JMI Revisionist History of Workplace Deviance

Abstract

In the mid-90s, we embarked on establishing the domain of Workplace Deviance. Though we were fortunate to meet our intended goals and have the impact we had hoped for, we have often thought about what we might have done differently. In this essay, we outline some of the things we wish we knew then that we know now. As we will describe, we perhaps should have chosen a different construct name, taken a theoretical rather than data driven approach to our typology, and developed a reflective rather than formative scale. We hope this essay based on our hindsight may be of value to future scholars seeking to establish new constructs in our field.

“Disgruntled former employee returns to his former workplace and opens fire on his coworkers, killing six then himself” It was the early 1990s, and weekly, national headlines reminded us of the latest shooting at yet another US Postal plant or workplace. In fact, from 1983–1993, there were so many incidents of U.S. Postal workers shooting and killing managers, coworkers, police offers and members of the public that these acts of rage came to be referred to as “going postal.” We were two years apart in Northwestern's doctoral program and both of us had included destructive workplace behaviors in our dissertations. We wondered what these antisocial workplace behaviors might have in common. Were they part of a broader construct? What exactly did we know about them, if anything, and what did we still need to learn?

We started thinking about how we might go about studying this important phenomenon. In organizational literature dominated at the time by functional topics, such as justice, negotiations, commitment, and culture, we were able to cobble together isolated papers in a variety of fields that reflected the dark side of organizational life. We found papers on employee theft (Greenberg, 1990), petty tyranny (Ashforth, 1994), sexual harassment (Gutek, 1985), employee drug and alcohol use at work (Mangione et al., 1975), and withdrawal (Gupta & Jenkins, 1983). It was clear this was a neglected aspect of organizational scholarship, and we felt it would be valuable for us to consider these behaviors as a whole, much like the citizenship scholars had done in clustering prosocial behaviors under one umbrella construct. We were warned it might be difficult to study however, given these were relatively low base rate behaviors that employees likely wanted to hide. But by 1995, we had taken several actions to bring attention and coherency to the domain of harmful workplace behaviors. We developed and defined the construct, which we labelled Workplace Deviance, we wrote several narrative reviews summarizing the nascent literature to date (Bennett & Robinson, 2003; Robinson & Bennett, 1997), we created a typology of its many behavioral manifestations using multi-dimensional scaling (Robinson & Bennett, 1995), and we developed validated scales assessing both interpersonal and organizational workplace deviance (Bennett & Robinson, 2000) that would hopefully facilitate the building out of this area of study.

Looking back on our early efforts, we are delighted to have succeeded in our initial goals. By last count, Google Scholar tells us that our early papers have garnered 12,000 citations so we are proud of that. However, there are a number of things that we wish we had known 30 years ago that we have learned by 2023. With the benefit of hindsight bias, we have often wondered if we should have used a different label than Workplace Deviance, developed a more theoretically derived typology, or created reflective scales rather than formative ones. Let us explain below, and in so doing, give food for thought to future authors pursuing new domains of study in organizational behavior.

The Branding Question: Should We Have Chosen the Term Workplace Deviance?

We have asked ourselves whether we could have used a better name for our construct. We invested a lot of time developing a very precise definition for our broad construct; we defined it as “voluntary, norm-violating behavior that threatens the well-being of an organization and/or its members” (Robinson & Bennett, 1995). We used the label workplace deviance to reflect the strong influence of the rich literature on deviance in sociology which guided our theorizing about norm-violating behaviors. If it worked in sociology, why not organizational behavior? This focus on deviance was especially useful because it enabled us to take into account the norms of the social context in which the behavior occurred, since norms, and therefore what is considered deviant, could vary across organizations or sub contexts. We also believed it would enable us to distinguish workplace deviance from unethical behavior, as workplace deviance would reflect violation of organizational norms whereas unethical behavior reflected much broader violations of fundamental principles or societal norms. We still believe our construct label is conceptually correct and aligned with these goals, but it was not without its downsides.

The problem we did not expect is that the “deviance” term would not work so well from a branding and use perspective. Did it sound “too out there” for organizations, maybe too aligned with connotations of criminality or perversion? We think this for a few reasons. First, it took us almost five years to publish our measurement paper after our typology. In fact, upon initial submission to Journal of Applied Psychology, our measurement paper was rejected. It was only on advice from more senior mentors that we waited for the editorship to change and tried again with the new editor with full disclosure of the paper's history.

