College and Research Libraries The Changing Nature of Jobs: A Paraprofessional Time Series Carol P. Johnson This study attempts to determine if paraprofessional jobs have changed significantly as a result of automation during the period between 1975 and 1990. Three raters reviewed job descriptions dated 1975, 1981, and 1990 from the technical services department of a small academic library using the Position Analysis Questionnaire, a widely used struc- tured job analysis tool. Analysis of the resulting data-standard z scores on thirteen overall job dimensions using ANOVA and omega-squared estimates-indicates no statistically significant differences among the jobs. Although these results show that the change in jobs over time is more idiosyncratic than expected, they may also illustrate the de-skilling effect of computers on library functions. D n 1984,Alan Veanernoted that "once a technology is applied to carry out very complex, rou- tine mental work, that work is driven downward in the work hierar- chy .... The change has provided mag- nificent professional enrichment oppor- tunities for librarians and has similarly enriched the jobs of support staff." 1 This insight raises an interesting question. With the downward shift in tasks from librarians to paraprofessionals, is there evidence of a corresponding upgrade in the parapro-fessional's responsibilities? Have paraprofessional jobs changed sig- nificantly as a result of automation? Much has been written in recent years about the working conditions and respon- sibilities of paraprofessionals by authors such as Charlotte Mugnier, Larry R. Oberg, and Cathleen C. Palmini. 24 Other researchers, among them Darla Rushing and Ann Prentice, concentrated on the impact of automation on the library or- ganization and specific departments. 5· 6 Scholars such as Harry Braverman and Shoshana Zuboff have written about the complicated effects of automation on the workplace from deskilling to the difficul- ties of managing computer-mediated work. 7· 8 Specific research by Suzanne Iacono and Rob Kling indicates that tech- nology neither automatically degrades nor improves a job.9 The differences are because of the ways in which work is or- ganized rather than because of the tech- nology. A search of library literature does not indicate any studies examining li- brary paraprofessional jobs over time to determine what changes, if any, have taken place as a result of technology. This study uses the Position Analysis Questionnaire (PAQ), a widely used structured job analysis tool, to examine, Carol P. Johnson is Director of Libraries and Audiovisual Services at the College of St. Catherine Library, St . Paul, Minnesota. 59 60 College & Research Libraries on a micro level, the job descriptions of three paraprofessional jobs in an aca- demic library technical services depart- ment at a small, private liberal arts col- lege. Dated 1975, 1981, and 1990, the job descriptions are examined to determine what, if any, changes occurred as a result of automation and to see if there are sta- tistically significant differences among the jobs over time. To some extent, these three jobs do reflect both a job progres- sion over time and changing responsibili- ties because of automation. Structured job analysis tools differ from the job evaluation systems used to determine organizational compensation. Research into statistical methods to analyze job data for the purposes of determining job similarities and differences began in the late 1970s and 1980s for Equal Employment Opportunity Commission purposes. According to Roberts' Dictionary of Indus- trial Relations, "job analysis is the system- atic investigation or study of a job or po- sition to determine the responsibilities, duties and qualifications required to per- form the job." Job evaluation is the "pro- cess of determining the classification, rat- ing or value of an individual job in rela- tion to the other jobs in an organization." 10 Job analysis, often an informal process, provides the data for job evaluation, and typically, job evaluation systems reflect the values and politics of the local orga- nization. This study attempts to study the jobs in question using a methodology that is not affected by local values and poli- tics. There are a number of more formal job analysis methods available for organiza- tions to use. Among these are Functional Job Analysis, the Job-Element approach, the PAQ, Critical Incidents Methodology, and Task Inventories. However, this study required a job analysis tool that produces quantitative data for comparison and lim- January 1996 its potential rater bias. The tool chosen was the PAQ, which rates jobs quantita- tively on 187 elements of work activity and the work environment. The instru- ment is based on the concept that human work can be analyzed "in terms of mean- ingful 'units' or job elements of a worker- oriented nature." 