More human than human: The consequences of positive dehumanization

Abstract

While dehumanizing language, or comparison of humans to animals or machines, is commonplace in administrative rhetoric, there is little evidence of its consequences, particularly when used in its positive form, with intent to praise, rather than denigrate. Using a survey experiment, the authors provide respondents with an employee evaluation of a hypothetical employee that includes comments from a supervisor with treatment and experimental groups being exposed to different types of language. Results suggest that dehumanizing language can alter perceptions of employee competence, but it comes with a tradeoff related to perceptions of their personality. This raises questions about how administrative rhetoric creates images of individuals within organizations, in both positive and negative ways.

Keywords:

• Administrative rhetoric
• dehumanizing language
• survey experiment

Notes

1 This study was approved by the Institutional Review Board at Boise State University on May 8, 2019.

2 A full table of demographic comparisons is available in Supplemental Appendix C.

3 For Latinos, this may be a function of how we asked the question, as a combined race and ethnicity question, rather than asking ethnicity separate from race.

4 Simple random assignment at the individual-level was conducted by Qualtrics Survey Software’s randomization feature.

5 Instructions were left fairly vague, as participants were first told “Please read the following performance evaluation of an employee, Brad Harris, carefully. You will be asked questions about this employee in the remainder of the survey.” As such, it cannot be assumed that participants view Brad as a coworker, subordinate, or just another person they are evaluating. This involved minor deception (participants were not told that Brad was a fictional employee), but participants were debriefed about this and the goals of the study upon completion.

6 All results displayed included respondents who are both working and unemployed. Results are robust to including only currently employed participants, which are presented in Supplemental Appendix B. Note that a large portion of respondents are employed (93.8%), though there is have data on the proportion of respondents in white vs. blue collar jobs. However, MTurk workers tend to be more likely to have white collar jobs than the general population (Castille, Avet, & Daigle, 2019), suggesting that many of these participants are indeed working in white collar jobs.

7 Perhaps attitudinal responses contain inherent bias (such as the likelihood of marginalized groups to agree with statements more than majority groups), but have been widely used in a host of attitudinal research. Additionally, since racial and ethnic minorities are routinely dehumanized, perhaps the treatments will work differently on them. To this end, all relationships are examined using white respondents only, and results are substantively and statistically similar. These results are available in tables B9 and B10 of the Supplemental Appendix.

8 While these are 7-point scales, and OLS is a common analytical technique for regressions using experimental treatments, results are robust to ordered logit models for tables 1 and 2. Analyses using seemingly unrelated regression to replicate tables 1 and 2 provide statistically and substantively similar results. These results are available in Supplemental Appendix B.

9 Cronbach’s α (0.84) indicates a reliable scale. We conduct analyses of each indicator separately, as well, which are available in Supplemental Appendix B.

10 A full table of means and standard deviations of all dependent variables is available in Supplemental Appendix C.

11 Cronbach’s α (0.88) indicates a reliable scale.

12 Full question text is available in Appendix A.

13 Cronbach’s α = .66 for human uniqueness, .73 for human nature.

Additional information

Notes on contributors

Stephen Utych

Stephen Utych is an Assistant Professor of Political Science at Boise State University. He earned his PhD from Vanderbilt University, and his research focuses on political psychology and political behavior.

Luke Fowler

Luke Fowler is an associate professor and director of the Masters of Public Administration program at Boise State University in Boise, Idaho. His research interests include state and local politics, intergovernmental relations, environmental policy, and policy implementation.