Reducing Racial Similarity Bias in Interviews by Increasing Structure: A Quasi-Experiment Using Multilevel Analysis

Abstract

We used a quasi-experiment to examine the relationship between interview structure and racial bias in simulated employment interviews. Although recent meta-analytic findings suggest that high-structure interviews (as compared with low-structure interviews) reduce same-race bias in interview ratings, this claim has not been tested using research designs that allow for stronger inference about the potential links between interview structure and bias. We showed videotaped interviews to 386 business students and, within both high- and low-structure conditions, determined levels of racial similarity bias using multilevel analysis. Our study used a fully crossed design within each condition, where all raters evaluated all applicants. As we expected, findings indicated that interviewers favored racially similar applicants less in high-structure interviews than in low-structure interviews. Our findings provide quasi-experimental evidence that increased interview structure may be effective to suppress racial similarity bias in employment interview ratings.

Abstract

Impact and Implications

Our findings illustrate the benefits of imposing structure in personnel selection interviews to reduce racial bias in interview ratings. Organizations that rely on high-structured interviews, rather than the more commonplace low-structure interview type, are likely to observe lower levels of racial similarity bias in ratings. The persistent and widespread use of unstructured interviews in organizations exposes applicants to unfair selection practices and increases the risk to organizations of employment discrimination litigation. Furthermore, these interviews withhold the benefits to organizations (e.g., better selection decisions) and applicants (e.g., fair employment practice) that accrue from interview ratings undistorted by bias. Given the focus of our research on fairness issues in employment and organizations, our findings may be important to help meet the following 2016–2030 United Nations Sustainable Development Goals: help organizations to promote inclusive, productive, and decent work for all (number 8); reduce inequality within countries (number 10); and promote peaceful and inclusive societies, provide access to justice for all, and build effective, accountable, and inclusive institutions at all levels (number 16).