Transgender-inclusive measures of sex/gender for population surveys: Mixed-methods evaluation and recommendations

Given that an estimated 0.6% of the U.S. population is transgender (trans) and that large health disparities for this population have been documented, government and research organizations are increasingly expanding measures of sex/gender to be trans inclusive. Options suggested for trans community surveys, such as expansive check-all-that-apply gender identity lists and write-in options that offer maximum flexibility, are generally not appropriate for broad population surveys. These require limited questions and a small number of categories for analysis. Limited evaluation has been undertaken of trans-inclusive population survey measures for sex/gender, including those currently in use. Using an internet survey and follow-up of 311 participants, and cognitive interviews from a maximum-diversity sub-sample (n = 79), we conducted a mixed-methods evaluation of two existing measures: a two-step question developed in the United States and a multidimensional measure developed in Canada. We found very low levels of item missingness, and no indicators of confusion on the part of cisgender (non-trans) participants for both measures. However, a majority of interview participants indicated problems with each question item set. Agreement between the two measures in assessment of gender identity was very high (K = 0.9081), but gender identity was a poor proxy for other dimensions of sex or gender among trans participants. Issues to inform measure development or adaptation that emerged from analysis included dimensions of sex/gender measured, whether non-binary identities were trans, Indigenous and cultural identities, proxy reporting, temporality concerns, and the inability of a single item to provide a valid measure of sex/gender. Based on this evaluation, we recommend that population surveys meant for multi-purpose analysis consider a new Multidimensional Sex/Gender Measure for testing that includes three simple items (one asked only of a small sub-group) to assess gender identity and lived gender, with optional additions. We provide considerations for adaptation of this measure to different contexts.

据估算，美国跨性别（trans）人群占总人口的0.6%，且该群体存在的显著健康差异已被文献记录，因此政府与科研机构正逐步完善性别/社会性别的测量体系，纳入跨性别包容性维度。目前针对跨性别群体调查提出的各类方案，比如宽泛的多选式性别认同列表、提供最大灵活性的手写输入选项，通常并不适用于大规模人口普查。此类普查需要精简的问题与少量分类以支撑分析工作。当前针对跨性别包容性人口调查的性别/社会性别的测量方案（包括现有正在使用的工具）的评估研究仍较为匮乏。本研究通过对311名受访者开展网络调查与后续追踪，并从最大多样性子样本（n=79）中开展认知访谈，对两项现有测量工具开展混合方法评估：其一为美国开发的两步式问题，其二为加拿大开发的多维度性/性别测量工具（Multidimensional Sex/Gender Measure）。研究发现，两项工具的项目缺失率极低，顺性别（cisgender，非跨性别）受访者未表现出任何理解困惑。然而，多数访谈受访者表示，两类问题集均存在使用障碍。两项工具在性别认同评估上的一致性极高（K=0.9081），但对于跨性别受访者而言，性别认同无法很好地替代其他维度的性别/社会性别测量。从分析中提炼出的可用于测量工具开发或适配的议题包括：性别/社会性别的测量维度、非二元性别身份是否归属跨性别范畴、原住民与文化身份、代理报告、时间性考量，以及单一项目无法有效测量性别/社会性别的局限。基于本评估结果，我们建议面向多用途分析的人口普查可考虑采用新的多维度性/性别测量工具进行测试，该工具包含三个简单问题（其中一个仅针对少数子群体提问），用于评估性别认同与生活性别，同时支持可选附加项。此外，我们还针对该工具在不同场景下的适配提出了具体考量要点。