Second, it is evident that this choice of label was not as well received as the label “counterproductivity” coined by Fox and Spector (1999). Indeed, Paul Spector acknowledged in a recent blog post that he and Susie Fox intentionally “branded” with a cool acronym (CWB) to gain attention. Their original 1999 CWB paper used an unvalidated scale which they describe as being derived from our 1995 typology. We were often asked by reviewers to clarify how our construct was different from CWB which was frustrating because they are indeed difficult to distinguish item by item and using the same two dimensions. And interestingly It seems to us that the label “CWB” is used more often in papers than the term “Workplace Deviance” even when authors use our measure. Having said that, we would not have used this term ourselves, as to us it felt too imprecise, seeming to capture more productivity-oriented behaviors at the exclusion of social behaviors. What we did not appreciate back then was that while precision matters, so too does the need for a good, branded term that could be easily embraced.

Why is this an issue? First, having both CWB and Workplace Deviance and various other terms used in the literature for the same construct has created unnecessary noise. Having consistency between terminology, definitions, and operationalizations is a basic tenet of good science. Having multiple labels, definitions and measures for the exact same thing muddies the waters. We would encourage scholars developing new constructs to think carefully about whether and how their construct is distinct from others and how they can best employ both precision and branding to lead the way.

Second, our choice of the term deviance led to a misunderstanding that we were erroneously suggesting that deviant behavior was always harmful. Indeed, some began to study “positive deviance” suggesting we had missed the mark by not studying norm-violating behaviors that were intended to help the organization. But they were mistaken. We weren’t interested in whether individuals might violate norms with an intent to help the organization. Our primary interest was on harmful behaviors; thus deviant (norm-violating) behavior that could potentially be harmful to individuals or organizations. We did not assume nor suggest that all deviant behavior was necessarily harmful or even intended to be harmful. It is noteworthy that related terms such as antisocial, counterproductive, or dysfunctional do not imply that there is an overlooked inverse positive aspect to the construct. Indeed, the opposite of antisocial is prosocial, an entirely different construct. For this reason, we feel like our use of the sociological term deviance dragged us into the weeds a bit.

What would we have called it instead, you may ask? A few contenders come to mind that might have been catchier, along with nice acronyms. For example, maybe we should have considered Destructive Organization Behavior (DOB), Organizational Insubordination Behavior (OIB) or our personal favorite, Harmful Actions at Work (HAW). Sadly it is too late now to rebrand now without making things even more muddy (e.g HAW, previously known as Workplace Deviance…)

Should We Have Created a Theoretical Derived Typology?

We developed our construct using multi-dimensional scaling (MDS), which identifies how the many manifestations of deviant workplace behavior are perceived by people in general to be similar and different to one another (Robinson & Bennett, 1995). We did so because we knew we needed to provide meaningful coherency and parsimony to all these behaviors if we were to study them. Rather than do so by rely on only the perceptions of the two of us, it was more valid to pull from the perceptions of a large swath of people. The result is a visual map that distributed the deviant behaviors according to their perceived similarities and differences, along with the underlying dimensions that reflect these similarities and differences. The farther apart the behaviors on the visual map, the more different they are perceived to be on average. Our resulting typology uncovered four categories of workplace deviance that included specific behaviors that clustered near one another: production deviance, property deviance, interpersonal deviance and political deviance. These behaviors taken together fell along two dimensions: how serious the behavior was and whether it targeted organizational members or the organization itself.

Although this typology has been well received, we have at times questioned if this was the right approach for several reasons. First, the typology emerged from the data, rather than being driven by theory. As a result, it has never worked well as a theory, and few have sought to develop a theory around our typology. This has meant that while people reference our typology as a starting point, our typology hasn’t theoretically led the field in this domain. Had we developed a theoretically derived typology, rather than one naturally emerging from the data, it is possible that the dimensions and types of workplace deviance could have been studied together with theory capable of explaining their antecedents as well as their consequences. For example, although we end up with two types of deviance: interpersonal and organizational, few studies have sought to make or apply parsimonious theory that explains their differing causes and consequences. Similarly, little advancement has been made to further understand the two dimensions of seriousness and targets or the four categories. As a result, the typology hits a dead end: it was useful for suggesting how individuals categorize harmful behaviors, and for pointing us toward behavioral measurement scales (Bennett & Robinson, 2000), but theoretically it did not move the field forward.

