Estimate the number of same-sex couples in the census: experimentation of a sex validation solution by the first name

In France, same-sex couples are overestimated when using Census data. This is due to a large amount of opposite-sex couples being wrongly considered as same-sex couples following an error in the reported sex of one of the partners. In order to identify those couples who likely are same-sex couples compared to those who are most likely opposite-sex couples who mismarked the sex item for one of the partners, we propose a solution using first names. An index associates to a particular first name the proportion of reported females among the holders of that name. It thus indicates whether an error in the reported sex is likely or not. We intend to demonstrate that this solution, when implemented in the French Census, should permit to produce high-quality estmates and new studies on same-sex couples.