A Reassessment of the Reliability of Difference Scores in the Measurement of Disconfirmation of Expectations

Abstract

Difference scores have been utilized in past tests of the disconfirmation of expectations model, and are currently used in measurement scales such as SERVQUAL. However, difference scores have been criticized as having low reliability. For example, Prakash and Lounsbury (1983) found very low reliabilities for disconfirmation scores for two products. It is argued here that there are several problems with their analysis that indicate that the reliability of difference scores may not be as low as they claim. An empirical example demonstrates reasonably good reliabilities for disconfirmation measures based on difference scores.