Related to the point above, we believe that due to its lack of theory, our typology hasn’t adequately accomplished one of the main goals we intended for it. We had hoped that our typology would a) pull together literatures on disparate deviant behaviors – for example at the time, there were disconnected studies looking at theft, alcohol consumption, and absenteeism- and b) encourage the field to study these behaviors in a more holistic way. In subsequent years, however, the opposite has happened; that is, a lot of different constructs have emerged, such as incivility, social undermining and abusive supervision that have proceeded to develop in siloed and often redundant fashion with one another. Indeed, a recent review identified 91 different constructs in our field that fit under the umbrella of workplace deviance (Zhong & Robinson, 2021).

Third, it is important to note that our MDS derived typology reflected how the behaviors appear to be similar, not necessarily how they may co-occur together in real life. Given that most of our studies examine co-occurrence, this is critical. Fortunately, our subsequent confirmatory factor analyses of these behaviors showed that these behaviors tend to co-occur, enabling us to develop measurement scales for them. However, most of the more serious behaviors, such as sabotage, political backstabbing, and violence, ended up being dropped because they did not occur frequently enough to be correlated with other behaviors. As a result, our measurement scales left out a large portion of the manifestations of workplace deviance that appear in the typology.

Should We Have Developed a Reflective Scale Instead of a Formative One?

One solution to the above problem would have been for us to develop a reflective measurement scale, rather than the formative one we developed (Bennett & Robinson, 2000). In other words, rather than seek out items that were behavioral manifestations of workplace deviance that occur across most contexts, such as I have falsified receipts” or “I have intentionally worked slower than I could have, we could and should have developed a measure for which each item reflected the key elements of our definition, which is that it a) violated key organizational norms and b) was potentially harmful and c) was volitional (not accidental). Examples of such reflective items would be, “I intentionally engage in harmful behaviors that break the norms of our organization.”

Our choice of scale was dictated by the norms of our field at the time, and we were unfamiliar with reflective scales. This is unfortunate because a reflective scale would have been superior for several reasons. First, it would have enabled us to capture the more seriously harmful behaviors that occur in the workplace, ones that had to be eliminated due to low variance in our formative measure. It is a problem for our original measures that we couldn’t account for the occurrence of more seriously harmful behavior, because, despite their lower frequency of occurrence, they may be much more important for us to understand given their high costs to organizations and to employees.

Second, a reflective scale would have enabled us to create a measure that more capably transcends workplaces, contexts, cultures, occupations, and time. As the world of work evolves from greater use of technology, remote work, teamwork, etc., so too have the specific manifestations of deviant behaviors of employees. Our field needs measures that can transcend geography and time. Our field is increasingly global and recognizing the cultural boundaries of our knowledge is critical. Likewise, the world of work continues to change at a rapid pace. Being able to use the same measures over time, to enable comparisons and to coherently build our knowledge is important.

Finally, it would have been very useful to have a measure that is sufficiently broad so that it does not compete for space with the proliferation of more specific behavioral measures of harmful deviance, such as incivility, social undermining, aggression, or ostracism (which as we now know often overlap tremendously with regards to scale content, Hershcovis, 2011). At this point, we have many overlapping constructs, often measuring the exact same things in studies examining the same outcomes using the same theories. We are finally moving toward consolidation of those various behaviors across studies, such as Zhong and Robinson (2021) and Zhong et al. (2023), which will enable us to learn across studies, and save precious research resources.

Although it is maybe a bit late, we have finally developed a new reflective measure of workplace deviance (Bigelow et al., in press). Not only will this measure be applicable to absolutely any culture, work context, occupation, or time, but it is also short. This means it will also be more practical for studies that require many measurement points, such as those employing event sampling methods or for those embedded in longer or more complex surveys.

In sum, we have both enjoyed and appreciated the opportunity to introduce the “dark side” of workplace life, and we have been pleased with how our work has been received in this area. Nevertheless, as we have argued, the benefit of hindsight bias has made us process a few regrets. We hope our insights provide some food for thought for others venturing into developing new domains in organizational behavior.

Even if this essay does not provide useful food for thought to future scholars, it was a fascinating experience for us to write it. It was fun reliving some of our early discussions and how we got started. In so doing, we came to realize we hadn’t remembered things exactly the same way; for example, we had different “origin stories” of how our shared interest in workplace deviance came about. It was also so interesting to look back at the literature at the time to see how far it has come and where it has gone. Until this essay, we had not appreciated the importance of reflecting on what we might have done differently now that we know what we know. Doing so can only make our own scholarship and that of the field stronger. We think writing one's own revisionist history would be especially useful as one approaches full professor standing, in thinking about what one might do differently with the rest of their career. After all, as the great writer George Eliot said, “it's never too late to be what you might have been!”