The Replication Crisis Could Be Worse Than We Thought

Citations aren't facebook likes, the academic world should stop treating them as such. Studies with bad methodology, dead ends and all shorts of flaws are often cited for a range of different reasons. Seemingly unreproduceable studies can be the result of bad measurements, but more interestingly could be because of previously unknown factors ie trace contaminants not accounted for.

Also, papers proven wrong ought to remain published. The idea behind retractions seems to be misinterpreted by some as an attempt to maintain journals as a library of ground truth. That was never the case and it never will be.

The real problem is the metrics we use for allocating funds, jobs and status.

Warning: this ScienceAlert article misunderstands the original paper and claims that non-replicated studies are 153 times as likely to be cited as replicated ones. This outrageous figure is not what the authors claimed.

Authors claim "cited 153 times less": ambiguous language, where they really meant "cited 153 fewer times". ScienceAlert writer turned that into "153 times more likely".