When I was a graduate student, I fell in love with choice models. After years studying the econometrics of the standard linear model, discrete choice offered so many new, cool twists. With contingent valuation (CV) studies abounding, data was plentiful and varied. Every CV dataset had its own kinks and quirks that begged to be addressed through innovative modeling techniques. Dissertation topics were not scarce. We economists like to assume. There are jokes written about this. My assumption, as I slaved over the statistical properties of the double-bounded CV model, was that CV data was good...