Practice effects with traditional mental test items

Abstract

A nationwide Federal employment program for recent college graduates required applicants to take a multiple abilities test battery. The abilities, each assessed by a separate test part, were Verbal, Judgment, Induction, Deduction, and Number. To equate alternate forms, a sixth test part was included in the test battery. This part could be an additional, parallel version of one of the five ability test parts. At the first test administration one form (A) was used operationally, and each of the five parts of two alternate forms (B and C) was administered to a randomly selected subgroup of test takers. Small but consistent score increases from the first test form to the second were observed. The greatest effects were for Induction and Deduction, next largest for Number, and least for Verbal and Judgment. At two subsequent administrations the order of alternate form administration was reversed (B and A, C and A), providing a counterbalanced design to assess the effects of alternate form, samples, and practice. Data from 66,303 test takers supported the hypotheses of practice effects. These data suggest that practice is most effective for item types constructed according to specific rules, next effective for test parts subject to speededness, and least effective for test parts tapping general information.