Abstract
Collecting job analysis ratings for a large number of jobs via surveys, interviews, or focus groups can put a very large burden on organizations. In this study, we describe and evaluate a streamlined, natural language processing-based approach to estimating (a) the importance of various knowledges, skills, abilities, and other characteristics (KSAOs) to jobs, and (b) how descriptive various interests are of work on a job. Specifically, we evaluate whether we can train a machine to accurately estimate KSAO ratings for jobs using job description and task statement text as the sole input. Data for 963 occupations from the U.S. Department of Labor’s Occupational Information Network (O*NET) system and an independent set of 229 occupations from a large organization provided the basis for the evaluation. Our approach produced KSAO predictions that had cross-validated correlations with subject matter expert (SME) ratings of knowledges, skills, abilities, and interests of .74, .80, .75, and .84, respectively (on average, across the 126 KSAOs examined). We found clear evidence for the validity of machine-based predictions based on (a) convergence among machine-based and SME-furnished ratings, (b) conceptually meaningful patterns of prediction model regression coefficients among the KSAOs examined, and (c) conceptual relevance of top predictor models underlying related clusters of KSAOs. We also found that prediction models developed on O*NET data produced meaningful results when applied to an independent set of job descriptions and tasks. Implications of this work, as well as suggested directions for future job analysis research and practice, are discussed.
Similar content being viewed by others
Notes
Although not of historical focus in the I-O psychology job analysis literature, interests play a critical role in occupational analysis in organizational practice and applied vocational, organizational, and educational psychology (Su, 2020). Vocational interests have long been a central component of vocational counseling and career exploration (Brown & Lent, 2013). Additionally, recent research has clearly linked vocational interests on theoretical and empirical grounds to job performance (e.g., Nye, Su, Rounds, & Drasgow, 2017; Schmidt, 2014; Su, 2020; Van Iddekinge, et al., 2011). As such, we deemed interests as important “other characteristics” to examine, given our focus on enterprise-wide job analysis efforts where differentiating among a diverse set of jobs is a central concern.
The Federal occupation to O*NET crosswalk used in this study was developed as part of earlier research to inform a Federal career exploration system, which in turn was based on crosswalk developed by OPM between Federal occupation series codes and the Standard Occupational Classification (SOC) (Russell, 2011). The aforementioned crosswalk was created through a multi-step human subject matter expert (SME) judgment process detailed in Russell (2011). The crosswalk focused on linking Federal and O*NET occupations that were as much a direct match as possible in terms of job duties performed. This does not mean that O*NET occupations not crosswalked to a given Federal occupation were not related to that occupation (e.g., some uncrosswalked occupations may have similar KSAO requirements), simply that they were deemed an insufficient match in term of job duties performed relative to the O*NET occupations ultimately crosswalked to them.
In the O*NET 22.0 database, 1,100 occupations had descriptions, 966 had KSA ratings, 974 had interest ratings, and 963 had core tasks. The subset of occupations examined here provides nearly full coverage of the 974 occupations that constitute occupations covered by O*NET’s data collection plan (https://www.onetcenter.org/taxonomy/2010/data_coll.html).
These sources can be found online at the following URLs: PCS: https://www.opm.gov/policy-data-oversight/classification-qualifications/classifying-general-schedule-positions/#url=Standards; JGS: https://www.opm.gov/policy-data-oversight/classification-qualifications/classifying-federal-wage-system-positions/#url=Standards), and HOGF: https://www.opm.gov/policy-data-oversight/classification-qualifications/classifying-general-schedule-positions/occupationalhandbook.pdf.
Given the large drop in words, we also examined models using a 1% threshold for eliminating words and found no substantive difference in model cross-validation results (e.g., cross-validated r’s did not differ by more than .01). This was expected given words with such a low base rate would not be expected to offer much predictive value upon cross-validation. We also attempted to examine models with even lower thresholds for eliminating words (0.75%, and 0.5%) but were not able to fit models because as the sample was split for modeling fitting purposes, several words wound up having no variance across occupations.
Interested readers can find more details on how components factor into the estimation of SPLS regression coefficients in the SPLS algorithm section of Chun and Keles (2010).
