As the final data set consists of multiple (6) responses from each respondent, there is a possibility of correlated errors across the multiple valuation responses [15, 27]. However, existing methods that account for potential correlation [15, 27] are not applicable to the characteristics of the data.9 Moreover, Alberini et al. [2] find that joint estimation (as presented by Poe et al. [27]) often provides only “negligible” gains in efficiency, confirming similar findings of Poe et al. [27] and Alberini and Kanninen [3]. Given the lack of tractable options for addressing potential error correlation in the survey data and the lack of clear empirical evidence that addressing such correlation would provide significant gains in efficiency, we proceed using the econometric model outlined above, based on the assumption that respondents answered each question independently, as instructed. Maximum likelihood estimation produces asymptotically unbiased parameter estimates, even if errors are correlated. Moreover, existing evidence suggests that any correlation between error terms, if such a correlation exists, would take the form of positive correlation.10 As Poe et al. [27] illustrate, accounting for positive error correlation increases the significance of willingness to pay differences of the type studied here. Accordingly, if positive correlation exists, the present results would provide conservative estimates of the significance of mean WTP differences.