Concurrent and Predictive Validity Estimates for the WISC-R IQs and ELPs by Racial-Ethnic and SES Groups

Concurrent and predictive validity estimates for both WISC-R IQs and Estimated Learning Potential (ELPs) are presented for 396 children from grades one through eight stratified by racial-ethnic groups (Anglo, black, Mexican American) and social classes (middle and lower). Reading and math achievement scores serve as criteria. Concurrent predictions of reading and math achievement clearly are made more accurately with IQs than with ELPs. Predicting achievement three years later also tends to be made more accurately with IQs than with ELPs. Twenty-six of the 48 pairs of correlations significantly favor the IQ-Achievement predictions while none significantly favors the ELP-Achievement predictions. The magnitude of the IQ-Achievement correlations is high, accounting for about 50% of the variance associated with both reading and math. Validity coefficients between the three racial-ethnic groups and between the two classes are very similar (i.e., there is little evidence of racial-ethnic or social class bias). The predictive relationships tend to be higher than the concurrent relationships.