Methods to address selection effects in the meta regression and transfer of ecosystem values

The analysis develops an approach for dealing with selection effects in the meta regression of ecological values. The approach is based on Heckman's [Heckman, James. 1979. Sample Selection Bias as a Specification Error. Econometrica 47 (1):153-161.] two stage procedure and is adaptable to cross section and unbalanced panel data. The approach identifies both a method of testing for selection effects and for consistent estimation if selection effects are shown to be statistically significant. The approach is illustrated with a meta regression of wetland ecosystem values. The application shows that selection is statistically and economically significant. Selection effects lead to baseline wetland values that are almost 4 times larger than values computed using the selection corrected parameters. Value adjustment factors for wetland services and methodological variables appear less prone to selection effects. The uncorrected value adjustment factors for wetland services and research methods are, on average, within 15% of the selection corrected value adjustment factors. (c) 2006 Elsevier B.V. All rights reserved.