Multiplicative versus additive models in measuring service quality

Abstract

One may measure service quality either with additive models such as SERVQUAL, SERPERF, and two-predictor model of performance and expectation, or with multiplicative models obtained by log-transformation of variables in additive models. This paper analytically compares multiplicative models with additive models. In addition, it empirically compares predictive performance of alternative models on overall service quality and customer satisfaction with a three-way factorial design: 2 forms (additive vs. multiplicative) $\times$ 3 types (SERVQUAL vs. SERVPERF vs. two-predictor model) $\times$ 2 levels (5 dimensions vs. 22 items). The empirical comparison uses the data gathered to evaluate service quality in five service industries (Hamburger Shop, Pizza Shop, Family Restaurant, Movie Theater, and Bakery Shop) with a standard SERVQUAL questionnaire. The data set includes 1,410 respondents’ evaluation of service quality. The analytical and empirical comparison shows that multiplicative models have three distinct advantages. First, multiplicative models are theoretically supported by the prospect theory and the Weber’s law. Second, multiplicative models allow one to identify the optimal levels of performance and expectation maximizing overall quality or customer satisfaction. Third, multiplicative models predict overall service quality and customer satisfaction better than additive models do.

Keywords:

• additive models
• multiplicative models
• service quality
• sensitivity
• SERVQUAL
• SERVPERF

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This research was supported by the Institute of Management Research, Seoul National University.