The debate over open access is missing the point

Being able to replicate scientific findings is crucial for scientific progress1-15. We replicate 21 systematically selected experimental studies in the social sciences published in Nature and Science between 2010 and 201516-36. The replications follow analysis plans reviewed by the original authors and pre-registered prior to the replications. The replications are high powered, with sample sizes on average about five times higher than in the original studies. We find a significant effect in the same direction as the original study for 13 (62%) studies, and the effect size of the replications is on average about 50% of the original effect size. Replicability varies between 12 (57%) and 14 (67%) studies for complementary replicability indicators. Consistent with these results, the estimated true-positive rate is 67% in a Bayesian analysis. The relative effect size of true positives is estimated to be 71%, suggesting that both false positives and inflated effect sizes of true positives contribute to imperfect reproducibility. Furthermore, we find that peer beliefs of replicability are strongly related to replicability, suggesting that the research community could predict which results would replicate and that failures to replicate were not the result of chance alone.

To analyse data from a trial and report the frequencies with which major and minor errors are detected at a general medical journal, the types of errors missed and the impact of training on error detection. 607 peer reviewers at the BMJ were randomized to two intervention groups receiving different types of training (face-to-face training or a self-taught package) and a control group. Each reviewer was sent the same three test papers over the study period, each of which had nine major and five minor methodological errors inserted. BMJ peer reviewers. The quality of review, assessed using a validated instrument, and the number and type of errors detected before and after training. The number of major errors detected varied over the three papers. The interventions had small effects. At baseline (Paper 1) reviewers found an average of 2.58 of the nine major errors, with no notable difference between the groups. The mean number of errors reported was similar for the second and third papers, 2.71 and 3.0, respectively. Biased randomization was the error detected most frequently in all three papers, with over 60% of reviewers rejecting the papers identifying this error. Reviewers who did not reject the papers found fewer errors and the proportion finding biased randomization was less than 40% for each paper. Editors should not assume that reviewers will detect most major errors, particularly those concerned with the context of study. Short training packages have only a slight impact on improving error detection.