Abstract

Faked Interviews can have a substantial impact on survey data quality and can lead to population estimates that are seriously biased. Especially in the context of policy research unidentified falsification can be problematic since the falsified data might produce results that lead to a misallocation of public funds. In order to prevent or rather identify such situations, this projects aims to develop methods for the identification of falsified interviews. Different statistical identification approaches will be evaluated in detail. In addition, new approaches and strategies will be developed and tested. These strategies will be used to identify suspicious patterns in the survey data via statistical methods and algorithms. Various indicators of falsification proposed in the literature as well as new ones will be considered. In the long run, this project aims to 1) provide general methods that can be extended to all IAB studies with interviewer participation, and 2) derive a comprehensive, efficient and cost-effective quality control strategy for IAB surveys to ensure high-quality data collections are being undertaken.