11 The designers as- sumed that there was a commonality across jobs resulting from the workers doing similar things and not as a result of the technology used or the product pro- duced.12 The PAQ's present form evolved over several decades from its earliest form, The Checklist of Work Activities, developed in 1958 by Ernest J. McCormick and G. J. Palmer. In its present form, job raters re- spond to each of the 187 questions using Likert-format scales. These questions are organized into six divisions: (1) informa- tion input; (2) mental processes; (3) work output; (4) relationships with other per- sons; (5) job context; and (6) other job characteristics. Results are provided for each of these categories, plus thirteen overall dimensions. Overall dimensions include: (1) decision, communication, and general responsibilities; (2) machine/ equipment operation; (3) clerical activi- ties; (4) technical activities; (5) service ac- tivities; (6) work schedule; (7) routine or repetitive activities; (8) work environ- ment; (9) physical activities; (10) super- vision; (11) public contacts; (12) hazard- ous environment; and (13) flexible sched- ule/ optional apparel. Each division in- cludes elements that tend to occur in com- bination on the job. The rater-scored sheets are analyzed by computer, and standard z scores are provided for each job I rater for each of the six divisions plus the overall dimensions. 13 Research into statistical methods to analyze job data for the purposes of de- termining job similarities and differences began in the late 1970s and 1980s for Equal Employment Opportunity Com- mission purposes. These articles focused on the type of job analysis tool used to describe the job (task oriented, worker oriented, or abilities oriented) and the sta- tistical techniques employed to analyze the data. 14 In a 1977 article, Richard D. Arvey and Kevin M. Mossholder proposed the use of analysis of variance (ANOVA) com- bined with the omega-squared estimate to compare jobs and determine similari- ties for validity generalization purposes. 15 In a later study, Arvey and fellow re- searchers found that given a reasonable sample size, the ANOVA model was an effective design in detecting true differ- ences among jobs, but the omega-squared estimates were more useful in determin- ing job differences even across small sample sizes. They determined that a rea- sonable rule of thumb would be that "val- ues near .30 indicate major job differences among jobs and values less than .15 indi- cate considerable job similarities."16 In an unpublished 1984 study, Arvey and fellow researchers hypothesized that one way to measure changes in jobs over time was to measure job description in- formation over time using the thirteen overall dimension z scores generated by the PAQY This article reports the find- ings of a statistical analysis of the PAQ results from these three positions using the ANOVA and omega-squared esti- mates. Methodology The authors derived the information for the analysis from job descriptions dated 1975, 1981, and 1990. The three jobs re- flect three paraprofessional library jobs in the cataloging department of a small, aca- demic library. ]obi In 1975, one position that could be de- scribed as paraprofessional existed in the cataloging department of the library. Master's-levellibrarians and clerks filled other positions. The position assisted the cataloging librarian in maintaining the public catalog and technical services-spe- The Changing Nature of Jobs 61 cific files, supervising and training stu- dent employees in filing cards in the main catalog, and searching for available copy cataloging using print tools. A high school education was required, and su- pervisory and typing skills were useful. The training curve was estimated at six months. ]ob2 By 1981, the number of paraprofessional positions increased and two levels ex- isted. Computerized cataloging via OCLC began in 1976. The paraprofessional in this job (level 2) located, reviewed, and modified cataloging copy in the OCLC database using the computer terminal. A second responsibility involved mainte- nance of the shelf list. Two years of col- lege-level study, one year of experience as a library technician, computer train- ing, and typing skills were necessary. ]ob3 By 1990, computers had replaced catalog cards, and database record maintenance and editing had replaced card filing. The main responsibility of the position in- volved database coordination, supervi- sion of catalog entries, and media cata- loging. The job now required two years of experience as a library technical assis- tant, level 1. Based on the changes in selection re- quirements and