Yes, We're Fishing—In Rough Waters for Hard-to-Find Fish

In their focal article, Ruggs et al. (2013) outline the missed opportunities for researchers within industrial–organizational (I–O) psychology in examining marginalized employees. The authors identify seven groups as having been overlooked by I–O psychologists and thus deserving greater attention in the future. They conclude their focal article by noting that, “Instead of being on the front line serving as scientists and allies for those who are marginalized and treated poorly, we have let these individuals take a backseat while we have gone fishing.” We disagree with this assertion. It is not that we have gone fishing and ignored marginalized employees. Rather, we have gone fishing, in rough waters, to locate hard-to-find fish. We are not purposefully ignoring marginalized employees. On the contrary, we are conducting research while battling numerical representation issues and ethical and administration issues. The focus of our commentary is to highlight these challenges and offer suggestions for addressing them in an effort to assist researchers in actually doing what the authors of the focal article are calling for them to do—to successfully engage in more focused research on these under-represented members of the workforce.

Numerical Representation Issues: Our Fish Are Rare

One of the biggest challenges to studying marginalized groups in the workforce, whether those described by Ruggs et al. or others, is that their numerical representation is usually relatively small compared to other groups. To go with the fishing analogy, there may be a lot of fish in the sea (i.e., employees in the workforce), but if we're only interested in the rare longnose surgeonfish (our marginalized group), there will be few fish to catch. For example, it is estimated that a mere 4% to 17% of the workforce is lesbian, gay, bisexual, and transgender (LGBT; Lee, Comrey, Ragins,  & Cornwell, 2001). Similarly, according to 2012 statistics from the Office of Disability Employment Policy, only about one in five workers in the United States is disabled. Similarly, assessing racial minorities will likely result in relatively low numbers in comparison to Caucasians (e.g., Native Americans represent 1.7% of general population; U.S. Census, 2010).

The rarity of marginalized workers in the workforce can create havoc for researchers in terms of random sampling and power issues. If researchers simply cast a wide net by randomly sampling the entire organization, unless the organization and the subsequent sample is very large, the number of individuals constituting the marginalized employee group in the sample would probably be insufficient to reliably conduct analyses. For example, disabled employees comprised fewer than 1% of the federal workforce in recent years (Rosenberg, 2008), despite it being one of the largest employers in the United States (Bureau of Labor Statistics, 2012).

One obvious solution to this issue of low numerical representation would be stratified random sampling, essentially oversampling the numerically small groups relative to their representation in the population. However, this strategy leads to other challenges. Individuals who are members of strata that are (nearly) completely sampled could feel like the study is attempting to identify and target respondents; there is no safeguard against identification as it happens via random sampling. This could influence response rates or the responses people make on the survey; further, it influences the perceptions of reviewers about the extent to which survey responses from these individuals can be trusted.

Another solution is akin to the fish aficionados who really want those longnose surgeonfish but can't seem to find them amid all of the other species of fish in the South Pacific. They can either go to habitats that are densely populated with the longnose surgeonfish or they can find conduits to access them (e.g., purchase from a specialty shop). Researchers can take similar approaches with obtaining samples of marginalized employees. For example, LGBT samples can be from pride festivals, religious minorities could be recruited from their places of worship, or racial minorities can be contacted through organizations explicitly focused on them. But, just like the fish aficionados who purchase their rare fish in a shop or fish in special habitats, sampling marginalized employees in this manner can come with a hefty price tag. In this case, the cost is measured in terms of representativeness of the participants and subsequent generalizability of the data obtained from them. That is, there might be something different about individuals who are members of such groups versus those who are not. Not all LGBT individuals attend pride festivals, and those who choose to may be fundamentally different than those who choose not to attend such events. Similarly, religious minorities who attend worship services or ethnic minorities who are members of heritage-based groups might differ from those who do not. This issue is not lost on reviewers. Indeed, reviewers question the validity of such convenience samples, often citing the representativeness issue as a critical problem with the research. As a prime example, reviewers will often question whether people who are willing to participate in LGBT studies (especially those derived from a convenience sample) and/or who identify as LGBT on a survey differ from LGBT individuals who are not willing to participate or to identify themselves as LGBT on paper. We agree that there are probably psychological processes associated with being “out” that are also associated with a nontrivial number of research questions. However, the extent to which being “out” as LGBT is a circumspect variable is disconcerting because there are no other demographic categories—marginalized or not—on which the self-identifying responses of people are so scrutinized. Yet reporting one's own ethnicity, marital status, religion, and the like are also exposing oneself to potential stigmatization and often revealing that which was concealed. Thus the price for obtaining data in this way can be seen in the potential difficulty with getting such research published.

