Measuring Service Quality with SERVPERF

SERVPERF, the performance component of the Service Quality scale (SERVQUAL), has been shown to measure five underlying dimensions corresponding to Tangibles, Reliability, Responsiveness, Assurance, and Empathy (Parasuraman, Zeithaml, & Berry, 1988). This paper describes a validation study, employing four different datasets, of a shortened 15-item version of the SERVPERF scale to be called SERVPERF-M. Exploratory and confirmatory factor analytic techniques were used to explore the dimensionality of the scale. Although internal consistency estimates for all scales were very satisfactory for all four datasets, the factor structure was somewhat unstable with Responsiveness, Assurance and Empathy tending to define a single factor in three of the sets and Reliability and Tangibles to define two other factors. Rasch analysis was employed to gain further insights into the behaviour of the items. These analyses suggested that the five factors can be treated as five different stages of service quality, rather than as five qualitatively different dimensions. The Rasch analysis also suggested that the items in both SERVPERF and SERVPERF-M are too easy to rate highly and that more “difficult” items need to be added to the scale. If this is done, it is likely that some of the confusion that exists over the dimensionality of this scale will disappear. Measuring Service Quality 3 Measuring Service Quality with SERVPERF The concept of service quality is not universally understood and is often used as an umbrella term to cover a range of impressions gathered by customers when dealing with vendors. These impressions, however, are important factors that influence buying behaviour and firms are very conscious of the need to improve this aspect of their operations, either by staff training or direct investment in facilities. If training programmes aimed at improving service quality are to be effective, there is a need to identify the specific constructs underlying this generic term. The present study reports on the construct validity of one of the main instruments used to measure this construct, the SERVQUAL scale developed by Parasuraman, Zeithaml, and Berry (1988), and a shortened form of this scale (SERVPERF-M) developed by the researchers for use in an Australian small-business setting. The psychometric properties of the SERVQUAL scale have been the subject of considerable research in recent times. The scale was developed from an initial pool of 97 items generated through a series of focus group sessions conducted with consumers (Parasuraman et al., 1988). The initial pool of 97 items was reduced to 22 to form the SERVQUAL scale with a reported reliability above .90 (Parasuraman et al., 1988). The scale was said to tap five different underlying dimensions of customer service termed Tangibles, Reliability, Responsiveness, Assurance, and Empathy. Other researchers have questioned the validity of the five-factor structure. Partial support was reported by Gagliono and Hathcote (1994) who obtained four factors corresponding to Personal Attention, Reliability, Tangibles, and Convenience. Babakus and Mangold (cited in Brown et al., 1989) found a single factor when SERVQUAL was used in a hospital setting. The five-factor structure was also rejected factor by Cronin and Taylor (1992) who conducted studies across five retail settings. Finn and Lamb (1991) tested the five-factor structure in a retail context. They reported that while the reliabilities for each of the five factors were acceptable with estimates ranging from .59 to .83, confirmatory factor analysis indicated that the data did not fit the model. Whilst the questions that make up the SERVPERF scale might well cover most of the broad domain of service quality, the issue of dimensionality is a fundamental concern since the utility of SERVPERF in guiding management and staff training decisions depends on the capacity to segment customer service into factors that are meaningful to staff and trainers in an Australian small business context. Whilst other studies have already explored the question of dimensionality, the present study took the additional step of not only identifying but also examining the relative difficulty of achieving high standards in the different dimensions in a training context. An associated aim of the present study was to see what improvements could be made to the scale that would lead to a better definition of the core dimensions of service quality. Three main statistical techniques were used to achieve these aims. The first was traditional item analysis focussing particularly on developing subscales that had high internal consistency. The SPSS RELIABILITY routines were used for this purpose. The second set of techniques included exploratory and confirmatory (LISREL) factor analysis. These were used to help define the main constructs underlying the SERVPERF scale and the relations among these constructs. The final technique involved the application of the principles of Rasch analysis to gain a clearer picture of the role of individual items and groups of items in the measurement of service quality. The study is presented as two parts. The first part is based on what will be termed Dataset 1 and concerns the initial analysis of the 22-item scale and the derivation of a 15-item SERVPERF scale. The second part is based on three further datasets and describes further validation of the shortened version of SERVPERF which is to be called SERVPERF-M. Measuring Service Quality 4