An Evaluation of Formula-Based Population Squared Cross-Validity Estimates and Factor Score Estimates in Prediction

Eight formula-based estimates of population squared cross-validity and nine estimates of factor scores were used in a prediction study involving 31 attitude type predictors and a criterion. These estimates were compared with the conventional cross-validation procedure and the use of data-level variables (raw scores or standardized raw scores) in two separate sample sizes of 150 and 250 cases. The major findings of this empirical study are: (1) Formula-based estimates of population squared cross-validity are as good as those obtained from the conventional cross-validation procedure, (2) factor score estimates are equally good in predicting population squared cross-validity, and (3) two formula-based estimates (Burket's and Rozeboom's) of population squared cross-validity in combination with one of the factor score methods appear to offer the most for practitioners concerned with prediction.