An Approach to Identifying Rare Couple Combinations

Abstract

Research questions in the social sciences often do not focus on individuals, but rather on combinations of persons within a family. For rare combinations, such as same sex couples and families with twins, it often proves difficult to draw random samples, since conventional screening methods reach their limits when the target population is too small. In such cases, another method promises to be faster and more cost-effective: combinations of family members can be identified in the population register by linking single entries, which enables us to filter certain combinations on the basis of attributes of several family members. The article presents this approach, using a survey of binational couples as an example. In a first step, difficulties of sampling rare couple combinations are discussed. Secondly, the paper presents the alternative approach, in which data of several family members are combined within the population register. The accuracy of this approach is evaluated, using the Berlin Couple Survey, in which name-based techniques supplemented the selection process. The results show that most combinations of couples with an ethnic German partner and a foreign-born partner can be identified accurately, with false identification being mainly due to naturalization and Germanized names. Finally, further possible applications of this sampling approach are discussed.