Another solution to addressing issues of numerical representation is somewhat contradictory to one of Ruggs et al.'s suggestions. Ruggs et al. recommend that researchers should not “lump” LGBT individuals together when researching; yet many of the suggested reasons for which individuals are discriminated against (e.g., perceptions of controllability, stigma) and the outcomes related to discrimination (e.g., distress) stem from similar processes. We agree that researchers cannot combine all groups when studying these issues and should consider groups in their unique contexts. However, we contend that theoretically sound solutions to issues facing various groups are likely to apply to other groups as well (as the underlying issue of why groups are treated differently applies to all groups). This is not to suggest that the challenges faced by lesbians are concomitant with those faced by gay men, nor the same as transgendered people, nor bisexual people. Nor are we suggesting that the struggles faced by LGBT individuals are the same as those faced by ethnic minorities, which are not the same as single workers, which are not the same as obese people, and so on. The historiosocial contexts in which members of these groups live, work, and play differs vastly. However, it is research on the underlying psychological principles of mistreatment and marginalization of groups, and not research on the particular groups themselves, that holds the answers to the problems faced by workers. We contend that these processes are the same even if the specific experiences and contexts differ.

For example, the underlying social psychological principles of stigma, discrimination, prejudice, and stereotyping (Fiske, 2002) are likely to be relevant to all groups that are mistreated. Although instances of overt discrimination are observed on a decreasing basis in contemporary society, covert or implicit biases are still prominent and negatively affect individuals (Dovidio & Gaertner, 1986; Fiske, 2002). Further, whereas categorization of people is a cognitive phenomenon that people come by naturally (Tajfel & Turner, 1985), it is also clear that such categorization is the root of stereotyping, prejudice, and discrimination (Fiske, 2002). Finally, it is clear that conceptualizing mistreatment as a stressor helps us understand the effects that mistreatment has on the individuals who experience it (Fitzgerald, Hulin, & Drasgow, 1995).

In addition, more work is needed on mistreatment and its various forms (e.g., incivility: Andersson & Pearson, 1999; sexual harassment: Fitzgerald et al., 1995; ethnic harassment: Raver & Nishii, 2010; bullying: Samnani & Singh, 2012; abusive supervision: Bowling & Michel, 2011; ostracism: Ferris, Brown, Berry, & Lian, 2008). There are a variety of sources of mistreatment in the workplace (e.g., supervisor, coworker, subordinate, client) as well as types of behavior (e.g., ostracism, incivility, etc.). Some questions that need to be addressed both theoretically and empirically are how differences in the source and form of mistreatment behavior differentially influence outcomes (Herschcovis et al., 2007) and whether these differ in perceived severity and subject appraisal (Lazarus & Folkman, 1984). Further, we have little understanding as to why some mistreatment is “approach” (e.g., abusive supervision, bullying) and some is “avoidance” (e.g., neglect, ostracism) and the factors that play into perpetration of one over the other. In fact, it is fair to say that although we know much about the negative effects of mistreatment on the targets, little is known about perpetrators and the possibly positive or negative effect of perpetration of mistreatment. Thus, we suggest researchers should further investigate the various underlying factors that lead to and result from dissimilar treatment of groups in addition to individual experiences in order to further the goal of lessening differential treatment in the workplace.

