On Thursday 19 April Edgar Erdfelder visits from the University of Mannheim about:
Statistical Power: A key concept in the emergence and surmounting of the replication crisis
There is little doubt that psychology currently faces a replication crisis. Although some psychological disciplines appear to be more involved than others, the problem is a general one, more or less typical for all branches of the social, behavioral, and biomedical sciences that evaluate their research hypotheses statistically. I argue that the replication crisis has two main causes, namely, (1) low power of statistical tests employed in typical studies and (2) p-hacking, that is, hunting for significant outcomes using multiple tests of significance on the same data. Both mechanism combined imply that false positive error rates may easily exceed 50% of the published significant research outcomes. Reasons and traditions fostering low statistical power and p-hacking will be discussed. Based on this discussion, I propose methodological rules and quality criteria for replication studies designed to help surmounting the replication crisis in the future.
Be the first to leave a comment. Don’t be shy.