(1076) Gender Differences in Letters of Recommendation for Fellowship Applicants: A Cross-Sectional Study

Introduction: Prior work found gender biases in letters of recommendation (LOR) for pulmonary and critical care medicine (PCCM) fellowship applicants at a single institution, with female applicants receiving favorable and longer LORs. Therefore, we sought to explore whether geographic variation exists in the use of gendered language within LORs for applicants applying to PCCM fellowship across different PCCM fellowship programs.

Methods: Fellowship applications submitted to 3 PCCM programs (South, Midwest, and West) in the U.S. in 2021 were included. Applications were deidentified with applicant accomplishments and demographics extracted. We identified the gender of the letter writer through a search of professional websites, and excluded applications and LORs if the gender of the applicant or author could not be ascertained. Linguistic Inquiry and Word Count (LIWC) analysis program was used to identify the different adjective frequency. The primary outcome was an impactful measure of the degree of support included in each LOR; secondary outcome was word count. We performed multivariable linear regressions, adjusting for applicant demographics, accomplishments, and letter author gender.

Results: Of the 1,705 applications, 6,239 LORs were included in the analysis. Most of the applicants (66.4%, N=1,115/1,679) and letter writers (68.6%, N=4,283/6,239) were male. Across all three programs, there was no statistically significant difference in the impactful outcome for female applicants as compared to males (Midwest: 0.14 more words 95% CI: -0.03-0.32; South: -0.01 more words 95% CI: -0.20-0.18; West: 0.10 more words 95% CI: -0.10-0.30).In both the Southern and Midwest programs, females had longer LORs as compared to male applicants (South: female =40.1 more words, 95% CI: 12.3-68.0, Midwest: female = 29.8 more words, 95% CI: 0.7-58.9). No statistically significant gender difference in letter length was found in the Western program.

Conclusions: There are regional differences in LORs, but only as it pertains to length of the LORs. Females did not stronger LORs as compared to male applicants. Future work is needed to explore the utility of LORs and how the information garnered is viewed by selection committees.