A final suggestion concerns methodology. That is, researchers should utilize and reviewers and editors should embrace qualitative research designs. This type of evaluation is more forgiving of small numbers and provides a rich and in-depth examination of the phenomena in question. Unfortunately, this methodology introduces another bias related to publishing in the key journals. Qualitative studies are in a sense a “marginalized group” and less likely to be published in top journals. This in itself indirectly keeps more research on marginalized groups out of the top journals.

Ethical and Administrative Issues: Overfishing and Permits

Another challenge to studying marginalized groups in the workforce involves administrative and ethical issues that accompany such endeavors. When fishing for rare fish, the ethical concerns may involve over-fishing to the point of extinction, which creates the administrative issues of having to get a hard-to-obtain permit to actually fish. With the study of marginalized employees in the workforce, the ethical issues involve privacy issues and concerns regarding confidentiality and anonymity. These concerns can create administrative nightmares in the form of being unable to identify such participants in an organization or obtaining an even smaller number of participants than should be expected to truly represent the makeup of the workforce.

As an example of how these ethical and administrative concerns can manifest themselves, consider sexual orientation. Most organizations do not keep demographic records on the sexual and gender1 orientations of its workers. Thus, the organization cannot identify potential LGBT participants from organizational records. Ethically, this is the appropriate stance for an organization, as sexual orientation is not a protected class in a large portion of the country nor is it usually job related. Nevertheless, it is often stigmatized by others, so obtaining and retaining such information would be inappropriate. Even when the characteristic of interest is protected (e.g., religion, disability), this information is not always readily available in employment records. This creates an additional barrier to good sampling procedures for these populations.

Concerns about anonymity and confidentiality may also be particularly salient for marginalized employees and lead to administrative woes. If a person is not out at work, it is difficult for that person to take part in a research study on LGBT issues unless all people are sent an invitation to participate in the study. Concerns on the participants' end, then, might occur when online surveys have unique addresses for each potential participant or when others might observe them participating in the study.

The solution to this issue might have to go beyond the researcher and be addressed by the practitioner as well. In the case of company-wide surveys, researchers might encourage organizational leaders to include a more exclusive array of demographic variables to allow the identification of many of the less-researched populations. Yet, this does come with the drawback of making the surveys more identifiable of any single participant. If this is the case, the choice of demographic variables should be based on the needs of the organization and their issues related to specific marginalized groups.

Another strategy for researchers conducting large studies might be to reach out to researchers whose area of expertise is in some of these marginalized groups and invite their collaboration on these projects. Partnering would benefit researchers by exposing them to different perspectives on their area of expertise and would benefit the expert by providing them with an opportunity to test their theories on a relevant sample. Researchers might also partner with community, professional, or advocacy groups that represent some of the interests of the relevant groups. Such groups might be able to provide access to a population and sample that is appropriate to the research question (although, as noted above, this could lead to concerns about the generalizability of the results). However, even if such groups are not the appropriate conduit for obtaining a sample, they are excellent resources for providing insight in survey design such that potential respondents feel protected and respected by the research process; such groups could also highlight recent developments in law and society that might influence the context in which the survey is taken.

Fishing Anyone?

We agree with the focal authors—I–O psychologists are out fishing. We contend that fishing can be a productive activity that can sustain life and benefit many. We believe that I–O psychologists need to reevaluate our fishing strategy and consider how our research questions will impact the groups we aim to study, how we can ethically study such groups while still maintaining high-quality scientific standards, and how we can draw conclusions about the underlying principles of mistreatment and marginalization. This is especially relevant for editors and reviewers who might need to assess the cost–benefit of not publishing a study due to methodological limitations that are a function of population characteristics. Until we know more about these groups that have been understudied—but not, in our opinion, overlooked—we cannot yet draw conclusions about the impact that these population characteristics have on the utility and generalizability of research on these groups.

• 1

  Gender orientation is not the same as sex. Sex refers to the assumed genotype of people, usually based on their external genitalia. Gender refers to the social construction of masculinity, femininity, and other gender roles, expectations, and stereotypes. Gender orientation therefore refers to the extent to which individuals ascribe to these particular gendered social constructions.