the tasks performed, the author expected that the ANOVA model would find significant statistical differ- ences among the jobs and reasonable omega-squared values for the job x di- mension effects. Three raters reviewed each job description (nine in all) to pre- vent biased results. They used the Job Analysis Manual for the Position Analysis Questionnaire. The raters, who were not experienced job analysts, consisted of one degreed librarian familiar with the job at the time in question; one degreed librar- ian who occupied a similar position in an- other library of the same size; and a third 62 College & Research Libraries TABLE I Job Evaluation Points for Each Job and Rater Job 1 (1975) Job 2 (1981) Job 3 (1990) Rater 1 Rater 2 Rater 3 503 688 673 667 657 613 771 640 685 library staff person familiar with the job (librarian or incumbent). The results were submitted to PAQ Services, Inc., in Utah for analysis. The PAQ results were then run on SPSS using the ANOVA model. The ANOVA model is used to detect significant differ- ences among jobs. For the AN OVA model, the dimension scores are considered a within-job factor, the jobs, and a between- job factor with raters nested between jobs. A significant F-value for the between-job factor indicates that there are statistical differences between the jobs. January 1996 A significant F-value for the job x di- mension interaction would indicate that the profiles of the jobs are significantly different; that is, the job dimensions dif- fer across the jobs. The omega-squared es- timate is calculated when the F-values in- dicate significant differences among jobs to determine the proportion of variance of the job dimensions (job similarities) and the proportion of variance because of job x dimension (job differences). Results PAQ-derived job evaluation points (mean) (see table 1) for the three jobs were: job 1 (1975), 621; job 2 (1981), 645; and job 3 (1990), 698 (see the time series chart in figure 1 and table 2 for the job profiles and z scores on the thirteen over- all dimensions) . For each job and rater, the PAQ provided z scores for the six di- visions-information output, mental pro- cesses, work output, relationships with other persons, job context, and other job FIGURE 1 Time Series Job Comparison on PAQ Library Technician Series 15..--------------------------------, 10 .5 0.0 -05 -10 -15 c Vl iO u QJ QJ 0 w u ·c: u 3 ·u; c iii .c > u ·u £. u m QJ QJ u G ~ .c (/) 0 IU u ~ (/) DIM QJ > u .~ c ~ s w 0 Vl >- .c a: .c a_ a_ > .~ m :0 a. :::J :::J a_ (/) u ·a. >-c 0 z Job One Job Two Job Three The Changing Nature of Jobs 63 TABLE2 Z Scores on Overall Dimensions for Jobs 1-3, Mean Scores Dimensions 1. Having decision, communication, and general responsibilities 2. Operating machines and/ or equipment 3. Performing clerical and/ or related activities 4. Performing technical and/ or related activities 5. Performing service and/ or related activities 6. Other work schedules vs. working regular day schedules 7. Performing routing and/ or repetitive activities 8. Being aware of work environment 9. Engaging in physical activities 10. Supervising/ directing/ estimating 11. Public and/ or customer and/ or related contacts 12. Working in an unpleasant/hazardous/ demanding environment 13. Having a nontypical schedule/ optional apparel style Job Dim1 Dim2 Dim3 Dim4 DimS Dim6 Dim7 Dim8 Dim9 Dim10 Dimll Dim12 Dim13 1 -.35 -.77 -.03 -.14 .16 .22 2 -.08 -.30 .50 -.47 -.29 .29 3 .07 -.34 .13 -.17 -.40 .39 characteristics-as well as the thirteen overall dimensions. Researchers tested the z scores for the thirteen overall dimen- sions (see table 3) to determine job differ- ences. Computed interclass correlation co- efficients for each job ranged from .69 to .84, sufficient for preceding with the ANOVA. These coefficients were mini- mally above the average reliability coeffi- cient (.68) reported for a sample of over 1,000 jobs involving over 3,000 pairs of analysts in a study of the PAQ interanalyst reliability.18 .48 .18 .16 -.81 .66 .93 -.78 -.79 .32 -.90 .10 .92 -.45 -.49 .60 -1.01 -.02 .65 -.70 -.47 .15 The ANOVA on the thirteen overall di- mension z scores (table 4) indicated no sig- nificant statistical differences among jobs, nor a significant job x dimension interac- tion. The job x dimension accounted for .008 percent of the variance; the source of variance as a result of jobs accounted for .004 percent. The value resulting from the ratio of variance indicates there is almost eighteen times the variance because of job similarities than job differences. Extremely low omega-squared scores ( <.15) indicated very strong job similarities, not differences. TABLE3 Z Scores on Overall Dimensions for Jobs 1-3, Raters 1-9 Job Rater Diml Dim2 Dim3 Dim4 DimS Dim6 Dim7 Dim8 Dim9 DimlO Dimll Dim12 Dim13 I 1 -.83 -.84 -.56 .04 .38 .21 .65 -.60 -.16 1.09 -.58 -.55 .22 1 2 -.08 -.51 .33 .12 -.03 .21 .50 -.65 1.03 .94 -.70 -1 .35 .59 1 3 -.15 -.95 .16 -.56 .13 .25 .29 -1.17 1.11 .75 -1.09 -.49 .16 2 4 .20 -.13 1.21 -.30 -.45 .22 .19 -.90 .46 1.35 .17 -.19 1.06 2 5 -.14 -.31 .30 -1.04 -.12 .25 -.02 -.91 .25 .65 -1.02 -.73 .39 2 6 -.28 -.45 .02 -.08 -.31 .40 .37 -.88 -.41 .76 -.50 -.57 .38 3 7 .32 -.42 -.21 -.18 -.67 .26 -. 