Getting it right with meta-analysis: Assessing heterogeneity and moderator effects in the presence of publication bias and p-hacking
It is well-known that publication bias and p-hacking, which are omnipresent in psychology and other sciences, lead to overestimated effect sizes in meta-analyses. Unknown is how and when they affect heterogeneity and moderator effects in meta-analysis. Because these outcomes of meta-analyses are used for making decisions, both in science and practice, methods are needed that can deal with publication bias and p-hacking. The proposed research (i) assesses consequences of publication bias and p-hacking for heterogeneity and moderator effects, and (ii) improves meta-analytic methods enabling researchers to obtain accurate estimates of overall effect size, heterogeneity, and moderator effects in their applications.
Prof. M.A.L.M. van Assen, Dr J.M. Wicherts & Prof. K. Sijtsma
1 September 2015 – 1 September 2019