Comparison of methodology in the collection of gender and race/ethnicity in hematology and oncology journal editorial boards.

Abstract

11061 Background: Reporting on disparities is strongly influenced by the methodology used to collect race/ethnicity and gender data. Incorporating gender and race into research has its challenges, as these variables are difficult to define. As underrepresentation of minorities and women continues to persist in many facets of academia, it is important to assess the accuracy of differing methodologies. While asking individuals to self-identify their race and gender remains the gold standard of reporting, low response rates and response bias have been shown to affect results. In our initial study on representation in editorial boards, gender and race/ethnicity were determined based on publicly available data which can lead to misclassification of editors. We aimed to add to our study by asking editors to self-report their gender and race in hopes to validate our methodology given the importance of considering gender and race in academia. Methods: Of the 60 highest impact journals in oncology, hematology, radiation oncology, and surgical oncology identified, race/ethnicity and gender determinations were made using two methods. All senior editors were sent a survey via email asking participants to self-report their gender, race/ethnicity, age, and job characteristics. Gender and race were also assigned to the editors by a diverse coding team based on publicly available data and the NIH's OMB Directive 15 as a framework. The self-reported data was then compared to data that was assigned by our team. Results: 66 of the 793 (8.3%) editorial board members included in the study responded to the survey. Of the 66 respondents, gender was assigned correctly 100% (66/66) of the time and race was assigned correctly 95.5% (63/66) of the time. Of the 66 respondents to the self-survey of the 793 editorial board members surveyed. A significantly lower proportion of men responded to the survey compared to the gender breakdown of the 793 editorial board members (54.5% vs 72.6%; p = 0.000279). The three incorrectly identified respondents self-identified as Native Hawaiian, White, and Middle Eastern. Conclusions: Multiple recent reports have demonstrated high rates of sexual harassment, gender bias, and exclusion in the field of oncology. Collecting data on racial/ethnic groups and gender is imperative to understand the academic landscape of oncology and work towards a more equitable environment. Notably, this data from our study supports the methodology of a diverse coding team assigning gender and race based on publicly available data and the NIH's OMB Directive 15 as a framework as an alternative to self-report. Our study also demonstrates the low response rates and significant discrepancies in the demographic of respondents seen in survey-based identification.