This paper describes how work by the United States Office of Personnel Management on the generalizability of employment test validities led to the development of a widely applicable meta-analysis method. The method focuses strongly on estimating the true variance of study correlations and effect size. This validity generalization procedure has been applied to over 500 research literatures in employment selection--each one representing a predictor-job performance combination. This procedure goes beyond the Glass methods of meta-analysis in providing methods for correcting both variances and means of correlations or effect sizes for the distorting effects of the artifacts of sampling error, measurement error, and range restriction. These meta-analysis methods have been successfully applied to non-employment selection topics. Twenty-seven references are listed. (BS)