Identifying Panel Conditioning in Self-Reported Satisfaction Data

Data derived from survey questions that ask respondents about their overall life satisfaction, are increasingly being used in (multidimensional) analyses of trends and fluctuations in a country’s prosperity. The more mainstream such data become, the more researchers are taking seriously potential measurement issues. This paper investigates the existence of a phenomenon called panel conditioning or panel effects, that is, the fact that, ceteris paribus, people might answer questions differently the longer they are in the panel. First, I will work through some identification issues that arise when trying to identify panel effects in panel data and that might lead to large biases when being ignored. Second, I will apply two main different strategies to identify panel effects in life satisfaction data. The analysis concludes that ceteris paribus, people rate their life satisfaction lower, the longer they have been in the panel.