2 min read

Peer review and community endorsement

In most academic fields there are journals it’s easy to publish in. Some of these are outright scams, but some are just not that fussy about the importance of results. In the experimental sciences, being able to publish negative or otherwise uninteresting results can be very important. Even in fields where ideas, rather than data, are important, being able to get research out into libraries is valuable –  though preprint servers such as arXiv are now filling that niche. 

Some of the papers published in, say, Communications in Statistics – Simulation and Computation are important and influential – eg, Pepe and Anderson’s classic on  structures of marginally specified models – but most of them will be of interest only to a few with common interests, and most of them, good or bad, are there because they’ve been rejected by other journals. 

Publication in this sort of journal doesn’t represent any sort of endorsement of the content by the broader community in that discipline – in fact, quite the reverse. Sometimes the lack of endorsement turns out to be to the discredit of the discipline, as with Margaret Pepe’s paper; more often, not. 

No-one’s going to set up an elaborate hoax to get a paper published in Comm Stat Simul Comp, let alone one of the actual scam statistics journals, because the response would be a mixture of blank looks and yawns. The claim that a hoax paper reflected the general views of the statistical community about the overgeneralised Beta distribution would be an obvious straw man, but in addition would be really boring. As Skeptic Magazine has shown, if the field is gender studies the result isn’t necessarily boring. It’s still a straw man.