To the Editor In her editorial1 published in JAMA Psychiatry, Dr Kraemer gives important insights into using covariates, thereby adding to her large body of highly valuable publications.
Notably, she also stated, “Tests that are not statistically significant should be regarded as indicative of poorly justified, designed, or executed hypothesis-testing studies….”1 This statement may trigger undesirable recommendations to scientists, reviewers, and editors. Most importantly, those deciding on dissemination of study results may be reluctant to report a statistically nonsignificant test if it is perceived as indicative of poorly justified, designed, or executed (PJDE) methodology, which would impair scientific dissemination and the validity of meta-analytic findings. Therefore, and in line with recent recommendations to reduce dissemination bias,2 we would like to reflect on this statement.
Gunther Meinlschmidt, Jan K. Woike, Marion Tegethoff. Disseminating Justified, Well-Designed, and Well-Executed Studies Despite Nonsignificant Tests. JAMA Psychiatry. 2016;73(1):88–89. doi:10.1001/jamapsychiatry.2015.2259