Potential dangers in the customary methods of conducting meta-analyses Recombinant versus urinary follicle stimulating hormone

The customary method of combining success rates in meta-analyses may often result in serious biases, leading to erroneous inferences. This arises because of an inadmissible pooling of frequencies from heterogeneous sources. The fundamental statistical principle, that the magnitude of an effect' should always be tested against the variation in that effect over the sample, may not therefore be satisfied. A simple, but rigorous, alternative method is described.