-----

The replication crisis of 2010s refers to a field-wise realization that common methods for producing and procuring knowledge in behavioral science is not as robust as one might hope. It was precipitated by a rise of seemingly outlandish claims, difficulties in replication, and cases of outright fraud. This crisis has led to several reforms that are intended to improve practice in the field. To analyze these reforms, Rouder applies the lens of Gerd Gigerenzer who worries that statistical practice is more about rituals than statistical thinking. These reforms, though well intended, have created new rituals that do little to promote good inference and interpretation. Rouder illustrates these points with studies that use the humble correlation coefficient to draw inferences on individual differences in behavior. As may not be obvious, the act of estimating and interpreting a correlation takes skill and insight. He argues that the most immediate key to good inference is understanding uncertainty in deep and subtle ways. Researchers should not mistake adopting replication-crisis reforms for judicious inference. Good analysis relies on researchers exercising thoughtfulness, creativity, restraint, and discipline in promoting a nuanced understanding of variability and structure in data.

-----