What can we learn from benefit transfer errors? Evidence from 20 years of research on convergent validity

Abstract

We develop a nonparametric approach to meta-analysis and use it to identify modeling decisions that affect benefit transfer errors. The meta-data describe the results from 31 empirical studies testing the convergent validity of benefit transfers. They evaluated numerous methodological procedures, collectively reporting 1071 transfer errors. Our meta-regressions identify several important findings, including: (1) the median absolute error is 39%; (2) function transfers outperform value transfers; (3) transfers describing environmental quantity generate lower transfer errors than transfers describing quality changes; (4) geographic site similarity is important for value transfers; (5) contingent valuation generates lower transfer errors than other valuation methods; and (6) combining data from multiple studies tends to reduce transfer errors. © 2013 Elsevier Inc.