This is an archived version of the Handbook. For the current version, please go to training.cochrane.org/handbook/current or search for this chapter here.

Methods have been described for weighting studies in the meta-analysis according to their validity or risk of bias (Detsky 1992). The usual statistical method for combining results of multiple studies is to weight studies by the amount of information they contribute (more specifically, by the inverse variances of their effect estimates). This gives studies with more precise results (narrower confidence intervals) more weight. It is also possible to weight studies additionally according to validity, so that more valid studies have more influence on the summary result. A combination of inverse variances and validity assessments can be used. The main objection to this approach is that it requires a numerical summary of validity for each study, and there is no empirical basis for determining how much weight to assign to different domains of bias. Furthermore, the resulting weighted average will be biased if some of the studies are biased. Direct weighting of effect estimates by validity or assessments of risk of bias should be avoided (Greenland 2001).