Going beyond statistical rituals: Improving the understanding of statistics using blog posts and workshops
Authors/Creators
- 1. Department of Child and Adolescent Psychiatry, Psychosomatics and Psychotherapy, University Hospital, LMU Munich
Description
In recent years, it has become increasingly clear that many empirical findings in psychological science are not replicable (Open Science Collaboration, 2015). This is problematic, because this suggest that psychological theories, which are often used for high-stake decision making, may be based on effects that, in reality, do not exist. The proximal causes of the replication crisis have been identified as publication bias (selective publication of significant results), questionable research practices (tweaking the data and analyses until one obtains a significant result), low power (too few participants to yield a high probability of obtaining an effect of interest) and hypothesising after results are known (HARKing; writing a rationale for a study after the data has already been analysed) (Bishop, 2019). To a large extent, these issues can be traced back to a poor understanding of statistics.