Customer satisfaction has long been recognized as a central objective in successful retailing strategies. In the evaluation of these strategies, retailers often request customers to complete satisfaction surveys. One of the strongest influences on customer satisfaction is a disconfirmation judgment, which is defined as the difference between the actual retailer performance and some standard of performance. As a result, a better understanding of the disconfirmation judgments may improve the implementation of customer satisfaction surveys.

With the evolution of research on customer satisfaction, three important issues pertaining to the measurement of disconfirmations have emerged. First, multi-item measures (scales) have become the accepted method of measuring disconfirmation judgments. Although the use of multi-item measures improves the psychometric properties of these scales, it also increases survey length, which may reduce response rates. Second, two approaches have emerged in measuring disconfirmations. One approach asks first for a pre-consumption assessment of the standard (e.g., How do I expect the retailer to perform?), and then for a performance rating (e.g., How did the retailer actually perform?), calculating disconfirmation as the difference score between the two measures. The second approach measures disconfirmation directly, asking in a single measure how the performance compared to the standard (e.g., How did the retailer perform compared to what I expected?). The two approaches differ not only in how they are measured, but also in whether the comparison standard is measured before or after consumption.

 A final issue arises from research indicating that several standards of comparison based on goals, expectations or norms may be used, with some studies advocating the use of multiple standards. Thus, in the design of customer satisfaction surveys using disconfirmations, retailers must address a number of measurement issues. Retailers must decide how many and what type of comparison standards to use. In addition, retailers must decide on whether to measure disconfirmations directly or with difference scores. To provide some insight into these issues, we examine the distinctiveness of goals, expectations and norms as both comparison standards and disconfirmation judgments and assess their uniqueness in a predictive relationship to satisfaction.

The studies employed three retailing situations: one measured consumer comparison standards in a hotel context, and the other two measured consumer disconfirmation judgments in a restaurant and university contexts. The results suggest that while goals, expectations, and norms are distinct as comparison standards, they lose their distinctiveness as direct measures of disconfirmation. Two explanations may account for this lack of dimensional stability. First, consumers may simplify the process by integrating the standards into a single evaluation context prior to constructing disconfirmation judgments. Second, consumers may assimilate the standards in the process of constructing disconfirmation judgments. Thus, while disconfirmations measured using difference scores should maintain the dimensionality of their respective pre-consumption standards, direct measures of disconfirmation may not. It is therefore critically important to make a conceptual distinction between a pre-consumption standard and a post-consumption standard used in formulating a disconfirmation judgment.

            Furthermore, the results indicate that while the use of disconfirmation judgments based on multiple comparison standards improves the predictive ability in models of satisfaction, the incremental predictive improvement for each additional disconfirmation was quite small. Based on a tradeoff between survey efficacy and survey length, we generally recommend that retailers use one or at most two comparison standards in satisfaction surveys. Compared to goal disconfirmations and norm disconfirmations, expectation disconfirmations had the greatest predictive ability in these data. However, due to the lack of distinctiveness among disconfirmation judgments and minor differences in the unique effects on satisfaction, there appears to be little difference among direct disconfirmation measures. Thus, the choice of a specific standard should be based on conceptual considerations rather than trying to select the “best” standard.

The measurement of multiple service dimensions using multiple comparison standards also raises issues in the analysis of these data. We introduce the correlated uniqueness model as an analytical framework well suited for this type of analysis, as it is more likely to result in admissible solutions and can be useful in modeling multiple disconfirmation judgments with two-dimensional items. Finally, notwithstanding the criticisms of the difference-score approach to disconfirmation measurement, direct measures of disconfirmation can also be problematic. As demonstrated here, direct measures of disconfirmation will likely change the dimensionality of multiple comparison standards, and as a result may lose some of the information contained in the pre-consumption standards. So the choice of disconfirmation measurement approach may boil down to which problem is worse, a loss in measurement reliability or the loss in construct dimensionality.