Improving the Quality of Volunteer Web Panels: Evaluating Propensity Score Adjustments Methods for the Leisure Panel

Tuesday, July 15, 2014: 10:00 AM
Room: 416
Distributed Paper
Vera TOEPOEL , Methods and Statistics, Ultrecht University, utrecht, Netherlands
With the rise of the Internet more and more data are collected via volunteer panels. These panels are not based on probability mechanisms and hence inferences are difficult to make. Post-survey adjustments techniques such as propensity score weighting are often used to improve data quality. However, which methods work best and which variables need to be taken into account  differ per survey. In an attempt to structure methods of post-survey weighting techniques for the volunteer Leisure Panel, different weighting methods are evaluated as well as different weighting variables, such as behavioral, attitudinal, lifestyle, and socio-demographic variables. Results are compared to outcomes of a probability-bases panel (LISS Panel) and figures from Statistics Netherlands.