Meta-Analysis,Benefit Transfer,and Methodological Covariates: Implications for Transfer Error

Meta-regression models in the valuation literature demonstrate that willingness to pay estimates vary according to methodological factors. Neither theory nor characteristics of policy sites dictate the treatment of associated covariates within benefit transfer, however, and the literature provides few insights into potential impacts of common empirical treatments. This paper introduces a method to systematically characterize the impact of methodological variables on transfer error. Using a repeated leave-one-out convergent validity framework, the analysis contrasts errors for a hypothetical ideal case in which correct methodological covariate treatments are known to the realistic case in which the correct treatment is unknown. Results indicate that the common assumption of mean values for methodological covariates leads to only a modest increase in mean transfer error relative to that found in the hypothetical ideal case.