ESCALA DE AVALIAÇÃO DA QUALIDADE DOS SERVIÇOS PRESTADOS POR ACADEMIAS DE GINÁSTICA (QUASPA) RATING SCALE FOR QUALITY OF SERVICES PROVIDED BY GYMS (QUASPG)

In order to identify factors that interfere with gym clients’ choice, permanence and withdrawal, the aim of this study was to build and validate an instrument to assess the quality of services provided by gyms. The study has four stages (question identification, evaluation of the analysis matrix clarity, importance and relevance, reliability analysis on instrument questions, and question analysis as to the instrument construct). A total of 51 criteria were identified and deemed important for service quality. The Analysis Matrix was composed of the following dimensions: structure, management, marketing and accessibility. The question and matrix evaluation ruled out 19 questions. The reliability evaluation excluded one question and the accessibility dimension. The Confirmatory Factor Analysis model evidenced adequate goodness of fit after the exclusion of eight questions and performance of six correlations. Composite Reliability, Average Variance Extracted, Discriminant Validity, and Cronbach’s alpha showed acceptable values. Thus, it is concluded that the instrument called “Rating Scale for Quality of Services Provided by Gyms” (QUASPG) has satisfactory psychometric properties, being composed of 21 questions distributed into three dimensions, namely: Environment, Management and Marketing.


Introduction
Gyms are inserted in a market where interventions are determined by trends, which lead them to continuously seek innovations. Fitness centers that do not adjust to the market have allegedly loyal clients who actually signed up only for convenience, for living nearby and for the professionals that work there, and not necessarily for the quality of the services provided 1 .
A way to follow market trends and have loyal customers is to gather information on the quality of the service provided by listening to customers who have a great knowledge about the company. Information can be leveraged to enhance products and services so that users' requirements are met. In addition, it can also help to discover new services, which makes companies grow their business 2 .
For companies to meet their customers' expectations, the former's missions and goals need to be put into practice, and actions are fundamental for quality strategic planning 3 . From this perspective, Campos 4 presents quality as a product or service that meets a customer's needs in a safe, reliable, accessible and timely manner.
In order to assist gym clients, scales to assess the quality of services provided by fitness centers were developed in several countries, including Turkey 5 , Korea 6 , Spain 7-11 , South Africa 12 , Canada 13 and Greece 14 . It is worth noting that the instruments were designed considering the culture of each region, which increases the specificity of results found by these instruments.
Concerns about the quality of services provided by gyms is reflected on high turnover rates 15 . In addition, managers also worry about withdrawal from physical exercise, which may be related to poor service provision involving inadequate customer service, insufficient workout equipment and failure to promote socialization among gym users.
Thus, considering reasons for gym clients' turnover or withdrawal, the objective of this study was to build and validate an instrument to evaluate the quality of services provided by fitness centers for the Brazilian reality.

Methods
This is a psychometric research composed of four stages. It has as main objectives to describe characteristics and measure individual or group variables. Thus, using a psychometric instrument as research method presents its advantages, such as sourcing of information on a large group of people in a short period of time, coverage of a large geographical area, and uniformity. The same instrument is applied to all subjects and ensures anonymity, which can make interviewees more comfortable with their answers, providing more coherent information and facilitating data analysis 17 .
The investigation was reviewed and approved by the Ethics Committee on Research Involving Humans of the State University of Londrina (Legal Opinion CEP/UEL: 555.415). All participants involved in the research phases signed a free and informed consent form, which briefly presented research objectives, methods and purposes.

st Stage: Identifying Items and Designing the Analysis Matrix
For the first stage, 90 clients and 30 gym managers from the city of Londrina (Paraná, Brazil) were invited to answer an open-ended questionnaire that aimed to identify aspects regarded as important in gym service provision.
Data were tabulated and categorized by means of content analysis 18 to establish criteria considered as relevant for gym service provision. This analysis allowed building the items that were part of the subsequent phases of the study.
To build the instrument matrix analysis, a content analysis on the items elaborated was performed 17 . The preliminary building of the instrument dimensions and indicators was done by the researchers of this study. It is worth noting that in this phase, three management professionals were consulted as well; they worked in business consulting and helped confirm the building of the matrix analysis items and dimensions. The analysis matrix items and dimensions were categorized after data collection 2 nd Stage: Assessing the Clarity, Importance and Relevance of the Instrument Items and Analysis Matrix This stage sought to eliminate possible inadequacies identified in the questions and the analysis matrix during the previous study stage. Thus, ten evaluators received a questionnaire, with four being Physical Education instructors working as gym coordinators, three administration undergraduate course professors, and three Physical Education professors with academic and/or professional knowledge in administration.
The instrument sent to the experts sought to evaluate language clarity and question importance and relevance. The evaluators should answer with the aid of a 0-10 liker scale, considering that with 0-7 points the question was discarded, and with 8-10 points the item was deemed valid 19 . Moreover, the analysis matrix built in the first study phase was verified by the experts, who analyzed the association between question and dimension 19,20 . For the question to reach an acceptable index, the item should have an agreement of 70% among the evaluators 21 .
Therefore, for the questions to be accepted to compose the third study stage, the items should reach a minimum mean of eight points for clarity, importance and theoretical relevance, and a minimum of 70% of agreement on the analysis matrix among the evaluators.

