erroneous conclusions. The problem with the selective publishing scenario is that the researcher does not know whether the data were generated with that scenario or rather that the data reflect true heterogeneity. Fixed-effect meta-analysis is only appropriate if the researcher is certain that the heterogeneity does not reflect heterogeneity in population effect size. If the heterogeneity caused by selective publishing is erroneously interpreted as heterogeneity of the underlying population effect size, then this heterogeneity artifact may lead entire fields astray. Researchers may focus on random aspects of the published studies as potential moderators of the studied effect and interpret them in substantive terms while all they represent is random noise coupled with a selective approach to scientific publishing.
