Approaches for Handling Missing Sexual Identity Data

• Activity Number: 471
• Type: Contributed
• Date/Time: Wednesday, August 3, 2011 : 8:30 AM to 10:20 AM
• Sponsor: Section on Survey Research Methods
• Author(s): HarmoniJoie Noel*+ and Taylor Lewis and Aaron Maitland and Heather Ridolfo
• Companies: Centers for Disease Control and Prevention and Centers for Disease Control and Prevention and Centers for Disease Control and Prevention and Centers for Disease Control and Prevention
• Keywords: missing data ; imputation ; sexual identity

Abstract

Health disparities research in the Lesbian, Gay, Bisexual, and Transgender (LGBT) community has received national attention as part of the new goals for Healthy People 2020. Using data from the National Survey of Family Growth Cycle 6 (2002) and the 2002-2008 National Health and Nutrition Examination Survey we found that less educated and Hispanic respondents were systematically missing data on sexual identity measures. Cognitive testing results showed that less educated and Hispanic respondents had comprehension issues with some of the sexual identity terms which could help explain their higher rates of missing data on the sexual identity measures. Systematic missingness on sexual identity measures could lead to biased univariate estimates and regression analyses of the relationship between sexual identity and health outcomes such as access to health insurance or disease prevalence. We will compare alternative approaches for handling missing data including casewise deletion, single and multiple imputation, and maximum likelihood estimation to evaluate the sensitivity of univariate and multivariate estimates to these different approaches.