Feldman Hope A, Papageorge Marianna V, Deboever Nathaniel, Goodrum Heath, Camp Samuel, Blackmon Shanda, Lawton Jennifer S, Antonoff Mara B
Division of Surgery, Department of Thoracic and Cardiovascular Surgery, The University of Texas MD Anderson Cancer Center, Houston, Texas.
Department of Surgery, Yale University School of Medicine, New Haven, Connecticut.
Ann Thorac Surg Short Rep. 2023 Aug 9;1(4):696-700. doi: 10.1016/j.atssr.2023.07.007. eCollection 2023 Dec.
Bias in letters of recommendation may have an impact on assessment of the cardiothoracic surgical applicant. Letters written for fellowship candidates might differ on the basis of the applicant's gender, although prior investigations have not explored this population. We aimed to evaluate gender differences in letters of recommendation written for cardiothoracic surgery applicants.
Letters of recommendation written for applicants between 2015 and 2020 to a large, 2-year academic cardiothoracic surgical training program were reviewed. Data regarding applicants and letter writers were captured. A library of terms to describe applicants was created. Natural language processing was employed to detect differences in use of these terms in letters written for men vs letters written for women. Letter content was compared across applicant and writer characteristics by Kruskal-Wallis and -tests.
There were 1707 letters of recommendation for 553 applicants analyzed: 430 letters for 147 (26.6%) women and 1277 letters for 406 (73.4%) men. Mean letter length was 498.4 (SD 209.3) words. Letters for men were longer (518.2 vs 491.7 words; .022). Men were more often described with terms of humility ( .01), professionalism ( .01), and integrity/honesty ( .01), whereas women were more frequently described by terms of family ( .01) and leadership ( .01). Men writers described men applicants more often with gendered language ( .01) and terms of humility ( .01), personality ( .01), and integrity/honesty ( .01).
We identified gender-based differences in letters of recommendation for cardiothoracic surgery applicants as well as differences based on the writer's gender. These findings are critical to consider in the assessment of applicants as efforts are needed to minimize bias and to support a diverse future of our specialty's workforce.
推荐信中的偏见可能会影响心胸外科申请人的评估。为申请专科培训的候选人所写的推荐信可能因申请人的性别而有所不同，尽管之前的研究尚未对这一群体进行探讨。我们旨在评估为心胸外科申请人所写推荐信中的性别差异。
回顾了2015年至2020年期间为一个大型的、为期两年的学术性心胸外科培训项目的申请人所写的推荐信。收集了有关申请人和推荐信撰写人的数据。创建了一个描述申请人的术语库。采用自然语言处理来检测在为男性撰写的推荐信和为女性撰写的推荐信中这些术语使用上的差异。通过Kruskal-Wallis检验和t检验，对不同申请人和撰写人特征的信件内容进行了比较。
共分析了553名申请人的1707封推荐信：为147名（26.6%）女性撰写的推荐信有430封，为406名（73.4%）男性撰写的推荐信有1277封。推荐信的平均长度为498.4（标准差209.3）个单词。为男性撰写的推荐信更长（518.2个单词对491.7个单词；P = 0.022）。男性更多地被描述为谦逊（P = 0.01）、专业（P = 0.01）和正直/诚实（P = 0.01），而女性则更多地被描述为家庭（P = 0.01）和领导力（P = 0.01）。男性撰写人在描述男性申请人时更多地使用性别化语言（P = 0.01）以及谦逊（P = 0.01）、个性（P = 0.01）和正直/诚实（P = 0.01）等词汇。
我们发现了为心胸外科申请人所写推荐信中基于性别的差异以及基于撰写人性别的差异。在评估申请人时，考虑这些发现至关重要，因为需要努力尽量减少偏见，并支持我们这个专业领域未来多样化的劳动力队伍。
