Cross-validation sample sizes

The squared cross-validity coefficient is a measure of the predictive valid ity of a sample linear prediction equation. It provides a more realistic as sessment of the usefulness of the equation than the squared multiple-correl ation coefficient. The squared cross-validity coefficient cannot be larger than the squared multiple-correlation coefficient; its size is affected by the number of predictor variables and the size of the sample. Sample-size t ables are presented that should result in very small discrepancies between the squared multiple correlation and the squared cross-validity correlation , thus facilitating the selection of sample size for predictive studies. In dex terms: cross-validity coefficient, least-squares regression, multiple c orrelation, prediction, sample size.