Accounting for Attribute Non-Attendance in Three Previously-Published Choice Studies of Coastal and Marine Resources

We revisit three recently-published papers that apply discrete-choice experiment methods to coastal and marine ecosystem goods and services, in light of attribute non-attendance (AN-A). We find that accounting for AN-A does not always improve model fit, but when it does, the improvement can be substantial. Estimated price and attribute coefficients change, but these changes do not follow a consistent pattern, either in direction or magnitude. Mean attribute increment value (i.e., willingness to pay, WTP) estimates change, but also with no discernible pattern. However, in several cases, generally in those cases where accounting for AN-A improves model fit, we observe substantial improvements in the confidence intervals on WTP, i.e., accounting for AN-A appears to produce much more precise WTP estimates. In short, we find that accounting for AN-A is not always warranted, but when it is, the key payoff appears to be more precise WTP estimates