Abstract
Prevalence studies with wearable devices are used to understand disparities in health-related physical activity behaviors and whether interventions are efficacious. However, studies have been limited to a binary definition of sex. This example analysis aimed to demonstrate how researchers can investigate differences in data beyond the sex-gender binary. Using a cross-sectional analysis of the All of Us Research Program dataset, participants' self-identified gender was categorized into Cisgender Female (n = 10,401), Additional Options (n = 27), Non-binary (n = 84), Transgender (n = 17), and Cisgender Male (n = 4,470). Fitbit data on active calories, steps, sedentary minutes, and very active minutes were analyzed following a valid statistical decision framework found in the companion editorial to this paper. Data were checked for normality using the Shapiro-Wilk test, and because data were not normally distributed, homogeneity was evaluated using the Brown-Forsyth test. The omnibus test for significant group differences was determined using the Kruskal-Wallis test, with significance accepted at p < 0.05. Effect sizes (ES) for omnibus test results were calculated using Epsilon squared. Results provide evidence for differences in physical activity metrics among gender groups (p < 0.001; active calories ES = 0.069, steps ES = 0.005, and very active minutes ES = 0.026). Cisgender males had higher active calories, steps, and very active minutes than cisgender females (40% more) and non-binary individuals (45% more). No differences were observed among other gender groups studied. These findings highlight that activity patterns vary beyond traditional binary classifications, emphasizing the need for gender-inclusive research in sport and exercise science. Specifically, the disparities observed underscore the importance of nuanced interpretations and tailored recommendations for diverse populations, addressing systemic gaps in supporting gender-diverse individuals in health and exercise behaviors.