Daniel T. Gilbert, Gary King, Stephen Pettigrew, Timothy D. Wilson, Comment on ‘Estimating the Reproducibility of Psychological Science’, Science, 351, Issue 6277, 4th March 2016, DOI: 10.1126/science.aad7243.
Another quick post, prompted by seeing a link to a New York Times article, which led to this piece in Science. It’s the latest turn in what has been called the ‘Replication Crisis’ in Psychology. Over the last few years there have been concerted attempts to repeat some prominent experiments. In principle it should be possible to replicate these and achieve comparable results, but in a striking number of cases the experimenters have found themselves unable to do so. Hence the crisis and the media attention: such failures threatened the integrity of the science and perhaps the scientists involved. But this piece seems important: it suggests that in various ways the rigour of the attempted replications may have been less secure than that of the initial experiments, and that some debatable choices about statistics were made. There will be more twists and turns in this story. On the one hand, replication is very difficult, it’s not often rewarding work, and it needs to be defended; on the other hand, it’s got to be done right, and on the whole I’m relieved that the discipline may not be as undermined as it looked for a while.