Can Behavior-Based Interviews Reduce Bias in Fellowship Applicant Assessment?

Abstract

Components of trainee applications may introduce bias based on race or gender. Behavior-based interviews (BBIs) rely on structured questions to elicit applicants' past experiences to predict future behavior. Our objective was to implement BBIs in one fellowship program and compare applicant assessment by race and gender when using a standardized assessment tool versus a BBI-based tool. In 2019 and 2020, we developed BBIs and BBI-specific assessments; 6 of 15 faculty were trained in this interview method. Applicants completed 6 interviews with either a BBI or unstructured format. All faculty completed a standardized assessment on applicants. BBI faculty also completed a BBI-specific assessment. Normalized average scores were calculated and used to rank applicants into quartiles. Race was categorized into White, underrepresented minorities (URMs; Black and Hispanics), and non-URMs (all others). Faculty and applicants were surveyed about BBIs. Seventy-five applicants were interviewed. Significant differences were found in standardized assessment scores (White 1.01 +/- 0.09, non-URM 1.02 +/- 0.08, URM .94 +/- 0.07; P=.02) and quartiles by race (P=.05), but not for BBI scores (White 0.98 +/- 0.09, non-URM 1.03 +/- 0.09, URM 1.02 +/- 0.1; P=.18) or quartiles by race (P=.17). There were no significant differences in score or quartile by gender for either tool. The majority of faculty and applicant survey respondents commented positively about BBIs. BBIs were successfully implemented and generally reviewed positively by faculty and applicants. BBIs reduced racial differences in applicant assessments. Applicant assessment may benefit from structured tools to mitigate potential biases.