| Summary: | Byers‐Heinlein et al. (2021) argue that a greater focus on reliability in infant research is much needed and desirable for the field. Moreover, they offer a number of possible solutions to increase reliability in infant research. Here, I do not discuss the suitability of the offered solutions. Instead, I further develop the authors' argument by considering the worryingly downstream consequences of low reliability for the very possibility of drawing meaningful inferences from the results of replication studies. I argue that the sources of error related to sampling and measurement greatly affect the estimates, resulting in confidence intervals being often very wide. Consequently, interpreting the differences between small‐scale original and replication studies becomes a meaningless task, which, I maintain, could be replaced by a meta‐analytic mindset and the habit of fully incorporating uncertainty in the reasoning about research findings.
|
|---|