Uncovering the Effect of Selected Moderators on the Disconfirmation-Satisfaction Relationship: A Meta-Analytic Approach
Social Science Research Network
2013). Uncovering the effect of selected moderators on the disconfirmation-satisfaction relationship: a meta-analytic approach. Innovative Marketing, 9(2) JOURNAL "Innovative Marketing" Abstract Customer satisfaction occupies a central role in marketing. Not surprisingly, researchers have produced an impressive body of literature that focuses on the causes and consequences of satisfaction. The antecedents of satisfaction have been investigated primarily through the disconfirmation paradigm
... holds that satisfaction is the result of conscious mental accounting comparisons undertaken by customers. Furthermore, empirical findings of the disconfirmation-satisfaction link, which are broadly congruent, suggest that when performance conforms to or exceeds initial expectations, a mental state of positive disconfirmation ensues, leading to satisfaction. Despite this insight, a major gap in our understanding concerns lack of generalizability of the disconfirmation model. Specifically, most studies have been conducted in the physical goods setting, thereby raising concerns about the applicability of this model for service exchanges which are more commonplace today. Services differ from goods with respect to intrinsic properties and the manner of delivery. As such, it is possible that the processes underlying customers' satisfaction judgments will differ between goods and services. To investigate generalizability of the disconfirmation paradigm, this paper reports the results of a meta-analysis that the incorporates effect of four moderating variables, i.e., (a) good or service; (b) measure of expectation; (c) definition of satisfaction; and (d) satisfaction scale, on the focal relationship between disconfirmation and satisfaction. The findings suggest that the effect of disconfirmation on satisfaction is weaker for services than it is for physical goods. By including other moderator variables in the analysis, we find that there is sufficient residual variance (in excess of 50%) to warrant further investigation of the expectationdisconfirmation paradigm. Implications of this research for theory development and the scope for further research are discussed.