Panel Conditioning in Difficult Attitudinal Questions

We investigate whether survey answers of trained respondents differ systematically from answers of novice respondents. Focusing on difficult attitudinal or preference questions, we find that novice respondents answer “don’t know” significantly more often. Importantly, however, there is no systematic evidence for a conditioning effect on measured prefer-ences (substantial answers). Overall, our results suggest that panel conditioning does not pose a serious problem, even in the case of rather difficult questions.