No abstract

No abstract

A new meta-analytic approach for assessing the generalizability of correlations or criterion-related validities is presented. The new approach incorporates specific procedures for estimating the sampling variance of corrected and uncorrected correlations. A major advantage of the new approach is the ease with which studies without any sample-based artifact data, with partial samplebased artifact data, or complete sample-based artifact data can be incorporated together into a practical procedure for meta-analysis. Monte Carlo studies indicated that the new procedure is consistently more accurate in estimating the mean and variance of true validities than the current correlation-based procedures. The implications of the new approach and its procedures for more appropriately estimating confidence limits, magnitudes, and the generalizability of correlations are discussed.A great deal of research in psychology and education has been focused on integrating the results of various studies to determine the degree to which bivariate relationships generalize across studies or situations or both. Such statistical investigations are commonly referred to as meta-analytic procedures (Hedges & Olkin, 1985;Hunter & Schmidt, 1990). More specifically, within the area of employment-test validation, procedures for determining the generalizability of correlation coefficients, covariances, and regression slopes are referred to as validity generalization (VG) procedures (

No abstract