The reliability, validity, and design of the Defining Issues Test.

Abstract

The Defining Issues Test, developed by Rest, was given to 365 subjects in three age groups, junior high (n =60,meanage= 13.9 years); high school (n = 200, mean age = 17.3); and college (n = 105, mean age = 20.2). The main findings of Rest and his associates were replicated. Each scale score discriminated significantly among age groups (p < .01), as did the Principled Reasoning (P) score. The validity of the P-score with respect to age group was estimated to be .48, and its reliability was estimated to be .70. The design of the test was criticized on the grounds that some stages have many more items associated with them than do other stages (hence, scale scores are not comparable with one another), that different dilemmas have different numbers of items from a given stage, that there was a significant order artifact for Stage 3 and Stage 4 items, and that this order effect interacted with age group. Finally, there was evidence in three of six dilemmas that a subject's decision was correlated with his stage of reasoning. It was suggested that the relation of stage to age could be clarified if the design of the test were changed to eliminate the possible effects of decision.

The Effects of Situational Factors on Moral Judgments

“…The result of these modifications was a more equal representation of the various moral stages among the issues, which, in Rest's (Note 2) original test, were biased in frequency toward the conventional and principled levels (as has also been noted by Martin, Shafto, & VanDeinse, 1977).…”

Age Trends in Judging Moral Issues: A Review of Cross-Sectional, Longitudinal, and Sequential Studies of the Defining Issues Test

“…In short, the cohortsequential analysis supports the cross-sectional data and the longitudinal data in suggesting that there is ontogenetic change on DIT P scores.7 Table 11 presents the design for what Schaie calls a "time-sequential" analysis, in which age and time of testing are the two factors. These are essentially three cross-sectional studies reported previously as unifactorial analyses by Martin et al (1977), Rest 6 Since the age factor was based on longitudinal data, a repeated-measures procedure was utilized on this factor but not on the cohort factor (see Baltes & Nesselroade 1972, p. 247). See Rest (1975) for details on sample and testing procedures.…”

The Development and Validation of a Multidimensional Measure of Moral Orientation

“…As previously mentioned, numerous early validation studies failed to control for variance in moral development scores that could be attributed to education and IQ (Martin, Shaftro, & Van Deinse, 1977). Since these studies, researchers have found a significant association between moral development scores and IQ (Dawson, 2002;Taylor, 1978).…”