A paradox in random-effects meta-analysis

12/21/2018

∙

by   Jiandong Shi, et al.

∙

0

∙

share

This paper reports a new paradox in random-effects meta-analysis. To our knowledge, this paradox has never been reported in the literature on meta-analysis. With the newly discovered paradox, it has been putting us in a dilemma on what final conclusion can be made. We hence advocate meta-analysts to be extremely careful when interpreting the final results from the random-effects meta-analysis, in particular when the number of studies is small and the heterogeneity is large.