The Predictive Validity of the Rokeach Value Survey for College Academic Achievement

With first-semester grade-point-average as the criterion, multiple regression was used to determine whether the Rokeach Value Survey was a successful predictor of academic performance for entering college freshmen. Although ranked values produced significant multiple correlations with the criterion, both alone and in combination with high school achievement and aptitude data, the predictive validity of values was not supported. When separate male and female samples were split into halves for cross-validation, the regression equations derived from the first half were poor predictors for the second half. Across halves, individual values showed low correlations with the criterion as well as reversals in sign. Replication the following year produced similar results. The consequences of the lack of any one value correlating strongly with the criterion is discussed as the most probable cause of sampling error.