The History Corner: Digital Humanities and the Psychology of Work
Nathan T. Carter, Megan R. Lowery, and Lane E. Siedor
It is with great excitement that I write to you in my first installment of the History Corner. I am proud to join a great group of people who have served as Historian in the past. Over the coming years, we plan to continue build upon the fantastic work by my predecessor, Jeff Cucina, who has initiated such projects as the SIOP Time Capsule and the Living History Series, in which the field’s luminaries are interviewed at the annual conference. I am also excited that my first History Corner article features two stellar graduate students from the University of Georgia, whose independent study in the area of “big data” inspired this article.
Digital Humanities and the Psychology of Work
Nathan T. Carter, Megan R. Lowery, and Lane E. Siedor
University of Georgia
As scientists studying the psychological experience of workers, I-O psychologists are frequently limited by the scope of the data available to them. We study the changing nature of work over time and its impact on workers (e.g., Wegman, Hoffman, Carter, Twenge, & Guenole, 2016) but are limited to the relatively immediate past—primarily the 1970s to the present day. This seems a major limitation considering humans have participated in organized work across the globe for millennia and major changes to how work is done, how it is viewed, and its place in our lives has gone through many shifts. Furthermore, we are often limited to studying more statistically common subpopulations of workers in contemporary settings. Despite this, there is much to be said for studying luminaries in various fields (e.g., Aguinis & O’Boyle, 2014) and in examining the context-dependent boundary conditions of our theories by examining less common occupational groups such as authors, artists, and great leaders.
The limitations noted above are rapidly being lifted in a movement often referred to as the digital humanities. This movement lies at the intersection of fields such as history, anthropology, the arts, and data science, allowing for a more integrated and systematic perspective on various human phenomena. In this column, we argue that the digital humanities perspective presents an incredible opportunity for I-O psychologists interested in asking unconventional questions or conventional questions in unconventional populations and settings.
In particular, we focus here on the promise of text-based analysis, a familiar area because of the recent surge of interest in “big data” (see Guzzo, Fink, King, Tonidandel, & Landis, 2015 and accompanying commentaries; Harlow & Oswald, 2016, and associated articles; Landers, 2016; Morrison & Abraham, 2015; and Tonidandel, King, & Cortina, 2016) but focuses on data sources traditionally associated with the humanities and historical research. Researchers may aim to analyze transcriptions of interviews in oral history collections, major works of literature (both quantitatively and qualitatively), memoirs, or text-based updates on social media to explore motives for behavior, elements of well-being, relationships between performance and a variety of individual differences, social movements, labor disputes, and histories of unrest in work matters.
Sources of Data
One of the most advantageous aspects of utilizing the digital humanities approach in psychological research lies in the emphasis on historical data that have been collected, digitized, and made freely available by numerous online sources. Table 1 shows a list of various free online resources available along with hyperlinks to their websites. In addition, more specialized archives exist, such as the Southern Oral History Project (http://sohp.org/) and the European University Institute’s Archives (http://archives.eui.eu/en/oral_history/). Aside from the readily available sources online, we highly recommend researchers seek out support for use of the digital humanities within the university setting. For example, the University of Georgia Libraries have launched a Digital Humanities Initiative (DIGI) to encourage related research, and universities such as the University of Illinois and Duke University have similar resources. These centers often offer digitization services for researchers and can be invaluable for finding others with similar interests. Below, we discuss two areas that we believe are rife for attention from I-O psychologists.
