The issue of multiple-choice (MC) vs. created-response (CR) test-item formats was reexamined at the eighth-grade level in three subject areas: general science, American history, and arithmetic. In each subject area, alternate forms with the same item-content but differing in which items were in which format were prepared from standardized tests. Between 269 and 289 students took each form. Measurement equivalence was substantiated by correlations corrected for attenuation between MC and CR items. Subgroups composed by sex, intelligence, and socioeconomic status (S.E.S.) showed no interactions with relative MC vs. CR discrimination, but one interaction was found with relative difficulty. In arithmetic, CR items were relatively more difficult (than MC items) for lower than for higher S.E.S. students. Comparison of overall item-test discrimination favored the CR items in arithmetic and history, but there was no difference in science. (Author)