Current two-step measures of gender identity do not prescribe methods for using expanded responses (e.g., multiple selections) among sexual and gender minority (SGM) people, though they want the opportunity to provide these responses. To increase statistical power using expanded gender identity responses, we created an automated algorithm to generate analyzable categories. Participants’ expanded gender identity responses and sex assigned at birth were used to create five categories (i.e., cisgender men, cisgender women, gender expansive individuals, transgender men, and transgender women) from a cohort of SGM people (N = 6,312, 53% cisgender individuals). Data was collected from June 2020 to June 2021. Chi-square tests were performed to assess the association between the algorithm-generated and participant-selected gender categories, and to identify demographic differences between participants in the algorithm-generated categories. Forty-six percent of our sample may have been classified into an “other” category without an algorithm due to writing their own response (5.7%), selecting “another gender identity” (5.7%), or selecting multiple (42.6%) or less commonly described (10.2%) gender identities. There was a relationship between the categories formed by our algorithm and participants’ single category selection (χ2 [20] = 19,000, p < .001). Concordance rates were high (97-99%) among all groups except for participants classified as gender expansive (74.3%). Without an algorithm to incorporate expanded gender identity responses, almost half of the sample may have been classified into an “other” category or dropped from analyses. Our algorithm successfully classified participants into analyzable categories from expanded gender responses.
