Creation of REFERQUAL
Tailoring and augmenting the SERVQUAL instrument has been identified as a necessary requirement when being applied to a new service industry [
25]. A client adherence classification structure was created in conjunction with several ERS organisers. The structure is self-reporting, exhaustive and mutually exclusive. The five statements are constructed to classify clients relating to the extent of adherence to the referral process:
A Referred clients who did not attend at all.
B Clients who started, but dropped out before completion.
C Clients attending, but had not completed the referral at the time of receiving the questionnaire.
D Clients who completed the referral but did not continue exercising.
E Clients who are still exercising having completed referral.
In addition to the self-reporting adherence criteria, clients were also requested to complete certain demographic information addressing factors raised by the literature such as gender, age, occupation [
26‐
29] and marital status [
30,
31]. Two further questions emanating from the literature were included at the beginning of the instrument as demographic issues due to the nature of the items containing no element of expectation and so could not be included as a perception minus expectation item:
'Do you consider yourself to be physically active whilst carrying out the duties demanded by your occupation?'
'Was reducing weight one of the reasons you were referred to the scheme?'
An adaptation of the Blair
et al. [
32] 7-day Physical Activity Recall (7PAR) questionnaire was also included in REFERQUAL to assess the exercise level of participants outside any occupational demand. The inclusion of the adapted 7PAR was aimed at offering the potential for differentiation between those participants who were sufficiently physically active and those who were not irrespective of adherence grouping. Furthermore, respondents were invited to report the location of the physical activity (
i.e. leisure centre or elsewhere), offering the researcher greater insight into whether the service quality of the operational aspects of the scheme or the overall management of the referral site may have most critically affected adherence.
Considering the lengthy and relatively complex nature of SERVQUAL, the 7PAR was simplified from five to three levels of physical activity (light, moderate and vigorous), although duration of exercise was retained. Definitions and examples of light, moderate and vigorous physical activity were given.
Many client-related factors established as being significant determinants to adherence in the literature, formed the foundation from which tailoring of the SERVQUAL instrument was undertaken. Part of the tailoring of the SERVQUAL instrument involved the creation of two new dimensions as a number of topics raised in the literature did not conceptually 'fit' into the existing dimensional framework. The first of these relates to the relationship between the client and the GP – an association distinct to the ERS industry from any previously examined. The second relates to personal perceptions of exercise. The existing 22 items of SERVQUAL were supplemented with the questions [see
Additional File 2].
Modification of SERVQUAL to eliminate negatively-worded questions included in the original instrument has received unanimous support, this was also taken into account when tailoring the items to the ERS Setting [
11].
Piloting methods
The pre-pilot REFERQUAL appended with a feedback sheet was distributed to six attendees at five ERSs and also to the corresponding scheme organisers. The five ERSs had been selected to participate in the main post-pilot study. Feedback was received with respect to the instrument's ease of use, wording and any other topic the respondents felt relevant. Some questions were subsequently removed the instrument whilst others were modified following feedback to enhance applicability and clarity or to eliminate duplicity.
Subsequently, one exercise class comprising 30 clients was invited to participate in the pilot study. The Exercise Professional leading the group distributed REFERQUAL by hand and responses were completed whilst the exercisers were at the centre. The Exercise Professional subsequently repeated the operation two weeks later with the same group. Twenty-seven of the 30 in the group completed both questionnaires, a response rate of 90%, the other three participants were absent for one of the two sessions. Again, respondents were also invited to comment on the instruments ease of use, wording and any general concerns relating to REFERQUAL.
Analysis
In order to determine test re-test reliability, correlation will be calculated via Pearson's 'r' or Spearman's 'rho' depending on the distribution of the individual items. One sample Kolmogorov-Smirnov (K-S) tests will be used to determine whether items are drawn from normally distributed data. K-S testing is a suitable measure of distribution regardless of sample size [
33], is consistent against all alternatives [
34] and frequently outperforms other measures [
35]. Internal Consistency will be measured via Cronbach's α for the overall instrument and the individual dimensions. Cronbach's α is widely regarded a reliable and versatile coefficient, particularly applicable on Likert scale items [
36].