Repeated cross-sectional surveys, such as general social surveys, aim to maintain consistency across each survey to ensure valid comparisons. Traditionally conducted face-to-face, these surveys face declining response rates and rising costs, increasing interest in self-administered mixed-mode designs. This study investigates the effects of switching from face-to-face to self-administered mixed-mode designs on selection bias and measurement bias using data from the 2023 German General Social Survey (ALLBUS). Respondents were randomly assigned to a face-to-face, concurrent, or sequential self-administered mixed-mode design. Selection bias was assessed through response propensities and sample composition, while measurement bias was examined at the variable level, focusing on question characteristics (showcard use, sensitivity, and question type). The sequential design more closely resembled the face-to-face benchmark in terms of selection bias. Measurement bias was lower for factual than attitudinal questions, and showcard use in face-to-face mode reduced measurement bias, suggesting that visual stimuli in that mode improve cross-mode consistency.
Alexandra Asimov (Tue,) studied this question.