Sh — -j-> no .2 « © s pG G3 C Eh © o ^ 3 -h © 32 .G a 2 3 © =£ .2 2 Sh *1 © ^ •*j > © 3 ''d © bfi-G 3 05 2 x D O b£ « 'a c3 *-» 3 2 gs « .2 +s «« g . •i—i ^ © S3 O Sh fel 3 ■"H < 3 3 © G u 5 feJD H O SH fH o ^ .■a © T3 qj 3 _Sh © > 3 © © £ ,© 3 3 © "£ © <~> 05 E" M 3 © c a Sh C fel Sh © C l 3 -G 2 c H c £ 05 Sm © 3 Sh Sh 3 © *43 £ G © © H-H © •> 3 05 Sh ■+H O S ^ s © © h3 © ^ «4H © c 05 33 s+H c© 33 3 3 © G £ *G © f—( < 33 © © 33 ^ ~ 3 a: fH _ 5" © Sh O Sh (U bC © u 3 ■+ © ^ 3 't © 05 3 O © © 05 33 © 3 33 © © © X g © bl Gh g 3 G G ^ © O .. 23 H sh &i “2 3 '■G © Sh Sh < 3; © O ^ 33 -M • H 33 © © y, © ^ © C Xh HH > 3 3 3 © © bl G © 05 © © 05 43 © . 5 cl, *> bl © ^ 33 3 © +H 3 © a bl,S v o G G ^ 3 G *3 3? = g »< a 'C