Representation and Samples

In the field of sampling, we explore strategies to achieve best possible representation of the target population. In this context, data quality refers to the accurate representation of the target population by respondents. Our focus lies on sampling methods and the conditions of data collection or fieldwork conducive to establishing representative data. Additionally, we conduct research on methods to address nonresponse and draw reliable conclusions from surveys despite missing data.

We currently focus on the following areas:

• Nonresponse bias: We delve into identifying personal traits or survey attributes that impact on participation and explore methods to enhance survey participation and to mitigate potential biases.
• Weighting: We develop methods to counterbalance biases arising from high non-response rates within specific demographic groups. We also investigate the computation of design weights for samples that combine multiple sampling frames.
• Imputation: We examine approaches for replacing missing values in survey data as effectively as possible. Our emphasis lies in imputation methods tailored to modular questionnaire designs, where certain items are intentionally presented to subsets of respondents, thereby generating planned missing values. We also conduct research on variance estimation under imputation.
• Respondent recruitment: We explore different approaches to recruiting survey participants and their impact on data quality. This includes innovative methods such as harnessing the potential of social networking sites (e.g., Facebook or Instagram) for recruitment purposes, e.g., for hard-to-reach groups such as refugees.