rd Stage: Assessing Instrument Reliability
This stage sought to assess the temporal stability of the scores of the instrument questions and dimensions that met the criteria established in the second study stage. To do so, the instrument was applied to 76 gym clients in Londrina (Paraná) on two occasions, with minimal interval of seven days, and maximum of 14 days between the first and the second applications. During both applications, the respondents were requested to identify themselves so that they could be contacted for questionnaire application in the second moment of this stage. For all instrument statements the participants should report their concept with the aid of a 5-point Likert scale, which represented the following concepts: 1 -Terrible; 2 -Bad; 3 -Regular; 4 -Good; 5 -Excellent.
For this stage data analysis, the intraclass correlation coefficient (ICC) was adopted to analyze instrument reliability. For the questions and dimensions to be considered acceptable, ICC indexes should reach a score equal to or higher than 0.6 22 .

th Stage: Construct Analysis
In this stage, Londrina's gym clients answered a questionnaire covering the sample sociodemographic aspects and the questions that reached acceptable indexes in the instrument reliability evaluation. For each instrument statement, the clients should use the Likert scale employed in the study previous stage.
Construct analysis used the instrument confirmatory factor analysis, which intends to confirm pre-established structural patterns. This was performed considering that, in this stage, there was previous information on the factor structure 23 , that is, the opinion of the experts investigated in the second stage, who qualitatively described the corresponding evaluation between items and dimensions.
The statistical analysis checked item normality, considering as acceptable those actimetry (sk) and kurtosis (ku) values inferior to 2 and 7, respectively. The existence of outliers was assessed through the Mahalanobis squared distance (D²) 23 . In addition, the factor weights of the questions were assessed, and ƛ≥0.4 values were excluded 24 .
Then, structural equation analysis was conducted; for goodness of fit evaluation, the following tests were performed: Chi-Square goodness of fit (X 2 ), and the lower the value the better the goodness of fit; Chi-Square on degrees of freedom (X 2 /g.1), considering as acceptable values those inferior to 5; Goodness-of-Fit Index (GFI) and Comparative Fit Index (CFI), whose values should be higher than 0.9; Penalty of Comparative Fit Index (PCFI) and Penalty of Goodness-of-Fit Index (PGFI), considering as ideal goodness of fit all values above 0.6; and finally, Root Mean Square Error of Approximation (RMSEA), whose values should be inferior to 0.08 23 . Modification indexes were checked so that due adjustments could be made to the first-and second-order models.
After adjustments to goodness-of-fit values, initial degrees of freedom values were compared to final ones in order to assess whether the final model goodness of fit was better than that of the initial model, considering that, for improvement, the value obtained in X 2 0.95 should be lower than the X 2 dif value. In addition, composite reliability (CR) was calculated to assess instrument reliability, considering that values equal to or higher than 0.7 present adequate reliability 23 Mean Extracted Variance (MEV) was calculated for model convergent validity, in which values equal to or higher than 0.5 indicate acceptable convergent validity 24 . Discriminant Validity (DV) was calculated in order to assess whether the items that integrate a factor does not correlate with other factors. It is worth stressing that DV is confirmed when the MEV of the factors are higher than or equal to the correlation square between these factors (r²) 24 .
Finally, to evaluate the scale internal consistency, Cronbach's alpha coefficient was employed, which assesses the inter-relation between items in the same domain. Values above 0.7 are deemed acceptable, and values higher than 0.9 are deemed excellent 25 .

Results and Discussion
In the first stage of the study, 49 items were identified and distributed into three dimensions, namely: Environment (20 questions), Management (18 questions) and Marketing (11 questions). The Environment dimension was also identified in instruments developed for the Turkish, Korean and South African realities 5,6,12 . However, even without presenting direct correlation, Management and Marketing dimensions were identified in other studies by means of items linked to dimensions that covered broad themes [5][6][7][8][9][10][11][12][13][14] . The dimensions found in other studies, regardless of having different nomenclature, show that the items that compose them correlate with those identified in this study.
The Environment dimension presented questions associated with stimulus, structure and cleanliness; clients consider as rating criteria all aspects that characterize the place, such as organization and distribution of objects, furniture and equipment, cleanliness of workout spaces 26 , in addition to different physical elements that call their attention [26][27][28] .
The Management dimension was associated with matters of planning, service, management and qualification, which are part of an integrated project for achieving organizational objectives 29 . These objectives consider a deep analysis on the company internal and external environments 30,31 , the direct relationship between company and client 32 , decision making, organization, leadership and business control 33 , besides knowledge and professional experience for the role 34 .
The Marketing dimension presented items reporting innovation, communication and strategy, which are connected to the establishment of profitable relationships with clients 31,35 , development of products and services 33 , presentation of feedback on the clients' matters of interest, and use of physical, financial and human resources to maximize market opportunities 32 .
Analyzing the answers provided by the consulted evaluators (Table 1), it was possible to identify that only 30 questions reached acceptable levels as to the evaluation of the agreement on the existing relationship between the analysis matrix questions and dimensions (>70%) 21 , and as to item clarity, importance and theoretical relevance (>8) 21 . The instrument reliability evaluation revealed that only "Conduct norms defined for use of gym equipment/spaces" did not show an acceptable temporal stability index (0.56 ICC), according to Vallerand 22 . The other items evidenced adequate intraclass correlation coefficient values, which ranged from 0.64 to 0.89.
Thus, the evaluation of the instrument reliability as to its dimensions and overall context (Table 2) excluded the question that did not reach an acceptable reliability index in this study stage. Analyses on instrument dimensions and overall assessment evidenced acceptable reproducibility indexes both in the first and second instrument applications (ICC >0.6) 22 . Finally, the confirmatory factor analysis was run with 29 questions distributed into the following dimensions: Environment (11 items), Management (13 items) and Marketing (5 items). The question normality analysis showed asymmetry values ranging from -1.402 to -0.153, and kurtosis values from -1.145 to 1.853, presenting normal distribution. Besides, after the normality analysis conducted through the Mahalanobis distance, the option was for keeping all individuals, assuming the normal distribution of the subjects' data 23 .

