Data linkage - respondents consent without selectivity?

Abstract

The lidA Study is designed as a longitudinal survey. The respondent's consent is mandatory for storing sample data. Moreover, the survey data shall be linked with social security data of the Federal Employment Agency and individual's health insurance claims data in case of the respondent's written consent. This essay pursues the issue of whether this methodologically challenging objective of obtaining 3 consents within one study could be met without any selectivity. The data basis is a cohort study with 2 cohorts of a representative sample of employed individuals subject to social security contributions. The sample was interviewed for the first time in 2011. The analysis dataset comprises 6 585 respondents. Selectivity analyses prove that the realisation of the first measurement's sample turned out to be representative as well as unbiased. As expected, more respondents stated their willingness to remain in the panel and also consented to linkage of social security data than those who consented to linkage of health insurance claims data. All 3 consents were given without resulting in any bias. Even linking all 3 consents does result in minimal effects of a few subgroups only. A significant number of respondents can be motivated to participate due to proper placement of the questions concerning consent and the provision of insight into the use of the data.