Accurate Measurement of Same-Sex Couples: Empirical Implications

Rebecca DiBennardo, University of California, Los Angeles

The Census represents the richest data resource measuring the LGBT population, but scholars note serious problems with its same-sex couple calculation. In 2010, the Census acknowledged measurement problems and suggested that it likely misclassified 40 percent of same-sex couples in Census 2000 and 28 percent in Census 2010. As a result, the Bureau released state-level “preferred” estimates for the number of same-sex couples in those years. Using these data, we develop Census 2000 county-level counts of same-sex couples and replicate a 2009 study that included a Census 2000 same-sex couple measure. After replicating regression analyses, substituting our adjusted estimates for the original tabulations, and comparing differences, our results show that the adjusted same-sex couple measure increases the magnitude of the effect of same-sex couples on voting behavior. These results demonstrate how same-sex couple adjustments impact findings from previous scholarship, and suggests that researchers should use adjusted data moving forward.