Letters of recommendation for pediatric surgery fellowship: Analysis of linguistic differences based on gender of the applicant

Background:: Gender bias in letters of recommendation (LOR) has been reported for candidate selection in academic medicine, typically with females frequently described with communal terms (e.g. helpful, kind, interpersonal) and males described more with agentic terms (e.g. assertive, intellectual, ambitious). This study examined the presence of linguistic gender differences in LOR for Pediatric Surgery Fellowship. Methods:: LOR submitted to a single pediatric surgery fellowship program between 2014 and 2018 were retrospectively reviewed and analyzed using a previously validated Linguistic Inquiry and Word Count Software (LIWC) program. Descriptive statistics and bivariate analysis were employed in our analysis. Multivariable logistic regression models were built to assess independent association of LIWC variables with applicant gender, LOR writer gender, and applicant-writer gender concordance. Results:: 1264 LOR from 325 applicants (51% female) were analyzed. Of the letter writers, 83% were male, 57% were professors, and 7.6% were Pediatric Surgery Fellowship Program Directors. The overall median average word count was 518 words, with no significant difference in LOR word count between applicant genders. Compared to male applicants, female applicants were described significantly more with work words (e.g. excellent, work; p = 0.04). Male LOR writers used authentic words (e.g. honest, humble; p = 0.006) and home words (e.g. family, house; p = 0.04) significantly more than female LOR writers. There were no significant differences in the use of agentic and communal words between genders. Conclusion:: While there are linguistic differences in LOR for candidates in the Pediatric Surgery match based on gender, previously described overt gender bias was not seen in this study. Interestingly, female candidates were described more with work words, like excellent; a reversal of previously described gender bias in academic medicine. These findings may be due to the unique interpersonal and multidisciplinary skills required in pediatric surgery and may represent a unique form of gender bias that warrants further study.