Meta-analysis by combining p-values: Simulated linkage studies

Meta-analysis has been little explored to make an overall assessment of lin kage from different studies. In practice, it is likely that published linka ge studies will only report p-values. We compared the performance of the wi dely used Fisher method for combining p-values with that of pooling raw dat a. More loci were consistently found by pooling raw data. In the absence of further information, combining p-values can provide an overall, but limite d, assessment of different linkage studies. However, meta-analysis would be better viewed as a preliminary step toward the goal of analyzing the poole d raw data. ((C)) 1999 Wiley-Liss, Inc.