Scientific publications typically report positive rather than negative empirical results [2], [3], suggesting that empirical results in the literature do not form a comprehensive or representative sample of all empirical results. It is widely agreed that publication bias, or the selective reporting of empirical studies on the basis of their statistical significance, may inflate effect sizes estimated by meta-analyses.[4]–[13]. A widely held view is that the optimal solution to counter publication bias is to collect all of the available evidence regardless of whether the results are significant or not. This view is contradicted in a recent paper by de Winter and Happee [1] (henceforth W&H) in which they present the results of a simulation study that was based on the premise that science is self-correcting. W&H concluded on the basis of their simulations that it may be more effective to publish results selectively on the basis of outcomes. We agree with W&H that scientific publication is associated with costs in terms of writing and publishing [14], which should be diminished. However, we disagree with their notion that too many published
