| Sumario: | The writers examine statistical approaches to dealing with the “fat tail” problem that arises in the dichotomous choice (DC) question format for contingent valuation. A disadvantage of the DC format relative to open-ended and iterative approaches is that the DC format does not estimate respondents' maximum willingness to pay (WTP). Instead, it only identifies whether the respondent's maximum WTP is larger or smaller than a specified bid amount. The fat tail problem can arise if the estimated distribution of maximum WTP attaches an unrealistically large proportion of the cumulative density function to very high WTP amounts. Truncating the distribution and the K.J. Boyle, M.C. Welsh, and R.C. Bishop (1988) normalization procedure seek to overcome this, but both approaches tend to depress the estimates of WTP. A new approach is the pinched logit model, which generates a “best guess” estimate of mean WTP. However, as this estimate involves extrapolation beyond the range of the data, it does tend to be highly variable.
|
|---|