![IconGems-Print](images/IconGems-Print.png)
Does Sexual Orientation and Gender Identity (SOGI) Question-Wording Influence Responses? An Experimental Test in a Non-Traditional Sample
James Dayton
ICF
Yangyang Deng
ICF
Lee Harding
ICF
Ronaldo Iachan
ICF
Matt Jans
ICF
Deirdre Middleton
ICF
Increasing visibility of sexual and gender minorities has led to heightened demand for accurate population statistics. Large-scale surveys increasingly measure sexual orientation and gender identity (SOGI). A measurement challenge is that questions must be understood by the non-LGBT population and meaningful to and respectful of the LGBT community. Poorly-worded questions incur measurement and nonresponse error risks. Motivated by varying prevalence estimates, this study assesses the impact of SOGI question wording on self-identification rates and item nonresponse. Respondents from two nonprobability surveys (MFour's Surveys on the Go® and Amazon Mechanical Turk [MTurk]), were randomly assigned to one of two sexual orientation and one of two gender identity measures used by three highly-influential health surveys: the California Health Interview Survey, Behavioral Risk Factor Surveillance System, and National Health Interview Survey. Results of the experiment are compared to each other and estimates from high-profile surveys. In addition to the experimental question wording results, this study demonstrates the potential for collecting SOGI data from nonprobability samples.