TRENDS IN SAT CONTENT AND STATISTICAL CHARACTERISTICS AND THEIR RELATIONSHIP TO SAT PREDICTIVE VALIDITY
Gary L. Marco, Carolyn R. Crone, James S. Braswell, W. Edward Curley, Nancy K. Wright · ETS Research Report Series · 1990
document changes related to the SAT and (b) to relate these changes to trends in SAT predictive validity.This study focused on changes in (a) test format, test content specifications, and actual test content, (b) statistical characteristics, and (c) equating procedures associated with SAT test forms taken by classes graduating from high school in years 1971 to 1985.The study covered the period from 1971 to 1974 as well as the 1975-85 period to represent-the -times when average .correlationswith college grades increased and decreased.The SAT item and test data used in the study came from forms administered in November and December of the senior year to members of the high school graduating cohort from 1971 to 1985.These forms accounted for over half of the scores of college-bound seniors who graduated from high school in this period.Changes in Test Content and Format and in Test-Development Procedures Primary among the changes that occurred in the test during the period studied was the shortening of the SAT-Verbal (SAT-V) and the SAT-Mathematical (SAT-M) in October 1974 from 75 to 60 minutes each.From 1970 to 1974 the SAT-V and the SAT-M each consisted of one 3D-and one 45minute section and were administered in one of two fixed section orders at each administration.In October 1974 the SAT-V and SAT-M were shortened to two 3D-minute sections each to permit the introduction of the Test of Standard Written English (TSWE) into the testing program.The revised SAT-V consisted of 85 items rather than 90 items; the SAT-M, like its predecessor, consisted of -2-