Formatting the two-step gender measure: Experimental insights from the United States

Background: Measuring sex and gender is central to demographic research and studies of social inequality. Best-practice guidelines for surveying English-speaking adults recommend a two-step approach, which uses two items to distinguish sex at birth from current identity. Additional research is needed to understand whether specific formatting decisions affect response distributions or measures of data quality. Objective: We explore three understudied formatting choices: (1) Should sex or gender terms be used for current identity responses, (2) should both items appear on the same page or different pages in online surveys, and (3) in which order should the response options be presented? By answering these questions, we provide practical guidance for researchers applying the two-step measure, and evidence on how robust the measure is to various formatting differences. Methods: We draw on three survey experiments conducted on two US adult samples: a sexual and gender minority (SGM) sample (n = 2,402) and a general population sample (n = 1,491). Conclusions: Our results suggest researchers using the two-step approach can improve data quality by (1) using gender terms for current identity, and (2) using a page break in online surveys. Varying the response option order does not have measurable effects on data quality in our samples, which frees researchers in most survey contexts to consider nontraditional response orders that have other advantages. Contribution: Some formatting changes can improve data quality, especially for SGM respondents, but small differences in how the two-step measure is formatted do not compromise comparability across studies or overall performance in general population surveys.