REFERQUAL: a pilot study of a new service quality assessment instrument in the GP exercise referral scheme setting

Exercise Referral Schemes (ERSs) in the UK involve a recommendation from a clinician to a patient concerning the up-take of physical activity, based upon certain pre-determined criteria. The patient then attends a local exercise centre, where an advanced exercise instructor will create an appropriate individualised programme, usually designed to last around 12–15 weeks. The publication of 'Exercise Referral Systems: A National Quality Assurance Framework' [1] was, in part, aimed at improving standards among existing ERSs and aiding the development of new systems. However, many schemes both pre- and post-publication of this document suffered from and struggled with issues relating to poor levels of adherence [2, 3]. Client perceptions of excellent service quality are highlighted as being crucial to the process of attracting new members and retaining existing members in the leisure management industry [4], a theme not uncommon within other service industries [5‐7]. The development of a tool to accurately assess service quality within the Exercise Referral setting, could, therefore, have a considerable impact on informing ERS co-ordinators of the factors that affect adherence and subsequent enable the establishment of strategic interventions to reduce the likelihood of client drop-out. The concept of service quality has been the subject of considerably lively academic debate for two decades, since being brought to the fore by Parasuraman, Zeithaml and Berry (PZB) [8, 9] in SERVQUAL [see Additional File 1], a multiple-item scale aimed at the measurement of service quality, structured around five dimensions:

[9]

The SERVQUAL instrument comprises twenty two questions tailored to assess customers' expectation of service quality, followed by a further twenty two items designed to identify the clients' perception of the same issues. Each item is scored on a seven point Likert scale, thus allowing the equation Q = P - E to be calculated [9].

Since PZB's [9] publication, SERVQUAL has been the target of substantial scrutiny whilst also dominating service quality assessment [10‐15], being applied to numerous settings, including health [16‐20] but not to ERSs. A considerable part of the academic debate over the past two decades has discussed the conceptual accuracy of SERVQUAL's perception minus expectation equation and whether performance-only measures would be more appropriate [21, 22]. This discussion is beyond the scope of this paper; however, it should be noted that in settings such as ERSs, where the perceptions of the absent client are as valuable (and possibly even more valuable) as the life-long adherent, performance-only measures are of no use [21, 22]. Furthermore, Sureshchandar et al. [23] go as far as to say there is "universal conformity that the twenty two items [of SERVQUAL] are reasonably good predictors of service quality in its entirety" (p. 2).

Applying SERVQUAL in its pure form (i.e. without modification) to any service is widely criticised in the literature [14, 24]. However, tailoring the instrument to a specific setting by adding additional items or modifying existing questions to supplement knowledge and understanding, SERVQUAL can give a unique insight into the service quality [25]. PZB [8, 9] confirm adaptations to SERVQUAL are necessary if an accurate measure of service quality is to be established across a diverse range of industries.

K-S tests were conducted on each of the 35 perception/expectation scores elicited from each completion (p < 0.05). In cases where both item scores were drawn from normally distributed data, correlation was calculated via Pearson's "r" (see Table 1).

Instances featuring one or both items being drawn from non-normally distributed data, correlation calculations were made via Spearman's rho (see Table 2). Significant, positive correlations are demonstrated for all of the items calculated using Pearson's "r" (mean = 0.957, SD = 0.02) and Spearman's rho (mean = 0.934, SD = 0.03). Furthermore, Cronbach's Alpha was calculated for all 7 dimensions (35 items) at 0.903 and subsequently for each individual dimension to demonstrate internal consistency (see Table 3). Internal consistency was calculated of the second of the two completions of REFERQUAL.

The reliability and GP dimensions score particularly highly at 0.802 and 0.857 respectively. Most others scored satisfactorily; however, the tangibles and responsiveness dimensions were a little low at 0.554 and 0.619 respectively.

The overwhelmingly supportive statistical analysis above combined with the feedback received from the respondents and scheme organisers confirmed REFERQUAL's applicability and suitability for further evaluation of service quality in the ERS setting. The high correlation scores generated across all 35 items are especially encouraging that REFERQUAL is a valid and reliable research instrument in this setting explaining, at worst, 74% of the variance.

The dimensional reliability scores are also heartening, especially considering the overall instrument internal consistency (α = 0.903). The two dimensions scoring slightly lower than the others (responsiveness and tangibles) are not considered to give cause for concern. Responsiveness is widely acknowledged to be a fundamental contributor to the understanding and assessment of service quality [7, 37] and REFERQUAL's modifications contain no new items that may have been responsible for this abnormality. Further investigations utilising this tool with much larger samples than this small-scale pilot should result in a more satisfactory report for this dimension.

However, the applicability of tangibles as a dimension within service quality assessment has received criticism, relating to client's perception of the concept as a proxy for evaluating service outcomes [37]. However, tangibles is generally retained in factor analysis [38] and the relative importance of the this dimension is thoroughly discussed within the literature, resulting in general affirmative agreement [12]. However, tangibles are the least critical of the proposed service quality dimensions [25]. Cleanliness, modern equipment and aesthetic appeal will rarely countervail poor quality products, unhelpful information and impolite staff. Future studies should incorporate factor analysis to establish whether tangibles remain distinct from the other 6 or whether this dimension requires further revision or augmentation.

Calculations within this investigation have been made via the perception/expectation score generated by the SERVQUAL format. Future studies should analyse both perception and expectation items separately to confirm the validity and reliability of this model particularly bearing in mind the fundamental advantage of the perception/expectation 'gap' model over the performance-only instruments – that the perceptions of those clients not visiting the referral sites at all could be gathered. This investigation solely sought the views of active exercises at various stages of completion of referral. Future studies should seek to investigation across the entire A-E spectrum outlined above.

Preliminary results indicate REFERQUAL to represent a promising model of service quality assessment within the ERS setting. Correlational findings are extremely supportive and are underlined by satisfactory reliability scores. However, future studies incorporating the views of the entire adherence spectrum featuring far greater sample sizes and subsequent factor analysis on the dimensional structure of REFERQUAL will offer far greater insight into the appropriability of this new model.