Table 2. Intraclass Correlation Index of instrument dimensions and overall evaluation
The confirmatory factor analysis initial model (Figure 1) presented inadequate adjustment quality in CFI (0.802), GFI (0.754) and RMSEA (0.088) indexes. However, X²/go (3.242), PCFI (0.738) and PGFI (0.645) indexes showed acceptable values in this construct evaluation phase 23 . Before starting the adjustment process based on the modification index, question Q01, referring to "Pleasant music for workout" was excluded from the CFA, since the correlation value presented was 0.28, which is considered as insignificant according to Hair et al. 24 . Moreover, in order to reach an acceptable goodness of fit in the first-order analysis, it was necessary to eliminate eight questions (Q23 "Policies for a good relationship between the gym and autonomous professionals (personal trainers)", Q29 "Promptness from the gym in providing services", Q10 "Well-defined gym administrative/hierarchical structure", Q5 "Good gym management", Q21 "Workout guidance must be provided exclusively by professionals with a degree in Physical Education", Q16 "Balance between the number of people and physical space(s) provided by the gym for workout", and Q28 "Provision of motivating elements such as sound and TVs"), as well as to make six correlations in the factor analysis ( Figure 2). In order to assess the correlation of Environment, Management and Marketing dimensions with Overall Evaluation, the second-order confirmatory factor analysis was adjusted ( Figure 3). The second-order model presented the same results as the first-order CFA final model, which were considered as acceptable (Table 3).   (Table 4). Subsequently, the internal consistency of the instrument overall evaluation and dimensions was analyzed. As for the construct overall analysis, Cornbrash's alpha stood at 0.927, which is rated as excellent. Considering the instrument detailed evaluation, there were internal consistencies classified as good in Environment (0.871), Management (0.872) and Marketing (0.837) dimensions 25 .
Concerning the values obtained in the instrument construct validation process, it was possible to confirm the three dimensions evidenced in the content analysis in the matrix identification. However, some items corresponding to stimuli (Environment dimension), planning, administration (Management dimension) and innovation (Marketing dimension) were suppressed throughout the analyses. In fact, the indicators seemed to be linked more to the criteria approached by managers than to gym users.
Thus, the construct correlated with the dimensions as follows: Environment, based on structure and cleanliness indicators; Management, related to service and qualification indicators; and Marketing, represented by communication and strategy indicators. It is worth nothing that the matters confirmed through confirmatory factor analysis were also mentioned in international investigations that aimed to build rating scales for services provided by gyms [5][6][7][8][9][10][11][12][13][14] .

Conclusions
These results allow concluding that the Rating Scale for Quality of Services Provided by Gyms (QUASPG) presented acceptable psychometric values. The instrument was composed of 21 questions distributed into Environment (8 items), Management (8 items) and Marketing (5 items) dimensions.
After the many stages for building the instrument, it was possible to observe that the items about client stimuli, as well as gym planning, administration and innovation, were excluded during the study phases. Such aspects may be related to the users' personal options and to related factors that are of gym managers' interest specifically.
Thus, the analysis matrix final model presented three dimensions, namely Environment, Management and Marketing, in addition to six indicators, with them being Structure, Cleanliness, Service, Qualification, Communication and Strategy, comprehending all criteria considered important for the quality of the services provided. Therefore, there is a concern about workout environment, how the gym is managed, and interpersonal relationship between the gym and its clients. Moreover, other relevant aspects include organization and distribution of objects, furniture, spaces, equipment, space cleanliness, service, professionals' knowledge and experience, provision of information to clients, and development of good relationships with those involved with the gym.
The instrument seeks to assess the quality of services provided by gyms, considering the services and products provided by these companies in the current market. Thus, it is evident that constant market changes, in addition to regional characteristics, can change users' perceptions and feelings about the services provided by gyms. Hence the importance of developing exploratory studies in order to enhance the instrument model proposed.