PUB - Publikationen an der Universität Bielefeld

In social sciences, longitudinal studies represent a common form of study design. However, this form of data collection may face the special problem of panel attrition over subsequent waves – a circumstance that can lead to a biased sample. As a consequence, longitudinal analyses may become flawed if the bias is unknown or not corrected. Previous research on panel attrition revealed that dropout can be related to several specific characteristics, such as sex, personality, or relationship status of the participants. As twins are a core feature of twin-studies – twin similarity a key tool in behavioral genetics – relational characteristics of this special siblinghood may play a crucial role in re-participating, too. Therefore, the aim of the present study was to investigate panel attrition with respect to certain personality traits (Big Five) and relational characteristics (sibling relationship and satisfaction with sibling relationship) to estimate potential biases as a basis for corrections in future studies. Analyses were not only performed on an individual level but also regarding twin pair resemblance as this is the main source of information in behavior genetic twin studies. The sample consisted of 4,097 twin pairs and 8,281 of their family members from wave one of the German TwinLife study. 60.9 % of the initial sample re-participated in wave 2. Results indicate that there are no consistent patterns in personality or relational characteristics across family members, age cohorts of the twins, and sexes regarding panel attrition. There also seem to be no consistent patterns regarding differences in twin similarity across re- and non-re-participators. Subsequently, selective panel dropout with respect to personal characteristics seems to be of minor importance in the TwinLife study and corrections may only be necessary in selective cases.