Florence E Turrentine1, Caitlin N Dreisbach2, Amanda R St Ivany3, John B Hanks1, Anneke T Schroen4. 1. Department of Surgery, School of Medicine, University of Virginia, Charlottesville, VA. 2. School of Nursing, University of Virginia, Charlottesville, VA; Data Science Institute, University of Virginia, Charlottesville, VA. 3. Department of Community and Family Medicine, Geisel School of Medicine, Dartmouth College, Hanover, NH. 4. Department of Surgery, School of Medicine, University of Virginia, Charlottesville, VA. Electronic address: ats2x@virginia.edu.
Abstract
BACKGROUND: Implicit bias has been documented in candidate selection within academic medicine. Gender bias is exposed when writers systematically use different language to describe attributes of male and female applicants. This study examined the presence of gender bias in recommendation letters for surgical residency candidates. STUDY DESIGN: Recommendation letters for 2016 to 2017 surgery resident applicants selected for interview at an academic institution were analyzed using qualitative text analysis, quantitative text mining, and topic modeling. Dedoose, QDA Miner, and RStudio analytic software were used for analysis. RESULTS: There were 332 letters of recommendation for 89 applicants (51% male) analyzed. Of 265 letter writers, 86% were male, 21% chairs, and 50% professors. Average word count was 404. Letter writers for male compared with female applicants had a significantly higher average word count (male = 421, SD 144; female = 388, SD 140, p = 0.035). Standout adjectives (eg exceptional), reference to awards, achievement, ability, hardship, leadership, scholarship, and use of applicant's name were most often applied to male applicants. Comments on positive general terms (eg delightful), grindstone words (eg hard-working), physical description, doubt raisers, and work ethic were most often applied to female applicants. Topic modeling and term frequencies revealed achievement words (performance, career, leadership, and knowledge) used more often with male applicants, while caring words (care, time, patients, and support) were used more often with female applicants. CONCLUSIONS: Gendered differences examined through language and text exist in surgical residents' recommendation letters. Implementing tools to help faculty write recommendation letters with meaningful content and editing letters for reflections of stereotypes may improve the resident selection process by reducing bias.
BACKGROUND: Implicit bias has been documented in candidate selection within academic medicine. Gender bias is exposed when writers systematically use different language to describe attributes of male and female applicants. This study examined the presence of gender bias in recommendation letters for surgical residency candidates. STUDY DESIGN: Recommendation letters for 2016 to 2017 surgery resident applicants selected for interview at an academic institution were analyzed using qualitative text analysis, quantitative text mining, and topic modeling. Dedoose, QDA Miner, and RStudio analytic software were used for analysis. RESULTS: There were 332 letters of recommendation for 89 applicants (51% male) analyzed. Of 265 letter writers, 86% were male, 21% chairs, and 50% professors. Average word count was 404. Letter writers for male compared with female applicants had a significantly higher average word count (male = 421, SD 144; female = 388, SD 140, p = 0.035). Standout adjectives (eg exceptional), reference to awards, achievement, ability, hardship, leadership, scholarship, and use of applicant's name were most often applied to male applicants. Comments on positive general terms (eg delightful), grindstone words (eg hard-working), physical description, doubt raisers, and work ethic were most often applied to female applicants. Topic modeling and term frequencies revealed achievement words (performance, career, leadership, and knowledge) used more often with male applicants, while caring words (care, time, patients, and support) were used more often with female applicants. CONCLUSIONS: Gendered differences examined through language and text exist in surgical residents' recommendation letters. Implementing tools to help faculty write recommendation letters with meaningful content and editing letters for reflections of stereotypes may improve the resident selection process by reducing bias.
Authors: Lindsay E Kuo; Heather G Lyu; Molly P Jarman; Nelya Melnitchouk; Gerard M Doherty; Douglas S Smink; Nancy L Cho Journal: JAMA Surg Date: 2020-09-02 Impact factor: 14.766
Authors: Alexandra Mannix; Sandra Monteiro; Danielle Miller; Melissa Parsons; Al'ai Alvarez; Sara M Krzyzaniak; Katarzyna Gore; Daniel Eraso; Dayle Davenport; Teresa M Chan; Michael Gottlieb Journal: AEM Educ Train Date: 2022-04-01