12 -.89 -.17 .72 -.71 -.79 .08 3 8 -.31 -.31 .30 -.03 .08 .62 .55 -1.02 -.43 .58 -.37 .08 -.12 3 9 .20 -.29 .31 -.31 -.61 .30 .06 -1.12 .57 .67 -1.02 -.69 .50 64 College & Research Libraries Hypothesizing that the thirteen overall dimension scores were too abstract or gen- eral, the author ran ANOVAS on the subscores of four selected divisional sets of z scores: Division 1 • Information input • Interpreting what is sensed • Using sources of information • Watching devices and/ or materials for information • Being aware of environmental con- ditions • Using various devices Division 2 • Mental processes • Making decisions • Processing information Division 3 • Work output • Using machines and/ or tools and/ or equipment • Performing activities requiring general body movements • Controlling machines or processes • Performing skilled and/ or techni- cal activities • Performing controlled manual and/ or related activities • Using miscellaneous equipment and/ or devices • Performing handling and/ or re- lated activities • General physical coordination Division 4 • Relationships with other persons Communicating judgments and/or re- lated information • Engaging in general personal con- tact • Performing supervisory and/ or co- ordinating and/ or related activities • Exchanging job-related informa- tion • Public and/ or related personal con- tacts January 1996 The ANOVA results show no signifi- cant differences for any of the four divi- sions. Discussion Test results indicate that there are no sig- nificant statistical differences among the 1975, 1981, and 1990 jobs using the ANOVAmodel to analyze the PAQ's thir- teen overall dimensions . These results do not support a conclusion that technology has significantly upgraded the parapro- fessional positions under review. Indeed, the results indicate change, but of an un- expected nature: job redesign to meet the needs of the time (old skills discarded, new ones learned), resulting in the re- placement of some duties by others. In- creased efficiency of work flow and op- erations, in this particular situation, re- sulted in a narrowing of functions, not an upgrading of responsibilities. Although not statistically significant, the total mean job evaluation points have increased over time (from 621 to 698) as a result of the increased educational re- Although not statistically signifi- cant, the total mean job evaluation points have increased over time (from 621 to 698) as a result of the increased educational requirements. quirements. Mort McPhail of Jeanneret and Associates produced a list of twenty- six PAQ elements likely to be affected by automation. He noted that technology can increase or decrease the worth of the job. Increases "in such ratings as training and experience required may be offset by de- creases in items dealing with decision making, combining and analyzing infor- mation, etc., and the jobs may have be- come simpler and less demanding to per- form."1 9 Also noticeable is a shift in respon- sibilities discernible by viewing the in- dividual PAQ "Item With Highest Per- centile Scores" sections of the PAQ re- ports (copies of these are available from the author). Supervision, which had the highest percentile score in the The Changing Nature of Jobs 65 TABLE4 ANOVA Results on the Thirteen Overall Dimensions 1975 job and is usually Source of Variation valued highly in job ---------------------------------------- df MS F o:l evaluation systems, de- Between Jobs 2 .08 -.004 dined. The highest per- Jobs (A) centile PAQ item for the Raters within jobs C(A) 6 .28 .28 .01 1990 job deals with rna- Within Jobs chines and equipment, not Dimensions (B) 12 2.21 20.78 .215 people. The next highest is inspecting, as opposed to students/trainees for the Job x dimensions Dimension x raters within jobs (B x C(A)) 24 72 . 15 1.42 .008 .11 .11 1975 job. The 1990 job is in- ....