Perceptions of workers through the years. Many databases go back very far, and databases also exist that are primarily focused on major events in history that were centered around workers, such as the Farmworker Movement Documentation Project at the University of California San Diego (https://libraries.ucsd.edu/farmworkermovement/medias/oral-history/). Such databases, as well as historical and anthropological work regarding working populations (e.g., Timothy Minchin’s Fighting Against the Odds: A History of Southern Labor Since WWII) represent a highly valuable potential source for data regarding the conditions and sentiments of workers. In fact, many anthropologists study human labor ranging from the present to the ancient. Certainly, there is a great potential for inroads to be made in fields of study such as the Anthropology of Work (see http://saw.americananthro.org/) and Industrial Archaeology (see http://www.sia-web.org/) that should not be ignored by I-O psychologists. An awareness of and involvement in these fields may greatly increase the impact and generalizability of work similar to Wegman et al. (2016) who studied changes in job characteristics in the recent past (1970s to present). Where direct accounts of workers are not available, more indirect sources could be examined, such as popular literature, music, and art, whose popularity may be an indirect indicator of public sentiment. Perhaps it is “pie in the sky” but we envision a psychology of work that can reach back further in time to mine data for a peek into the psychological experience of workers that have long since passed.
Rare or typically unavailable worker populations. For many good reasons, the majority of I-O psychology research is based on what we would consider representative worker populations, which allows for generalizations about a large number of workers. But the importance of occupational subpopulations is certainly not lost on I-O psychologists, and the increasing focus on situational influences is a good indication of the need for its consideration. However, many subpopulations of workers are difficult to reach for commonly employed field and survey methods, particularly the “stars,” or those individuals with disproportionately high performance, visibility, or social capital (Call, Nyberg, & Thatcher, 2015) that may be particularly rare within a given population. Of course, psychology is no stranger to qualitative studies of major figures, such as United States presidents, and clinical psychologists have generally lead this charge in the past (e.g., Gartner, 2008; McAdams, 2011). A digital humanities perspective, however, can bring new rigor and broadened scope to such study and allow these major figures to be differentiated along a number of lines, such as the setting of their work, their demographics and history (think biodata), and the source of the data by the analysis of biographies, autobiographies, memoirs, diaries, and interviews. The foci of these studies could include presidents, queens, kings, others occupying major political leadership positions, CEOs, musicians, artists, novelists, and so on. In addition, the experiences of marginalized workers such as immigrants, sex workers, and drug dealers could be explored through oral histories and other published accounts of their experiences.
Descriptive analysis. Researchers may want to start by conducting basic text analyses or summaries of their collected works. Descriptive analyses may include determining frequencies of words or phrases (i.e., n-grams), collocation (i.e., words that frequently occur in conjunction with one another), concordance (i.e., the context of a given set of words), or entity recognition (Duke University Libraries, 2016). Although these analyses are easily conducted, they provide only descriptive information whereas other types of analyses might allow richer inferences to be drawn. In using basic text analysis, I-O psychologists may answer questions such as: “Do men and women differ in the way they talk about their jobs?” “Has the way individuals describe their jobs or work changed over time?” or “Are there certain aspects of work that are more frequently discussed than others?” Furthermore, there are a host of tools available for descriptive text analyses including Linguistic Inquiry and Word Count (LIWC; http://liwc.wpengine.com/) engine, or the Google n-gram engine (https://books.google.com/ngrams), to name a couple.
Sentiment analysis. Next, researchers may wish to determine the sentiment of written work by computationally identifying and categorizing opinions expressed or attitudes toward a particular object, idea, movement, organization, and so on. Sentiment analysis largely determines whether typical attitudes toward specific targets has been expressed as positive, negative, or neutral within different literature pieces, and whether these opinions are comparative or not in nature. Sentiment analysis may provide very valuable, qualitative information that can be connected across sources or years to get a better understanding of how attitudes about work have changed over time or how attitudes differ across the world. For example, work by the United Nations Pulse Lab, highlighted by Gloss et al. (2016), used the tone of conversations on social media to further understand global unemployment trends. However, many preexisting tools for conducting sentiment analyses such as Pattern, a module for Python, and the “Rsentiment” package in R (Bose & Goswami, 2016) may not suit the needs of a specific research project. To explain, the tools and dictionaries used in sentiment analyses have to match the underlying theoretical argument set forth within the project. For example, if researchers wanted to investigate the shift in valence of attitudes toward wellness programs over time, they would have to utilize a program/dictionary that has been taught to identify words or phrases frequently associated with wellness programs. Thus, if there is not a preexisting tool to suit a researcher’s specific needs, there is an element of machine learning and data science that will be required to construct an appropriate analysis tool. To this end, Hernandez, Newman, and Jeon (2016) provide an excellent chapter that describes such a process and is highly recommended reading.
