OBELISS

Randomized controlled experiments are increasingly popular across the social sciences as the gold standard for identifying causal mechanisms underlying a wide range of social phenomena including collective action, social preferences, or market behavior. Because behavioral experiments are typically conducted with small groups sampled from student populations, a common critique is that they lack both scale and external validity. We propose to develop facilities for online experiments with participants recruited from panel surveys, more specifically the LISS panel, a long-running, high-quality and representative panel in the Netherlands. This would not only allow researchers to run online experiments with larger groups and representative samples, but crucially, also to capitalize on the data already available on panel respondents to enrich experimental designs with information on social-economic backgrounds, opinions, demographics, and more. In addition, all data collected In the LISS panel can be merged to CBS microdata. Another promising opportunity is to let panel respondents participate repeatedly in online experiments, in parallel to measurements of the panel survey.

Integrating facilities for experiments in the LISS panel would provide a unique infrastructure for ground-breaking social science research. While tools for online experiments are widely available (e.g., oTree), integrating such tools with the data infrastructure of LISS to allow a smooth real-time interaction between online experimental interfaces and the data already available from LISS is less straightforward. Concretely, the expected output of this project would be a well-documented public code base to be used by researchers and panel engineers for developing controlled online panel experiments.