The development of more realistic choice experiments has taken on board a number of suggestions in the broader hypothetical bias literature. One issue, in particular, is the increasing interest in finding ways to bridge the gap between the stated choice response and real choosing, as a way of increasing the confidence with which an individual would hypothetically purchase or use an alternative that is actually chosen in the choice experiment. In this paper we investigate the relationship between the respondent's response to a certainty question, defined on a 1-10 scale of surety, and features of the choice experiment that may have a bearing on the degree of confidence that can be placed on the stated choice, controlling for exogenous effects such as socioeconomic characteristics and attitudes to vehicle emissions. The focus on response certainty in this paper is as an external validity test. We find, using a generalised ordered logit model, compelling evidence that the number of acceptable alternatives and hence associated levels of attributes, together with the contrast of attribute levels of each designed alternative relative to an experienced status quo (or reference) alternative, play an important role in establishing certainty of response in a real market. The evidence should be taken on board in the future design of more realistic choice experiments.
response certainty ordered choice random thresholds automobile choice stated choice experiment internet survey context influences reference alternative