skip to main content skip to footer

Validation of the SAT and New Item Types Against Four-Year Academic Criteria

Author(s):
French, John Winslow, 1918-
Publication Year:
1957
Report Number:
RB-57-04
Source:
ETS Research Bulletin
Document Type:
Report
Page Count:
65
Subject/Key Words:
College Board, College Entrance Examinations, Grade Prediction, Predictive Validity, Scholastic Aptitude Test (SAT), Scores, Test Format

Abstract

The College Board felt it was desirable to consider the possibility of altering the test materials in the SAT to permit the reporting of scores in addition to the present verbal and mathematical scores. Towards this end, eleven newly developed or newly adapted tests were tried out in combinations of four at ten colleges. An earlier report presented the findings for the validation of these tests and the SAT against freshman average and grades in specific freshman courses. This report compares the freshman validities with four-year criteria, including the cumulative average, graduation vs. non-graduation, comprehensive examination, and major-field grades. The pattern of test validities for the four-year criteria closely resemble those for the freshman criteria. These data show the high school record to be less good at predicting major-field work than it is for predicting freshman average grades. Tests of government and literature information were the most successful among the experimental tests. When corrections were made for restriction of range and test length, these tests were more valid for the cumulative four-year grade average than the SAT-V or SAT-M. Multiple correlations based on these data show that a test composed of six 25-minute subtests would be only .01 less reliable and .08, on the average, more valid than the present two-score SAT. In addition it is shown that a composite having more than two scores would make possible better differentiations between major-field areas than is possible at present.

Read More