Relationship or network analysis. Network analysis has already begun to elucidate concepts across our field, such as in theories of leadership (e.g., Carter, DeChurch, Braun, & Contractor, 2014) and in understanding how attitudes form and evolve over time (Dalege et al., 2016). This more advanced type of analysis investigates patterns of relationships among different actors within groups. Network analysis has primarily been used to study patterns of communication networks, influence networks, or friendship networks, and we believe it will also be useful at the intersection of digital humanities and the psychology of work. For example, researchers may answer questions such as: “How can an organization drive innovation through creativity networks?” “What characteristics of individuals are needed within a network at key times in organizational change?” “How can network analysis be used to investigate mergers and acquisitions?” or “How can organizations identify integral people for hiring and promotions based on their network connectivity?” Furthermore, in several fields there has been an identification of the utility of examining collaboration networks (via coauthorships), and other types of networks could certainly be deduced from historical accounts and anthropological documentation. Additionally, researchers may be interested in constructing synonymy networks (Guame, Duvgnau, Prevote, & Desalle, 2008) in order to identify how similar words are based on clustering, and may help explain differences and similarities in how people have talked about work over time, providing a statistical and graphical methodology for finding commonality in unique word usage (similar to the identification of a latent variable among item responses). Network analysis is likely to be incredibly informative and, with the rise in prevalence of use within our field, some researchers are already familiar with tools or programs used for data analysis.
Inference of psychological constructs. Finally, researchers may wish to use techniques to identify and scale the psychological characteristics of persons being studied. That is, we can study individual differences (e.g., personality) expressed within the digital humanities and identify individuals in terms of the language they use using tools such as the IBM Watson personality text analysis program, which returns personality trait scores based on a given set of text. For example, work by Dr. James W. Pennebaker of the University of Texas at Austin has focused on the intersection of linguistics and psychology in identifying markers of certain aspects of personality or behavior. Interestingly, a study by Newman, Pennebaker, Berry, and Richards (2003) predicted the occurrence of lying from a participant’s linguistic style. Another study utilized a meaning extraction method to analyze open-ended self-narratives and identify dimensions with which people tend to describe themselves aside from typical itemized personality measures (Chung & Pennebaker, 2008). In turn, I-O psychologists may take a similar approach to identifying linguistic markers of leadership, using personality scores as a predictor of particular outcomes or as covariates, and estimating the attitudes and emotions of the subjects of study. Recently, Hogan X—a division of Hogan Assessments—has partnered with Receptivi.ai, a tech company that uses machine learning and natural language processing to turn unstructured data into personality insights based on the work by Pennebaker. The strategic partnership hopes to use Hogan Assessments’ years of personality research to make drawing personality signals from text analytics easier for researchers and practitioners. With a little hard work, the options are seemingly endless and is in keeping with the lexical hypothesis upon which many theories of individual differences are based.
Although we ourselves admittedly have much to learn about this new area and the methods involved, we believe the digital humanities approach to the analysis of historical documents presents an exciting, cutting-edge methodology for I-O psychologists to explore. Additionally, this type of research can help to catalyze collaboration with other researchers and promote an interdisciplinary approach to answering some of the most challenging questions within the psychological sciences. Further, it allows for I-O psychologists a more direct path to fulfill past calls to action for person-centric (Weiss & Rupp, 2011), humanistic (Zickar, 2010), and histiographic (Zickar, 2015) approaches to the study of work and workers. We hope this article will spark action and/or debate in the field to incorporate a digital humanities perspective into our research.
Aguinis, H., & O’Boyle, E. (2014). Star performers in twenty-first century organizations. Personnel Psychology, 67(2), 313-350.
Bose, S. & Goswami, S. (2016). RSentiment: Analyse sentiment of English sentences [software]. R package version 2.0. https://CRAN.R-project.org/package=RSentiment.
