Treatment of Single Likert-Type Items in Surveys

• In Event: Latent Analyses With Surveys in Education Research

Abstract

Cooper (1976) and Whitney (1978) posed tests for polarity of responses to a single Likert-type item. Hsu (1979) posited a test of disagreement in respect to the same frequencies. For small samples, Cooper and Hsu’s tests depend on the difficult calculation of exact multinomial probabilities. Herein, the purpose of this paper is to 1) correct a couple of computational problems with Cooper’s paper, 2) provide computational extension so that regardless of sample size and number of scale steps, multinomial, or multinomial equivalent, probabilities are calculated for Cooper and Hsu’s tests, 3) proffer effect size estimates for all tests, and 4) provide a universally available method for calculation of these tests, as such is not readily available in statistical packages.

Authors

• John D. Morris, Florida Atlantic University

• Mary G. Lieberman, Florida Atlantic University