>"The problem with saying "Half of this research is worthless" is that it's impossible to know which of the research will be worthless before you do the researc

>"The problem with saying "Half of this research is worthless" is that it's impossible to know which of the research will be worthless before you do the research." Once you know the common problems, it is actually quite easy to spot design flaws or a history of hiding unwanted results. When you know what to look for you can filter out crappy research in a few seconds. It could probably be automated to a large extent. For example, a common thing in biomed is to have a "headline claim" that x protein/drug/etc influences y behavior/health outcome/etc. But then there is never a scatterplot of x vs y. Another is lack of "blinding" (hiding the experimental conditions from the researcher/subjects). There are more, but you get the picture. The presence of these baffling practices quickly demonstrates they don't know what they are doing, and the research should not be funded.