__ ____________________________________ ___. deed different, reflecting an apparent re- arrangement of work activities brought on by automation. The history of library automation can explain much about the changes in these three jobs. Automation for the smaller li- brary began with cataloging functions. In the late 1970s, the majority of libraries be- came members of cooperative national cataloging bibliographic utilities, such as OCLC. Shared online cataloging pro- duced the catalog cards and machine- readable archival tapes, and changed the job structure in some cataloging depart- ments. The shift to online cataloging meant that more copy cataloging was done, par- ticularly as the number of contributing libraries increased. Copy cataloging (ed- iting a specific record to meet local library requirements) did not require the skills of a degreed librarian. Rather, it required an individual able to understand catalog- ing terminology, use computers, and learn the coding requirements of the bib- Paraprofessionals performed the same duties as some librarians: copy catalog- ing and supervision. Automation at this time contributed to the confusion of roles between professional and paraprofes- sional. A review of division 4 (relationships with other persons) z scores supports the above contention that job responsibilities have changed. It is evident that the 1990 position requires more judgment commu- nication, less general personal contact, less supervisory responsibility, and less public contact than the two previous jobs. The 1976 automation of cataloging func- tions increased the job level (1981) by in- cluding copy cataloging; but by 1990, a narrowing of job function had occurred as librarians learned to adapt office auto- mation strategies of work flow and effi- ciency to cataloging functions. The 1990 job provides more responsibility for da- tabase management but decreased super- visory responsibility. liographic utility. Those libraries that Conclusion chose to employ nondegreed library staff This study began as an effort to under- to perform copy cataloging incorporated stand the impact of computers on the these requirements by increasing the edu- paraprofessional jobs in the technical ser- cational and training requirements for the vice department of one library, anticipat- job. All other job requirements remained ing a statistically significant difference. the same. The author believed that the addition of This is also reflected in the duties and responsibilities of the 1981 position. "Re- lations with Professional Personnel" was the highest-ranking PAQ item for this job. OCLC copy cataloging to the 1981 posi- tion, and definitely the addition of data- base record maintenance and media cata- loging to the 1990 position, would make 66 College & Research Libraries a statistically significant difference. This did not occur. Perhaps this is the result of the unique job history or pattern for this particular library, or it may result from the PAQ's worker orientation (as opposed to task orientation), which may not discriminate sufficiently to measure the "real" job difference. Duplication of the study in other libraries should be un- dertaken to see if similar results are found. In this particular case, technology, of- ten viewed as a way to upgrade the sta- tus of library positions, does not seem to have raised the job level. One reviewer of this article noted that it seems "counter intuitive" that increasing levels of re- quired education do not make a differ- Technology, a better educated work force, fewer accredited library schools, and the continuing democ- ratization of the information process will all continue to exacerbate the problem in the future. ence. Although qualifications (KSAs) are used as data in the PAQ, in this case the unique blend of responsibilities in each job resulted in some dimensions, highly valued in job evaluation systems, decreas- ing. Others, also highly valued, stayed the same or increased. The combination of increases, balanced against decreases was not enough to produce a statistically sig- nificant difference. All libraries attempt to recruit staff with high qualifications; however, this is a subjective judgment area . With the changing work environment, the ques- tion of what qualifications are necessary to perform the work at each level of the library organization is a legitimate sub- ject for research. This study attempted to find a tool that would examine jobs over time with as little bias as possible, but further studies are needed to answer that question. The paraprofessional jobs un- der review have not been significantly up- graded because the responsibilities that January 1996 upgrade positions and typically lead to better pay (typically, autonomy; author- ity, including supervisory responsibility; and decision-making) are not reflected in the jobs under review. The national adoption of MARC as the preferred way for most libraries to cata- log library materials has resulted in the standardization of cataloging rules and practices. MARC copy cataloging com- puterizes and standardizes the level of decision making. This permits non-MLS staff to take over the bulk of copy cata- loging. As Shoshana Zuboff noted, "the purpose of the intelligent technology at the core of a computer system