We created an Open Science Framework respository for this article at https://osf.io/jbtkp/. This repository includes R code for (a) initial processing of the O*NET data, (b) cleaning of O*NET occupation text in preparation for modeling, (c) training of SPLS models and application to O*NET occupation test sets, and (d) consolidation of results across KSAOs. In addition to code, we provide key data files we started with and that are produced by running the code referenced above.
Follow-up analyses revealed setting these bounds had no substantive impact on conclusions drawn in this study. We imposed these bounds to allow for more direct comparison to results in O*NET and to permit future researchers to derive predictions that were bounded by values used the O*NET ratings scales for KSAOs.
To facilitate quick, visual identification of the top language-based predictors for each KSAO, we have also provided supplemental materials that show word clouds of the top 50 positive predictors in the final SPLS KSAO models (see Supplement C).
We thank an anonymous reviewer for highlighting this positive finding.
When interpreting these results, note that residuals will be larger for interests, as they are based on a 7-point rating scale, whereas the KSAs are based on a 5-point rating scale.
To be clear, the judgment of “conceptual meaningfulness” here is somewhat subjective. It is based on a review of these results by a subset of the authors of this manuscript, all of whom have extensive research and applied experience in the areas of individual differences and occupational analysis. Thus, the perspective offered here is based on decades of applied research experience.
We thank an anonymous reviewer for suggesting we conduct further analyses to explain the relatively high profile correlations among uncrosswalked occupations.
In the context of text analysis, a “dictionary” is simply a set of words that have been identified as useful for measuring or predicting some construct or concept of interest.
References
Amorim, E., Cancado, M., & Veloso, A. (2018). Automated essay scoring in the presence of biased ratings. Proceedings of NAACL-HLT 2018, 229–337. Association for Computational Linguistics.
Baranowski, L. E., & Anderson, L. E. (2005). Examining rating source variation in work behavior to KSA linkages. Personnel Psychology, 58(4), 1041–1054.
Blei, D. M., Ng, A. Y., & Jordan, M. I. (2003). Latent Dirichlet allocation. Journal of Machine Learning Research, 3, 993–1022.
Bobko, P., Roth, P. L., & Buster, M. L. (2008). A systematic approach for assessing the currency (“up-to-dateness”) of job-analytic information. Public Personnel Management, 37, 261–277.
Brannick, M. T., Levine, E. L., & Morgeson, F. P. (2007). Job and work analysis: Methods research, and applications for human resource management (2nd ed.). Sage.
Brannick, M. T., Pearlman, K., & Sanchez, J. I. (2017). Work analysis. In J. L. Farr & N. T. Tippins (Eds.), Handbook of Employee Selection (2nd ed., pp. 134–161). Routledge.
Brown, S. D., & Lent, R. W. (2013). Career development and counseling: Putting theory and research to work, 2nd edition. Wiley.
Campion, M. C., Campion, M. A., Campion, E. D., & Reider, M. H. (2016). Initial investigation into computer scoring of candidate essays for personnel selection. Journal of Applied Psychology, 101, 958–975.
Carter, G. W., Cook, K. W., & Dorsey, D. W. (2009). Career paths: Charting courses to success. Wiley-Blackwell.
Carter, G. W., Dorsey, D. W., & Niehaus, J. W. (2004, April). The use of transactional data in occupational analysis: Text-mining of on-line job listings. In J. M. Ford (Chair), Automated text analysis in I/O psychology: Research to practice. Symposium conducted at the Annual Conference of the Society for Industrial and Organizational Psychology, Chicago.
Casner-Lotto, J., & Barrington, L. (2006). Are they really ready to work? Employers’ perspectives on the basic knowledge and applied skills of new entrants to the 21st century workforce. Partnership for 21st Century Skills.
Cawley, G. C., & Talbot, N. L. C. (2010). On over-fitting in model selection and subsequent selection bias in performance evaluation. Journal of Machine Learning Research, 11, 2079–2107.
Chun, H., & Keles, S. (2010). Sparse partial least squares regression for simultaneous dimension reduction and variable selection. Journal of the Royal Statistical Society: Series B, 72, 3–25.
