Abstract

As it has recently been shown by the identification of a large number of falsified survey interviews, there is a need to strengthen and improve the interviewer control for IAB surveys. The aim is to detect falsifiers more effectively and early, which seeks to improve the data quality permanently. One possible control strategy is the detailed documentation or collection of para data. Those para data document specific behavior of the interviewer during the interview such as mouse movements, changes between response categories or responding times. The advantage of this method is that on the one hand, the interviewer is not aware of this control, on the other hand, no consent of the respondent for this documentation is necessary. Using machine-learning algorithms or data mining methods, this data can be analyzed efficiently and inexpensively.