Race and Gender Bias in Clerkship Grading

Abstract

Phenomenon: Existing literature, as well as anecdotal evidence, suggests that tiered clinical grading systems may display systematic demographic biases. This study aimed to investigate these potential inequities in-depth. Specifically, this study attempted to address the following gaps in the literature: (1) studying grades actually assigned to students (as opposed to self-reported ones), (2) using longitudinal data over an 8-year period, providing stability of data, (3) analyzing three important, potentially confounding covariates, (4) using a comprehensive multivariate statistical design, and (5) investigating not just the main effects of gender and race, but also their potential interaction. Approach: Participants included 1,905 graduates (985 women, 51.7%) who received the Doctor of Medicine degree between 2014 and 2021. Most of the participants were white (n = 1,310, 68.8%) and about one-fifth were nonwhite (n = 397, 20.8%). There were no reported race data for 10.4% (n = 198). To explore potential differential grading, a two-way multivariate analysis of covariance was employed to examine the impact of race and gender on grades in eight required clerkships, adjusting for prior academic performance. Findings: There were two significant main effects, race and gender, but no interaction effect between gender and race. Women received higher grades on average on all eight clerkships, and white students received higher grades on average on four of the eight clerkships (Medicine, Pediatrics, Surgery, Obstetrics/Gynecology). These relationships held even when accounting for prior performance covariates. Insights: These findings provide additional evidence that tiered grading systems may be subject to systematic demographic biases. It is difficult to tease apart the contributions of various factors to the observed differences in gender and race on clerkship grades, and the interactions that produce these biases may be quite complex. The simplest solution to cut through the tangled web of grading biases may be to move away from a tiered grading system altogether.

Keywords:

• Grading
• clerkships
• bias
• race
• gender
• MANCOVA

Acknowledgments

The authors wish to thank Logan Butler and the Medical Education Outcomes Center for their assistance providing the data used in this study, and Bethany Schowengerdt for her assistance editing the manuscript.

Ethical approval

Ethical approval for this work was granted by the University of Minnesota Institutional Review Board on September 4, 2019. Study number: STUDY00007525.

Previous presentations

A virtual poster summarizing this research was presented at Best Practices in Health Sciences Education, University of Minnesota, Minneapolis, Minnesota, May 1–5, 2022.

Notes

1 A note on language: Although often used interchangeably, sex and gender are related but distinct concepts, as are race and ethnicity. In the U.S., certain terms such as “minority” and “Underrepresented in Medicine” (URM) have also historically been used to describe the race and/or ethnicity of an individual. To understand the total body of literature exploring assessment biases, we reviewed literature that referenced any of these related concepts. In order to most accurately reflect the original work, we have used the language present in the original articles when referring to the existing literature. It is important to note, however, that expectations for language use have evolved since the literature was originally published, in part because imprecision of terms can muddy our understanding of the potential connection between socio-demographic characteristics and outcomes such as grades. Additionally, some of the terms that have historically been used, such as “underprivileged” or “minority,” are now recognized as having dehumanizing, pejorative, and/or stigmatizing connotations. In our study, we analyzed data specifically regarding participants’ self-reported gender and race, while endeavoring to avoid imprecise and potentially stigmatizing terminology when reporting findings and drawing conclusions.

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.