Health, Exercise, and Sports Sciences ETDs

Publication Date

Winter 12-24-1976

Abstract

Previous research has shown that the responses to items in a test of the affective domain are not affected by context when the items are embedded in different random subtests. Other studies in the affective domain show that the mean responses of items are influenced by the favorableness item compositions of the subtests. In multiple matrix samples, subtests of an entire test are taken by subgroups of a population. The influence of favorableness item composition on the total score of a given dimension in the affective domain has not been studied. This study offers evidence toward the following questions: (1) Do different favor­ableness item compositions of an affective test influence the means of a total dimension score? (2) Do different favorableness item compositions of an affective test influence the individual item responses of the items composing the total dimension score?

The favorableness item composition was determined by the item means estimated for 60 subjects taking an entire attitude inventory. Forty subjects per subtest then took one of the favorable, random, or unfavorable subtests in which all 10 items from the dimension were embedded. A one-way analysis of variance revealed a nonsignificant f-ratio for the means of the 10-item dimension scores of the three subtests. Two-tailed two-sample Kolmogorov-Smirnov tests were used to test for differences in the ten individual item responses between the favorable and unfavorable, favorable and random, and random and unfavorable subtests. Since only three of the 30 Kolmogorov-Smirnov tests comparisons between responses to individual items were significant, it was concluded between responses to individual items were significant, it was concluded that the responses to the individual items were not affected by being embedded in the different subtests.

Data snooping suggested that possible trends may exist in the data. When ascetic items were arranged in order of favorableness of response to the items in the total test, it was noted that there appeared to be an increasing difference in favorableness of response in the favorable and unfavorable context. These differences in mean response, i.e., the item means in the favorable context minus the item means in the unfavorable context, were analyzed to determine whether there might be a trend relative to the order of favorableness in the entire test. Also, the same analyses were used for the differences in mean response between the favorable minus random and the unfavorable minus random settings. A nonparametric test for increasing or decreasing trend revealed a significant trend for the item means in the favorable minus unfavorable subtests.

Document Type

Thesis

Degree Name

Physical Education, Sports and Exercise Science

Level of Degree

Doctoral

Department Name

Health, Exercise, and Sports Sciences

First Committee Member (Chair)

Dale L. Hanson (Co-chair)

Second Committee Member

Lynnette B. Plumlee (Co-chair)

Third Committee Member

Kenneth Carl Lerston

Fourth Committee Member

Peggy Blackwell

Share

COinS