3 ms·

Is this what is called “p hacking”? Looking a among many different criteria until you find one that is statistically significant?

• ImaCake 6y ago
  This is a good question to ask and I am not sure why you are being downvoted. The answer is that p-hacking was a real problem back when these kind of large multi-gene frequency analysis studies were just starting out. Not intentionally, mind you, just that the people running the studies were biolgists first, and statisticians only on an ad hoc basis. In the late 90s and early 2000s there was a lot of debate about the poor quality of statistical analysis in the field and things started to improve. Today these studies have a bunch of descriptivist and bayesian tools they can use to sort out real positives and negatives from false ones. The result is that these studies now discover real and important genes that impact complex diseases. We are finally getting close to the dream of personalised medicine. A good further read is the 1998 paper that describes the problem in detail: http://ib.berkeley.edu/labs/slatkin/popgenjclub/pdf/terwilliger-weiss1998.pdf http://ib.berkeley.edu/labs/slatkin/popgenjclub/pdf/terwilli...