Handling respondent uncertainty in choice experiments: Evaluating recoding approaches against explicit modelling of uncertainty

Using data from two Choice Experiment environmental valuation surveys we investigate several different ways of handling respondent uncertainty. In both surveys respondents are asked to state their certainty of choice after each single choice set. We evaluate three different recoding-of-answers methods adapted from the Contingent Valuation literature. Furthermore, we evaluate two models which directly capture the effect of respondent uncertainty by parameterization of a scale function. In one model the scale parameter is a function of respondents' stated uncertainty level. In the other it is a function of respondent and choice set characteristics found to be significant determinants of stated uncertainty. All approaches are compared to a benchmark utility model specified as a random parameter error component logit model. While some of the recoding approaches apparently reduce noise in the data, they generally have no significant effect on attribute Willingness-To-Pay estimates, and standard errors on WTP estimates tend to increase. The explicit modelling of the scale parameter using stated uncertainty reveals that unobserved variation decreases as certainty of choice increases. While the model performance does not improve much, this approach offers a structurally and intuitively much more appealing way of accounting for uncertainty in choices in CE surveys.