... organizing principles of statistics: bias, chance, and causation. Meta-analysis, Feinstein argues, ruins all this eort. It does so because it seeks to “convert existingthings into something better. ... of many dierent studies into one summary measure. is is, to some extent,unavoidable in that clinicians and researchers need to try to pull together dierent studiesinto some useful summary ... repeated and detailed, can be – indeed oen is – wrong; thus valid clinical judgmentscan only be made aer removing confounding factors (Miettinen and Cook, 1981;Rothman and Greenland, 1998).Randomization,...