Commentary by Dijksterhuis, author of original study on behaviour priming that failed to replicate in most recent @RegReports, falls into usual post hoc explanations and lack of falsifiability. 1/n psychologicalscience.org/redesign/wp-co… #replicability #opensciecne
All degrees of freedom that are required to make the effect significant post hoc were foreseeable a priori. Eg exclude those who reporting being suspicious of the study's intent. If this data was collected + the original theory state it's an unconscious effect, 2/n
then why wasn't this employed as an exclusion criterion? Author reports not expecting such a high rate of suspicion, but it if was a priori relevant the rate wouldn't matter or could have had a design contingency. 3/n
Either the original theory always stated the effect was necessarily unconscious or this idea is novel and exploratory, but it can't be both. How many more decades of this work before we stop calling studies exploratory and allow for #falsifiability? 4/n
"we may ask ourselves whether investing in a large
scale replication project such as the current one is the best and most economical
way to investigate older findings" - say this BEFORE the replication was done or not at all. I doubt this Q would arise if results were +ve. 5/n
scale replication project such as the current one is the best and most economical
way to investigate older findings" - say this BEFORE the replication was done or not at all. I doubt this Q would arise if results were +ve. 5/n
