The main goal of our research is to contribute to the reliability of online survey data by examining the possibilities of classical demographic weighting by including new, thematically relevant attitude variables. The direct predecessor of the project is the research entitled Election forecast from survey data, where we developed an approach that took respondent availability and political interest into account during weighting. Our current project further tests this approach in other, more sensitive topic areas.

The study focuses on four socially relevant areas – alcohol consumption, fast food consumption, tooth brushing habits, and gasoline consumption – where respondents’ biased self-reports regularly deviate from real-life patterns. In collaboration with Panelstory, we are working on methodological approaches that examine whether including attitudes related to the topic in the weighting can improve the accuracy of estimates.

The primary purpose of the openly accessible weighting procedures and program codes created during the project is to allow them to be tested and evaluated in a market research environment. In the longer term, these results may contribute to the more well-founded application of online survey methodology in areas where respondent bias is particularly present – such as healthcare, sustainability or consumer behavior research.