Chung, D, Chun, H., & Keles, S. (2013). spls: Sparse partial least squares (SPLS) regression and classification. (Version 2.2–1). Available from: https://CRAN.R-project.org/package=spls
Costanza, D. P., & Fleishman, E. A. (1992). Fleishman Job Analysis Survey (Part III). Management Research Institute.
Dawis, R. V., & Lofquist, L. H. (1984). A psychological theory of work adjustment. University of Minnesota Press.
Department of the Army (2019). Army Research Institute for the Behavioral and Social Sciences – Request for White Papers: Data Science for Enhancing Job Design (Solicitation Number: W911NF-18-S-0005-FY1906)
Dierdorff, E. C., & Norton, J. J. (2011). Summary of procedures for O*NET task updating and new task generation. National Center for O*NET Development. Retrieved July 8, 2017, from https://www.onetcenter.org/dl_files/TaskUpdating.pdf
Efron, B., & Hastie, T. (2016). Computer age statistical inference: Algorithms, evidence, and data science. Cambridge University Press.
Feinerer, I., & Hornik, K. (2017). tm: Text mining package. (Version 0.7–1). Available from: https://CRAN.R-project.org/package=tm
Fleisher, M. S., & Tsacoumis, S. (2012a). O*NET analyst occupational skills ratings: Procedures update. Human Resources Research Organization. Retrieved July 8, 2017, from https://www.onetcenter.org/dl_files/AOSkills_ProcUpdate.pdf
Fleisher, M. S., & Tsacoumis, S. (2012b). O*NET analyst occupational ability ratings: Procedures update. Human Resources Research Organization. Retrieved July 8, 2017, from https://www.onetcenter.org/dl_files/AnalystProcUpdate.pdf
Fleishman, E. A., Constanza, D. P., Marshall-Mies, J., Wetrogan, L. I., & Uhlman, C. E. (1995a). Knowledges. In N. G. Peterson, M. D. Mumford, W. C. Borman, P. R. Jenneret, & E. A. Fleishman (Eds.), Development of prototype Occupational Information Network (O*NET) content model (pp. 4–1 - 4–23). Utah Department of Employment Security.
Fleishman, E. A., & Mumford, M. D. (1991). Evaluating classifications of job behavior: A construct validation of the ability requirement scales. Personnel Psychology, 44, 523–575.
Fleishman, E. A., & Reilly, M. E. (1992). Handbook of human abilities: Definitions, measurements, and job task requirements. Consulting Psychologists Press.
Fleishman, E. A., Wetrogan, L. I., Uhlman, C. E., & Marshall-Mies, J. (1995b). Abilities. In N. G. Peterson, M. D. Mumford, W. C. Borman, P. R. Jenneret, & E. A. Fleishman (Eds.), Development of prototype Occupational Information Network (O*NET) content model (pp. 10–1–10–39). Utah Department of Employment Security
Frey, C. B., & Osborne, M. A. (2017). The future of employment: How susceptible are jobs to computerization? Technological Forecasting and Social Change, 114, 254–280.
Friedman, L., & Harvey, R. J. (1986). Can raters with reduced job descriptive information provide accurate Position Analysis Questionnaire (PAQ) ratings? Personnel Psychology, 39, 779–789.
Gael, S. (1988). Job descriptions. In. S. Gael (Ed.), The job analysis handbook for business, industry, and government, Volume 1 (pp. 71–89). Wiley.
Gruder, E. J. (2012). Identifying appropriate sources of work information. In M. A. Wilson, W. Bennett, S. G. Gibson, & G. M. Alliger (Eds.), The handbook of work analysis: Methods, systems, applications and science of work measurement in organizations (pp. 31–40). Routledge/Taylor & Francis Group.
Harvey, R. J., & Lozada-Larsen, S. R. (1988). Influence of amount of job descriptive information on job analysis rating accuracy. Journal of Applied Psychology, 73, 457–461.
Hastie, T., Tibshirani, R., & Friedman, J. (2009). The elements of statistical learning: Data mining, inference, and prediction (2nd ed.). Springer.
Holland, J. L. (1997). Making vocational choices: A theory of vocational personalities and work environments (3rd ed.). Psychological Assessment Resources.
