Gender-specific starting point bias in choice experiments: Evidence from an empirical study

In this paper, we test whether preferences and willingness-to-pay estimates obtained in a Choice Experiment study are susceptible to starting point bias as is often the case in Dichotomous Choice Contingent Valuation studies. On the basis of a multinomial probit model, we find that preferences are indeed susceptible to starting point bias. In a split-sample design our results suggest that varying the price levels displayed in a so-called Instructional Choice Set presented prior to the actual preference eliciting choice sets, significantly impacts respondents' preferences and willingness-to-pay for protecting Danish nature areas from new motorway development. In particular, our results show that the bias is gender-specific. Only female respondents are significantly affected. Results further reveal that the impact of the starting point bias decays as respondents evaluate more and more choice sets. This supports the Discovered Preference Hypothesis, and on the basis of this we suggest a number of ways to potentially mitigate the starting point bias.