Why most social science "studies" should simply be dismissed out of hand

We can always data mine for correlations until we get one we like. So when a socialist does a "study" proving that capitalists are psychopathic, or a libertarian "demonstrates" that lax environmental regulation makes us healthier... just ignore them. (These things might be true, but you can figure the studies themselves are worthless.)

The only studies that one should pay any attention are one's where the study's author expected the opposite result, but reluctantly reports that the study contradicted his expectations. So when Roland Fryer set out to show bias in police shootings, but found none, that is worth paying attention to. Similarly, when Gregory Clark attempted to show that social mobility had increased in modern times, but found it hadn't, again, that is worth noting.


  1. What you propose sounds suspiciously similar to the genetic fallacy, such as when a former atheist becomes a Mormon or when a former Christian preacher becomes a communist.

    1. All comments about something you say sounding "suspiciously similar" to some other idea should be dismissed out of hand.

  2. It'd be interesting to make a list of such findings. Another one that comes to mind is Robert Putnam's finding in Bowling Alone that diversity is bad for communities.

  3. It's not the genetic fallacy. In fact it follows from the argument John Ioannidis makes in Why Most Published Research Findings Are False: