Implicit Gender Bias in Third-Year Surgery Clerkship MSPE Narratives

To assess implicit gender bias in surgery clerkship evaluations of third-year medical students at a large, academic hospital in the Southeast.

University of North Carolina at Chapel Hill School of Medicine has multiple branch campuses where students can complete their surgical clerkship including 1 large academic center, 1 hybrid academic and community-based practice, and 3 community-based hospitals. All residents and faculty evaluations of medical students who completed the surgery clerkship from March 1, 2018 to February 28, 2019 were analyzed. Evaluations were anonymized and names and pronouns were removed to mitigate evaluator bias. A word dictionary was created guided by previous literature and categorized descriptive adjectives into 4 categories: ability, grindstone, standout, and personality traits. Adjectives used to describe students, and references to the student using gendered pronouns or gender-fair language were coded and quantified as percentage of total evaluation word content. These percentages were compared between male and female students. A subsequent analysis was completed to assess the effects of gendered pronouns on linguistic patterns.

A total of 583 evaluations from the surgery clerkship were available for 183 students (51.9% female, 48.1% male). When gender-fair language was used, there was no difference in the adjectives used to describe female and male students. Male evaluators were more likely to use female gendered pronouns compared to male gendered pronouns (3.1% vs 2.3%, p = 0.028). When gendered pronouns were present, evaluations of female students were more likely to contain grindstone adjectives but less likely to contain standout terms compared to evaluations of male students (4.4% vs 2.8%, p = 0.006; 0.6% vs 1.3%, p = 0.006).

For students who have completed their surgical clerkship, the language patterns in evaluations differ between female students and male students. When the female pronoun was used, narratives contained more grindstone adjectives and fewer standout adjectives. Our results are consistent with previous literature and may be a manifestation of "othering" or a compensatory means of describing female students. This is potential manifestation of implicit gender bias.