is to substitute algorithms or decision rules for individual judgments. This substitu- tion makes it possible to formalize the skills and know-how intrinsic to a job and integrate them into a computer pro- gram." 20 This study indicates that automation has not upgraded technical services li- brary paraprofessional positions as much as the author had thought. The functions remain the same, but the tools change (and the level of the person doing the work). However, this study did not in- vestigate the intrinsic rewards that tech- nology may add to jobs by including re- sponsibilities that are more interesting and central to the operation of libraries. There are new jobs in libraries that did not exist prior to the introduction of com- puters; it would be interesting to identify these and compare them with the more traditional jobs of libraries. The future li- brary will have both. Real improvement in the status of paraprofessionals is de- pendent upon the level of authority, au- tonomy, and skill assigned to the jobs they fill. This will not occur without further study of the necessary qualifications for all levels of library work and the appro- priate ways to obtain those qualifications. Technology is a means, not an end; and it is not the answer to the increasing un- certainty that besets the profession. The deskilling effects of computers certainly complicate an already complex situation. Technology, a better educated work force, fewer accredited library schools, and the continuing democratization of the infor- mation process will all continue to exac- erbate the problem in the future. Re- The Changing Nature of Jobs 67 searchers need more data to clarify both paraprofessional and professional posi- tions and their respective roles in librar- ies. Job analysis using tools such as the PAQ may provide useful data to aid in the discussion. Notes 1. Alan B. Veaner, "Librarians: The Next Generation," Library Journal109 (Apr. 1, 1984): 623- 25. 2. Charlotte Mugnier, The Paraprofessional and the Professional Job Structure (Chicago: ALA, 1980). 3. Larry R. Oberg, Mark E. Mentges, P.N. McDermott, and Vitoon Harusadangkul, "The Role, Status, and Working Conditions of Paraprofessionals: A National Survey of Academic Librar- ies," College & Research Libraries 53 (May 1992): 215-38. 4. Cathleen C. Palmini, "The Impact of Computerization on Library Support Staff: A Study of Support Staff in Academic Libraries in Wisconsin," College & Research Libraries 55 (Mar. 1994): 119-27. 5. Darla H. Rushing, "Caught in the Middle: Systems, Staff and Maintenance in the Medium- Sized Academic Library," Journal of Library Administration 13 (1990): 157-73. 6. Ann E. Prentice, "Jobs and Changes in the Technological Age," Journal of Library Adminis- tration 13 (1990): 47-57. 7. Harry Braverman, Labor and Monopoly Capital; The Degradation of Work in the Twentieth Cen- tury (New York: Monthly Review Pr., 1974). 8. Shoshana Zuboff, In the Age of the Smart Machine: The Future of Work and Power (New York: Basic Bks., 1988). 9. Suzanne Iacono and Rob Kling, "Changing Office Technologies and Transformations of Clerical Jobs," in Technology and the Transformation of White-collar Work, ed. Robert E. Kraut (Hillsdale, N.J.: L. Erlbaurn Associates, 1987). 10. Roberts' Dictionary of Indu strial Relations, 4th ed. (Washington, D.C.: Bureau of National Affairs, 1994). 11. Ernest J. McCormick, Paul R. Jeanneret, and Robert C. Mecham, "A Study of Job Charac- teristics and Job Dimensions as Based on the Position Analysis Questionnaire (PAQ)," Journal of Applied Psychology Monograph 56 (Aug. 1972): 247-368 . 12. Benjamin Schneider and Neal Schmitt, Staffing Organizations, 2nd ed. (Glenview, Ill.: Scott, Foresman and Co., 1986). 13. Ernest J. McCormick, Robert C. Mecham, and Paul R. Jeanneret, Technical Manual for the Position Analysis Questionnaire, 2nd ed. (PAQ Services Inc., 1989). 14. Edwin T. Cornelius III, Theodore J. Carron, and Marianne N. Collins, "Job Analysis Mod- els and Job Classification," Personnel Psychology 32 (1979): 693-708. 15. Richard D. Arvey and Kevin M. Mossholder, "A Proposed Methodology for Determining Similarities and Differences among Jobs," Personnel Psychology 30 (1977): 363-74. 16. Richard D. Arvey and others, "Detecting Job Differences: A Monte Carlo Study," Person- nel Psychology 34 (1981) : 709-30. 17. Richard D. Arvey and others, Skills Obsolescence: Psychological and Economic Perspectives (unpublished paper, 1984). 18. McCormick, Mecham, and Jeanneret, Technical Manual for the Position Analysis Question- naire. 19. Mort McPhail, PAQ Items Likely to Be Impacted by Office Automation (Logan, Utah: Jeanneret Associates, 1990). 20. Shoshana Zuboff, "New Worlds of Computer-Mediated Work," Harvard Business Review 60 (Sept./Oct. 1982): 142-52.