Howard, A. (Ed.) (1995). The changing nature of work. Jossey-Bass.
Kern, M. L., Park, G., Eischstaedt, J. C., Schwartz, H. A., Sap, M., Smith, L. K., & Ungar, L. H. (2016). Gaining insights from social media language: Methodologies and challenges. Psychological Methods, 21, 507–525.
Kjell, O. N. E., Sikström, S., Kjell, K., & Schwartz, H. A. (2022). Natural language analyzed with AI-based transformers predict traditional subjective well-being measures approaching the theoretical upper limits in accuracy. Scientific Reports, 12, 3918.
Kosinski, M., Stillwell, D., & Graepel, T. (2013). Private traits and attributes are predictable from digital records of human behavior. PNAS Proceedings of the National Academy of Sciences of the United States of America, 110, 5802–5805.
Kosinski, M., Wang, Y., Lakkaraju, H., & Leskovec, J. (2016). Mining big data to extract patterns and predict real-life outcomes. Psychological Methods, 21, 493–506.
Kuhn, M., & Johnson, K. (2013). Applied predictive modeling. Springer.
Lance, C. E., Butts, M. M., & Michels, L. C. (2006). The sources of four commonly reported cutoff criteria: What did they really say? Organizational Research Methods, 9, 202–220.
Landers, R. N. (2017, April). A crash course in natural language processing. The Industrial-Organizational Psychologist, http://www.siop.org/tip/april17/crash.aspx
Landers, R. N., Brusso, R. C., Cavanaugh, K. J., & Collmus, A. B. (2016). A primer on theory-driven web scraping: Automatic extraction of big data from the Internet for use in psychological research. Psychological Methods, 21, 475–492.
Lievens, F., Sanchez, J. I., & De Corte, W. (2004). Easing the inferential leap in competency modeling: The effects of task-related information and subject matter expertise. Personnel Psychology, 57, 881–904.
Liu, B. (2015). Sentiment analysis: Mining opinions, sentiments, and emotions. Cambridge University Press.
Manning, C. D., & Shutze, H. (1999). Foundations of statistical natural language processing. MIT Press.
Manning, C. D., Surdeanu, M., Bauer, J., Finkel, J., Bethard, S. J., & McClonsky, D. (2014). The Stanford CoreNLP natural language processing toolkit. Proceeding of the 52nd Annual Meeting of the Association of Computational Linguistic: System Demonstrations (pp. 55–60). Baltimore, MD.
McGraw, K. O., & Wong, S. P. (1996). Forming inferences about some intraclass correlation coefficients. Psychological Methods, 1, 30–46.
McKenny, A. F., Aguinis, H., Short, J. C., & Anglin, A. H. (2018). What doesn’t get measured does exist: Improving the accuracy of computer-aided text analysis. Journal of Management, 44, 2909–2933.
McKenny, A. F., Short, J. C., & Payne, G. T. (2013). Using computer-aided text analysis to elevate constructs: An illustration using psychological capital. Organizational Research Methods, 16, 152–184.
Michalke, M. (2017a). Using the koRpus package for text analysis. Retrieved July 29, 2017a, from https://cran.r-project.org/web/packages/koRpus/vignettes/koRpus_vignette.pdf
Michalke, M. (2017b). koRpus: An R package for text analysis. (Version 0.10–2). Available from: https://reaktanz.de/?c=hacking&s=koRpus
Mikolov, T., Sutskever, I, Chen, K., Corrado, G. S., & Dean, J. (2013). Distributed representations of words and phrases and their compositionality. Advances in Neural Information Processing Systems. Retrieved July 8, 2017, from https://arxiv.org/abs/1310.4546
Morgeson, F. P., & Campion, M. A. (1997). Social and cognitive sources of potential inaccuracy in job analysis. Journal of Applied Psychology, 82, 627–655.
Morgeson, F. P., & Campion, M. A. (2000). Accuracy in job analysis: Toward an inference-based model. Journal of Organizational Behavior, 21, 819–827.
