Racial and Ethnic Bias in Letters of Recommendation and Personal Statements for Application to Otolaryngology Residency.

Introduction: The persistent lack of racial and ethnic diversity within the field of otolaryngology calls for an analysis of potential bias within the residency application system. Letters of recommendation (LORs) and personal statements (PSs) are the most important subjective application constituents. This subjectivity predisposes these components to implicit bias. In applications to various surgical subspecialties, prior linguistic studies assessing bias in reviews of LOR show race-based differences. Thus far, racial and ethnic linguistic differences in LORs for otolaryngology applicants have not been analyzed in the literature.

Methods: LORs and PSs were abstracted from otolaryngology - head and neck surgery applications in the Electronic Residency Application Service for the 2019-20 and 2020-21 application cycles. Linguistic Inquiry and Word Count 2015 was used for quantitative analysis of emotional, cognitive, and structural components of written text.

Results: Race-pair analysis of the 2019-2021 application cycles revealed higher mean "teaching" scores for LORs for Asian, black, Hispanic, and white applicants when compared to applicants who self-identified as Other. White applicants had lower scores for the terms "research" and "analytic" when compared to Asian and black applicants, respectively. Analysis of PSs revealed greater scores for an "authentic" writing style for white versus Asian applicants. White applicants were found to have higher scores for "tone" compared to black applicants.

Conclusion: Minor racial and ethnic language differences exist in both LORs and PSs. A statistically significant difference was observed among LORs, with the "teaching" term used more frequently for Asian, black, Hispanic, and white applicants compared to self-identified Other individuals. For PSs, statistically significant differences were observed among white applicants, who wrote about themselves using more "authentic" language when compared with Asian applicants and who also had higher scores for "tone" compared to black applicants. Although these differences were statistically significant, the practical impact of the variances is likely small.