Contrasting conventional with multi-level modeling approaches to meta-analysis: Expectation consistency in UK woodland recreation values

Abstract

The paper presents a variety of meta analysis models of woodland recreation benefit estimates, contrasting conventionally estimated models with those provided by novel, multi-level modeling (MLM) techniques (Goldstein, 1995). Our conventional models suggest that studies carried out by certain authors are associated with unusually large residuals within our meta-analysis. However, the MLM approach explicitly incorporates the hierarchical nature of meta-analysis data, with estimates nested within study sites and authors. Allowing for this reveals that these residuals are not a significant determinant upon values, suggesting that, at least in this aspect, estimates may be more robust than indicated by less sophisticated models. However, previously noted differences in benefit estimates between alternate valuation methods persist across our various analyses and remain a cause for concern.