Add new comment

Dr. K and Dr. G will agree with me that a suggestive study is a much different animal from a conclusive study.  
Too true.  90+ % of "studies" done in the medical world are very weak, IMO.   Retrospective and case control studies, the majority of studies done, at best can demonstrate correlation.  Usually this is because doing a proper study is too expensive or will take too long.  Naturally this doesn't prevent the press from reporting "X causes Y". 
Even the gold standard randomized/controlled/blinded studies accept a (calaulated) error rate of 5% for Type I error and 80% for Type II error.  Even in an ideal context, when dealing with very complez questions, it is also really hard to eliminate all sources of systemic bias even with randomization and it is often also extremely difficult to arrange things so that you're really comparing apples to apples.