Retrain in Data Analytics

Guzzo et al. describe the 21st century world as “awash in data” (p. 491). We argue that the existence of large quantities of data is neither new nor particularly exciting; what makes the big data revolution so remarkable is our unprecedented ability to make sense of vast amounts of information. Contemporarily we can store, analyze, and draw meaningful conclusions based on gigantic data sets, which would have been unfeasible to grapple with using yesterday's computers. In other words, when we talk about big data, we are inherently (if not implicitly) discussing big data analytics (Shaw, Reference Shaw2014). Thus, our first recommendation for remaining relevant as I-O psychologists is to retrain ourselves in big data analytics.

As social scientists, the current big data paradigm shift invites us to reframe our cognitive architectures surrounding research methods and data analysis. Specifically as psychologists, we have been trained to use certain (arbitrary) rules of thumb, such as a minimum sample size of 20 and an alpha level of .05. Although we generally understand these heuristics to be potentially misleading and problematic (e.g., Cohen, Reference Cohen1994), we still unanimously cling to them, upholding them as our scientific standards. What is perhaps unsettling for some about the growing popularity of big data is that it challenges us to rethink our homegrown scientific principles. After all, 20 participants are a proverbial drop in the bucket when you have a sample size of 20,000, or 200,000, or even 2,000,000.

Consequently, I-O psychology is currently at a crossroads; as scholars, we are being challenged to explore unfamiliar research methods in order to meet the demands of big data. Big data analysis supersedes functional silos, uniting interdisciplinary scholars who study a vast array of topics using common methods. Therefore, as a discipline it is critical for us to embrace methodological innovations from outside the field of I-O; for instance, the field of network science has a plethora of analytic and visualization tools that are well suited to big data. For example, relational event network modeling provides an adaptable framework for modeling longitudinal social interaction data (Butts, Reference Butts2008). Relational event models provide a look at process as though it were a continuous movie rather than several snapshots in time. These models provide the opportunity to understand the dynamics of collective interactions (Leenders, Contractor, & DeChurch, in press). We do not mean to imply that psychologists do not bring a lot to the table. Our substantial emphasis on theory-based research and aversion to data dredging (i.e., “dust-bowl empiricism”), among other things, make us important contributors to social science research using big data analytics.

Reapproach Organizational Theory Through a Big Data Lens

Rapid technological advancements are not only changing the way that people interact with one another but also changing the way that we can approach organizational theory. Researchers have access to data that were previously unobtainable or very cumbersome to access, process, and analyze. Access to these new informational sources pushes us to think about (a) answering questions that we have not been able to answer previously and (b) asking new questions that require us to rethink or expand our current theories.

First, we have the opportunity to answer longstanding questions. For example, researchers have struggled to understand the mechanisms that explain how social systems emerge, maintain, and dissolve. Due to heavy reliance on survey methods, we are often limited in opportunities to observe the evolution of interactions in social systems. Big data has opened a new avenue for researchers to investigate the emergence of phenomena in ways that were never before available (Lazer et al., Reference Lazer, Pentland, Adamic, Aral, Barabasi, Brewer and Van Alstyne2009). As an example, this opportunity is clear in teams research. Although research on teams is often concerned with emergent constructs (i.e., trust, cohesion, shared mental models), researchers have been unable to measure the nature of emergence directly with traditional quantitative measurement approaches (e.g., self-report surveys; Kozlowski, Chao, Grand, Braun, & Kuljanin, Reference Kozlowski, Chao, Grand, Braun and Kuljanin2013). Instead, measures of interactions and team processes have relied on individual perceptions, which are then aggregated to the team level to represent an emergent construct. Big data, combined with analytic techniques such as the relational event models mentioned above, can provide the means for researchers to observe how dynamic interactions among team members coalesce into emergent properties over time.

Second, we have the opportunity to expand our theories. For instance, imagine you are a job candidate, and you are sitting down to take an employer's selection test. However, instead of the typical paper-and-pencil intelligence or personality exams, your potential employer presents you with an app-based video game (à la Angry Birds) to play for 20 minutes. Over the course of that time, the employer collects several megabytes of data, which are subsequently analyzed to reveal your standing on a variety of job-relevant characteristics such as creativity, decision-making ability, and intelligence. This scenario, however futuristic and far-fetched it may seem, is a description of an actual employee selection process designed by the Silicon-Valley-based startup Knack. By combining innovative assessment methods and the power of big data, Knack has pioneered a new theory-driven form of employee selection (Noguchi, Reference Noguchi2015; Peck, Reference Peck2013), which—among other assets—all but negates the issue of job candidate social desirability. Knack is just one example of a company that has exploited big data analytics in order to augment and reengineer a traditionally time-consuming and costly human resource process. This case study highlights how big data changes how we think about fundamental I-O concepts (e.g., selection, creativity, decision making, etc.). We must continually expand our theories to include new conceptualizations and operationalizations of old constructs, as well as entirely novel social phenomena (e.g., virtual communities, online gaming).

Moreover, big data is changing the landscape of research, and it is essential that I-O psychologists embrace this change and advance current thinking about organizational phenomena. We are slowly beginning to see responses to this change in the field. For example, the recently announced journal Academy of Management Discoveries aims to publish empirical research that “establishes a [new] phenomenon” important for understanding organizations (Van de Ven, Reference Van de Ven2015). This journal is open to different philosophies of science, thus opening the door for research driven by new or poorly understood phenomena that are not yet explained by existing theories. Academy of Management Discoveries is a prime example of a platform for social scientists to explore big data analytics in order to answer longstanding questions, and we believe that it is critical that I-O psychologists actively participate and immerse themselves in this burgeoning area of exploration and discovery.

Reengage in Interdisciplinary Research and Teaching

As aforementioned, Knack is a pioneering people analytics startup that employs big data methodologies. In part, the secret to Knack's innovativeness can be found in the company's commitment to interdisciplinarity; a functionally diverse team of neuroscientists, psychologists, and data scientists designed Knack's avant-garde, big-data-analytic assessment methods (Noguchi, Reference Noguchi2015; Peck, Reference Peck2013).

According to Guzzo and colleagues, big data “is the great equalizer in terms of interdisciplinary research and practice.” They suggest that sources of big data can be “of value to other disciplines, thus creating occasions for the comingling of perspectives, theories, and analytics techniques” (p. 493). We wholeheartedly agree that big data can provide a point of integration with other disciplines. However, a considerable amount of lip service arguably gets paid to the notion of interdisciplinarity without nearly as much actual, cross-functional research being carried out. Thus, our final recommendation is to leverage the big data revolution by conducting interdisciplinary research with scholars from other disciplines. To clarify, Guzzo et al. note that researchers from different disciplines may be interested in the same dataset, and thus I-O psychologists should share their data. We suggest that I-O psychologists take this a step further and conduct interdisciplinary research, which requires scientists from different disciplines to work together to ask new questions and use new approaches that could not otherwise be asked without the synthesis of two or more disciplinary perspectives (Hall, Feng, Moser, Stokols, & Taylor, Reference Hall, Feng, Moser, Stokols and Taylor2008).

The wave of the future is clear—scientists across many disciplines are taking notice of the power of big data analytics. However, harnessing the power of big data is often easier said than done, especially for scientists trained in the social realm. Therefore, a main priority related to interdisciplinarity needs to be adjusting classical I-O psychology curricula to incorporate data science education. In order for I-O psychologists to continue meaningfully engaging with their interdisciplinary peers, we need to afford our young scholars the opportunity to learn contemporarily relevant research and analytic methods in a structured educational environment rather than dogmatically adhering solely to our traditional research and statistical methods.