Is the distribution of resolvable uncertainty Type I extreme value? A test for random coefficient models using choice probabilities

Probabilistic stated choices allow respondents to express uncertainty about their intended decisions. They are typically paired with random-coefficient models (RCM) that impose Type I extreme value (EV1) resolvable uncertainty. This paper proposes a test of these assumptions based on the restrictions imposed by the EV1 distribution on the population distribution of interquantile ranges of ex ante returns. Our test is rejected in four studies published in leading economics journals. In these applications, EV1-based estimators often perform reasonably well for median willingness-to-pay (WTP), but are less reliable for distributional objects such as the full WTP distribution.