Mumford, M. D., & Peterson, N. G. (1995). Skills. In N. G. Peterson, M. D. Mumford, W. C. Borman, P. R. Jenneret, & E. A. Fleishman (Eds.), Development of prototype Occupational Information Network (O*NET) content model (pp. 3–1 - 3–75). Utah Department of Employment Security.
National Center for O*NET Development. (n.d.) Content model. O*NET Resource Center. Retrieved July 8, 2017, from https://www.onetcenter.org/content.html
Nye, C. D., Su, R., Rounds, J., & Drasgow, F. (2017). The relationship between interests and performance: An updated meta-analysis. Journal of Vocational Behavior, 98, 138–151.
O’Connor, B. P. (2000). SPSS and SAS programs for determining the number of components using parallel analysis and Velicer’s MAP test. Behavior Research Methods, Instrumentation, and Computers, 32, 396–402.
O’Neil, C. (2016). Weapons of math destruction. Crown Publishing Group.
Pan, Y., Peng, Y., Hu, T., & Jiebo, L. (2017). Understanding what affects career progression using LinkedIn and Twitter Data. Special Session on Intelligent Data Mining. IEEE Big Data Conference, Boston, MA.
Park, G., Schwartz, H. A., Eichstaedt, J. C., Kern, M. L., Kosinski, M., Stillwell, D. J., Ungar, L. H., & Seligman, M. E. (2015). Automatic personality assessment through social media language. Journal of Personality and Social Psychology, 108, 934–952.
Pennebaker, J. W., Boyd, R. L., Jordan, K., & Blackburn, K. (2015). The development and psychometric properties of LIWC2015. University of Texas at Austin.
Praama, B. Y., & Samo, R. (2015). Personality classification based on Twitter text using Naïve Bayes, KNN, and SVM. 2015 International Conference on Data and Software Engineering (ICoDSE). Yoayakara, Indonesia.
Preotiuc-Pietro, D., Carpenter, J., Giorgi, S., & Ungar, L. (2016). Studying the Dark Triad of personality through Twitter behavior. In Proceedings of the 25th ACM International on Conference on Information and Knowledge Management (CIKM '16). New York.
Primoff, E. S. (1975). How to prepare and conduct job element examinations (U.S. Civil Service Commission Technical Study 75–1). Government Printing Office.
Putka, D. J., Beatty, A., & Reeder, M. (2018). Modern prediction methods: New perspectives on a common problem. Organizational Research Methods, 21, 689–732.
Raymark, P. H., Schmit, M. J., & Guion, R. M. (1997). Identifying potentially useful personality constructs for employee selection. Personnel Psychology, 50, 723–736.
Reeder, M. C., & Tsacoumis, S. (2017a). O*NET analyst ratings of occupational skills: Analysis cycle 17 results (2017a No. 0003). Human Resources Research Organization. Retrieved July 8, 2017, from https://www.onetcenter.org/dl_files/AOSkills_17.pdf
Reeder, M. C., & Tsacoumis, S. (2017b). O*NET analyst ratings of occupational abilities: Analysis cycle 17 results (2017b No. 0002). Human Resources Research Organization. Retrieved July 8, 2017, from https://www.onetcenter.org/dl_files/Wave17.pdf
Riloff, E. (1995). Little words can make a big difference for text classification. Proceedings of the 18th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval (pp. 130–136). Seattle, WA.
Rinker, T. W. (2013). qdap: Quantitative discourse analysis package. (Version 2.2.5). University at Buffalo. Available from: http://github.com/trinker/qdap
Robinson-Morral, E. J., Hendrickson, C., Gilbert, S., Myers, T., Simpson, K., & Loignon, A. C. (2018). Practical considerations for conducting job analysis linkage exercises. Journal of Personnel Psychology, 17, 12–21.
Rounds, J., Armstrong, P. I., Liao, H., Lewis, P., & Rivkin, D. (2008). Second generation occupational interest profiles for the O*NET system: Summary. National Center for O*NET Development. Retrieved July 8, 2017, from https://www.onetcenter.org/dl_files/SecondOIP_Summary.pdf
Rounds, J., Smith, T., Hubert, L., Lewis, P., & Rivkin, D. (1999). Development of Occupational Interest Profiles (OIPs) for O*NET. National Center for O*NET Development. Retrieved July 8, 2017, from https://www.onetcenter.org/dl_files/OIP.pdf
Russell, T. L. (2011). Linking Federal occupational series to the O*NET-SOC 2010 classification. In D. J. Putka & R. A. McCloy (Eds.), Building an interactive federal career exploration and discovery system: Phase I (pp. F11-58). Human Resources Research Organization.
