Targeting the Replication Crisis and Improving the Credibility of Research Findings in Clinical Psychology. A Commentary on Pittelkow et al

• Published in: Clinical psychology (New York, N.Y.) N.Y.), 2021-06, Vol.28 (2), p.226-228
• Main Author: Hengartner, Michael P.
• Format: Article
• Language: English
• Subjects: Bayesian Analysis; Clinical Psychology; Credibility; Experimental Replication; Experimentation; Human; Intervention; Mental Disorders
• ISSN: 0969-5893; 1468-2850
• DOI: 10.1037/cps0000028

Comments on the original article by Pittelkow et al. (see record 2021-72420-014) regarding replication target selection in clinical psychology. The failure to replicate research findings in psychology is a major scientific issue with important practical implications (Pashler & Harris, 2012), especially with respect to psychological interventions in people with serious mental health problems (Hengartner, 2018). The study by Pittelkow et al. (2021) in this issue is thus timely and commendable. They offer a standardized approach based on Bayes factors (BFs) and subsequent qualitative evaluation to determine which intervention studies in clinical psychology have uncertain or questionable evidential support and are thus in need of replication. Overall, for 42.6% of the statistically significant effects studied, the BFs indicated weak or no evidence for the claimed effect. As I detail below, this is likely an underestimate of the weakness of the evidence, since BFs do not take into account systematic research biases that inflate effect estimates. Nevertheless, this alarming finding corresponds reasonably well with the poor replicability of many research findings as demonstrated in previous studies. In the following, I will discuss some of these studies and offer suggestions as to how we could further improve the credibility of research findings. (PsycInfo Database Record (c) 2021 APA, all rights reserved)