No abstract

A distribution of several hundred multiple correlation coefficients (R's) obtained at colleges and universities throughout the country among ACT scores and high school grades on the one hand, and firstyear college grades on the other, ranges from .29 through .80. Why is there such variability? Since the ACT tests are the same and presumably the high school grades measure roughly the same variables in all instances, the answer must depend on differences from campus to campus in one or more of the following factors: (a) the criterion is different, (b) the sample is different, or (c) the R's themselves unreliably represent the true association between predictors and criteria, because of careless reporting, atypical student samples, the use of very small samples, or other unusual circumstances. This report examines possible explanations of the fact that grades at some institutions are more predictable than those at others. Reliability of Predictors, Criteria, and PredictabilityThe reliabilities of our predictors, ACT Composite scores and high School Average (HSA, the average of the four ACT student-reported high school grades), are in the low .90's (American College Testing Program, 1965). The reliability of over-all first-year college grade point-average (GPA) can be estimated as .80. Thus, the highest cor-1 Based on a paper presented at the 1968 annual meeting of the National Council on Measurement in Education, February 10, 1968, Chicago, Illinois.

No abstract

No abstract

No abstract