Publication bias in sport and exercise psychology: The games we play

The purpose of this study was to determine if publication bias exists in the sport and exercise psychology literature. All relevant articles from the 1987, 1992, and 1997 issues of five sport and exercise psychology journals were included in the analysis. The articles were coded by two reviewers according to whether null hypothesis tests were conducted and, secondly, if these tests resulted in a rejection of the main stated null hypothesis. Of those articles including tests of significance, approximately 98% had at least one significant finding and approximately 80% rejected the main stated null hypothesis. Taken together, along with comparisons to a sample of unpublished dissertations and theses, these findings indicate that publication bias is alive and well in the sport and exercise psychology literature. The implication of such findings is that we are overestimating the effectiveness of our interventions and the degree to which variables covary.