Sanchez, J. I., & Levine, E. L. (2000). Accuracy or consequential validity: Which is the better standard for job analysis data? Journal of Organizational Behavior, 21, 809–818.
Schmid, H. (1994). Probabilistic part-of-speech tagging using decision trees. Proceedings of International Conference on New Methods in Language Processing, Manchester, UK.
Schmidt, F. L. (2014). A general theoretical integrative model of individual differences in interests, abilities, personality traits and academic and occupational achievement: A commentary on four recent articles. Perspectives on Psychological Science, 9, 211–218.
Short, J. C., Broberg, J. C., Cogliser, C. C., & Brigham, K. H. (2010). Construct validation using computer-aided text analysis (CATA) an illustration using entrepreneurial orientation. Organizational Research Methods, 13, 320–347.
Su, R. (2020). The three faces of interests: An integrative review of interest research in vocational, organizational, and educational psychology. Journal of Vocational Behavior, 116 Part B, 103240. https://doi.org/10.1016/j.jvb.2018.10.016
The Psychometrics Centre. (n.d.a). Apply Magic Sauce. Retrieved July 8, 2017, from https://applymagicsauce.com/
The Psychometrics Centre. (n.d.b). Predicted traits. Retrieved July 8, 2017, from https://applymagicsauce.com/documentation_traits.html
Tonidandel, S., King, E. B., & Cortina, J. M. (2018). Big data methods: Leveraging modern data analytic techniques to build organizational science. Organizational Research Methods, 21, 525–547.
Tsacoumis, S., & Van Iddekinge, C. H. (2006). A comparison of incumbent and analyst ratings of O*NET skills. Human Resources Research Organization. Retrieved July 8, 2017, from https://www.onetcenter.org/dl_files/SkillsComp.pdf
U.S. Department of Labor, Employment and Training Administration. (2018). O*NET® Data Collection Program, Office of Management and Budget Clearance Package Supporting Statement. Part A: Justification. Author. Retrieved April 15, 2018, from https://www.onetcenter.org/dl_files/omb2018/Supporting_StatementA.pdf
U.S. Office of Personnel Management (2018). FedScope Employment Cube June 2018. Washington, D.C.: Author. Online: https://www.fedscope.opm.gov/employment.asp
Van Iddekinge, C. H., Roth, P. L., Putka, D. J., & Lanivich, S. E. (2011). Are you interested? A meta-analysis of relations between vocational interests and employee performance and turnover. Journal of Applied Psychology, 96, 1167–1194.
Wegman, L. A., Hoffman, B. J., Carter, N. T., Twenge, J. M., & Guenole, N. (2018). Plotting job characteristics in context: Cross-temporal meta-analysis of changes in job characteristics since 1975. Journal of Management, 44, 352–386.
Wilson, M. A., Bennett, Jr., W., Gibson, S. G., & Alliger, G. M. (Eds.) (2012). The handbook of work analysis: Methods, systems, applications and science of work measurement in organizations. Routledge/Taylor & Francis Group.
Youyou, W., Kosinski, M., & Stillwell, D. (2015). Computer-based personality judgments are more accurate than those made by humans. Proceedings of the National Academy of Sciences of the United States of America (PNAS), 112, 1036–1040.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary Information
Below is the link to the electronic supplementary material. R code and key data files used in this article have been made publicly available via the Open Science Framework and can be accessed at https://osf.io/jbtkp/.
Rights and permissions
About this article
Cite this article
Putka, D.J., Oswald, F.L., Landers, R.N. et al. Evaluating a Natural Language Processing Approach to Estimating KSA and Interest Job Analysis Ratings. J Bus Psychol 38, 385–410 (2023). https://doi.org/10.1007/s10869-022-09824-0
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10869-022-09824-0