Just how much published research is bunk?
There seems to be a lot more scrutiny being pointed toward bad research recently, and today’s link continues that theme. This time, an article in the Chronicle of Higher Education discusses an attempt to replicate every study over a year long period, in three major psychology journals, and see just how many of those results can actually be replicated. Initial indicators imply that the answer might turn out to be “not many at all”.
This, to me, is another one of those symptoms of the broad, underlying problems with the publish or perish tenure review process in academia. On one hand, you have the pressure to fabricate flashy and interesting results in order to obtain high impact publications and build your career – an issue I’ve touched on before in this blog. On the other hand, you have an environment that actively looks down upon “replication of effort”, so good quality reviews or repeat studies are marginalized, and even denied publication, because they aren’t “sexy” enough. Retractions get minor billing, checking others’ work is treated by tenure review boards as a waste of time, and the quality of scientific research steadily degrades.