Call, M., Nyberg, A., & Thatcher, S. (2015). Stargazing: An integrative conceptual review, theoretical reconciliation, and extension for star employee research. Journal of Applied Psychology, 100(3), 623-640.
Carter, D. R., DeChurch, L. A., Braun, M. & Contractor, N. (2015). Social network approaches to leadership: An integrative conceptual review. Journal of Applied Psychology, 100, 597-622.
Chung, C. K. & Pennebaker, J. W. (2008). Revealing dimensions of thinking in open-ended self-descriptions: An automated meaning extraction method for natural language. Journal of Research in Personality, 46, 96-132.
Dalege, J., Borsboom, D., van Herreveld, F., van den Berg, H., Conner, M., & van der Mass, H. L. J. (2016). Toward a formalized account of attitudes: The Causal Attitude Network (CAN) model. Psychological Review, 123, 2-22.
Duke University Libraries. (2016). Introduction to text analysis: Analysis methods and tools. Retrieved from http://guides.library.duke.edu/
Gartner, J. D. (2008). In search of Bill Clinton: A psychological biography. New York, NY: St. Martin’s Griffin.
Gloss, A., Foster, L., Rupp, D., Scott, J., Saari, L., Osicki, M., C… Maday, D. (2016). United n Nations policy brief decent work for all: Leveraging big data for a human-centered approach to sustainable development. The Industrial-Organizational Psychologist, 53(4), 147-152.
Guame, B., Duvgnau, K., Prevote, L., & Desalle, Y. (2008). Toward a cognitive organization for electronic dictionaries: The case for semantic proxemy. Proceedings of the Workshop on Cognitive Aspects of the Lexicon (COGALEX), 86-93. doi: 10.3115/1598848.1598863
Guzzo, R. A., Fink, A. A., King, E., Tonidandel, S. & Landis, R. S. (2015). Big data recommendations for industrial-organizational psychology. Industrial and Organizational Psychology: Perspectives on Science and Practice, 8, 491-508.
Harlow, L. L., & Oswald, F. L. (2016). Big data in psychology: Introduction to the special issue. Psychological Methods, 21, 447-457.
Hernandez, I., Newman, D.A., & Jeon, G. (2016). Twitter analysis: Methods for data management and a word count dictionary to measure city-level job satisfaction. In S. Tonidandel, E. King, & Cortina, J. M. (Eds.), Big data at work: The data science revolution and organizational psychology. New York, NY: Routledge.
Landers, R. (2016). A crash course in I-O technology: A crash course in data visualization platform Tableau. The Industrial-Organizational Psychologist, 54(2). Retrieved from http://www.siop.org/tip/oct16/crash.aspx
McAdams, D.P. (2011). George W. Bush and the redemptive dream: A psychological portrait. New York, NY: Oxford.
Morrison Jr., J., & Abraham, J. (2015). Reasons for enthusiasm and caution regarding big data in applied selection research. The Industrial-Organizational Psychologist, 52(3), 134-139.
Newman, M.L., Pennebaker, J.W., Berry, D.S., & Richards, J.M. (2003). Lying words: Predicting deception from linguistic style. Personality and Social Psychology Bulletin, x29, 665-675.
Tonidandel, S., King, E., & Cortina, J. M. (2016). Big data at work: The data science revolution and organizational psychology. New York, NY: Routledge.
Wegman, L.A., Hoffman, B.J., Carter, N.T., Twenge, J., & Guenole, N. (2016). The changing nature of jobs: A meta-analytic investigation. Journal of Management. [Advance online publication].
Weiss, H. M., & Rupp, D. E. (2011). Experiencing work: An essay on a person-centric work psychology. Industrial and Organizational Psychology: Perspectives on Science and Practice, 4, 83-97.
Zickar, M. J. (2010). Recognizing the need for a humanistic movement within industrial-organizational psychology. Industrial and Organizational Psychology: Perspectives on Science and Practice, 3, 97-99.
Zickar, M. J. (2015). Digging through dust: Histiography for the organizational sciences. Journal of Business and Psychology, 30, 1-14.