SERVQUAL Model with Interrelated Dimensions. The Case of The Pharmacy in Mexico

: This research is focused on understanding the interrelation between the different dimen- sions of the SERVQUAL model in the pharmaceutical area, which have differences in per- ceptions of service domains depending on the cultures in which they are immersed. To study the proposed relationships, we collect data of 100 pharmacy customers of Mex- ico and use SEM to test the relationship between SERVQUAL dimensions, purchase fre- quency, gender, and quality perception. Our results show that the SERVQUAL dimensions are interrelated, SERVQUAL dimen- sions and gender predict the perception of quality but not the purchase frequency. These results provide key information in order to understand in a better way the service quality in the pharmaceutical area. We develop a theoretical framework that is based on interrelationships between the dif- ferent dimensions of SERVQUAL, assuming that they are not independent of each other. This way of understanding SERVQUAL dimensions can provide relevant information to un- derstand better how individual perceptions affect the service quality in the pharmaceutical area.


Introduction
The quality of service perceived by users in health is an important concept for organizations and constitutes a measure of performance obtained from the organizational strategies carried out to improve service systems (Tzeng and Chang, 2011).
Service quality is a classic area of research in management and marketing science. However, its specific application to the pharmaceutical area is low. This is because the research in dimen-sion of pharmaceutical service quality are few and not evaluated the impact of these dimensions. This leaves a great field to explore due to the characteristics and effects that the consumption of the different products that a pharmacy sells can cause to users, public administration and society (Rojas et al., 2018).
Understanding the quality of service is key to being able to improve the management within the organizations. Grönroos (2020) points out that quality of service is the result of an evaluation process where the consumer compares their expectations with the perceived service when they receive it. Due to the amplitude and generalization of this definition, several authors have tried to develop a consistent instrument that can measure this variable. However, because of its complexity, there is no general agreement, and over time this variable has been interpreted in different ways (Um, 2018). In health, the SERVQUAL model is structured in functional quality (Fatima et al., 2018). In this context, the client's perceptions about the quality of the service result from a comparison of their pre-service expectations with their actual service experience. Based on this perspective, different scales have been developed for the measurement of service quality, which are mainly known as SERVQUAL (Haming et al., 2019). Five dimensions of service quality apply to organizations providing services in general (Shafiq et al., 2017): (i) tangibility, that is, physical facilities, equipment and staff appearance; (ii) fiability, that is, the ability to perform the promised service in a reliable and accurate manner; (iii) guarantee, that is, knowledge, courtesy and ability of employees to instill trust in the client towards the service provider; (iv) responsibil-ity, that is, the willingness to help customers and provide a fast service; and (v) empathy, that is, the provision of personalized and individualized customer service (Zeithaml et al., 2020). However, these five dimensions of SERVQUAL are generally assumed as orthogonal and independent constructs, whose relationships are exploitable through structural equation models (SEM), which up to now has implied the proposal of models without interrelations.
The aim of this study is to describe relationships between constructs, considering the quality of service as a latent dependent variable, determined by a sequence of interrelationships between the tangible elements that determine empathy, and finally security, as necessary elements to reach a response in the quality of service perceived by users in the pharmaceutical area. This paper is organized as follows: Section 2 is a literature review about of quality of service adapted to phar-maceutical area in the Mexican market. Section 3 exposes the proposed methodology, whereas Section 4 illustrates it with a real-world case study in the Mexican pharmaceutical area; and Section 5 provides discussions, conclusions to the results obtained in this study, as well as their limitations and future research.

Literature review
The service marketing literature devoted much attention to the relationship between perceived quality service and customer satisfaction as a component of "loyalty chain" , see Vatjanasare-gagul (2019) and Song and Qu (2017). These contributions have illustrated that the quality of service should be considered an attitude highly related to satisfaction, but not equivalent (Pansari and Kumar, 2017). The quality of the service can be defined as the degree and direction of the discrepancies between the perceptions of service delivery and the previous expectations of cus-tomers or users (Parasuraman et al., 2002). Currently, most researchers agree that quality service is a record of customer satisfaction (Zeithaml et al., 2020).
The use of different instruments such as SERVQUAL, which measures the gaps between expectations and reality, has helped establish explanatory models based in dimensions of quality service related to the client's or user's observations and how the client or user finally receives the service delivered (Patel, 2014). In this way, Dehghanpouri et al. (2020) has proposed models that begin with the SERVQUAL measurement scale, consisting of dimensional structures or constructs: tangibles, reliability, responsiveness, assurance, and empathy for measure service quality. Furthermore, this type of conceptual models of the relationship between service quality and customer satisfaction are developed based on the literature and serve to explore various hypotheses. The above is possible under the approach of SEM, wich emerges as a result of the conjunction of two traditions. On the one hand, an econometric perspective focused on prediction, on the other, a psychometric approach that models concepts as latent (unobserved) variables that are indirectly inferred from multiple observed measures (indicators or manifest variables), see Tarka (2018) and Hair et al. (2019). However, generally this dimensions are considered as not correlated with each other, or without interdependence (Jones and Shandiz, 2015).
In the pharmaceutical area, the issue of measuring service quality has been approached from different perspectives. For example, Parmata (2016) proposes a measurement of perceived qual-ity, using a valid scale with four dimensions (reliability, guarantee, responsiveness and com-munication) adapted from SERVQUAL. In other study Maidana et al. (2018), have shown the reliability and validity of the content, the criteria and the conformation of constructs for adap-tations of the SERVQUAL dimensions to the pharmaceutical context of self-services. There is evidence that in the pharmaceutical context, not all dimensions of SERVQUAL tend to have the same importance in determining the quality of service perceived by the user (Pereira et al., 2019). Moreover, it is very likely that there could be interrelationships between some of these dimensions (Khanjankhani et al., 2016).
In this study, we will try the following hypothesis: · Hypothesis 1: The dimensions of SERVQUAL in the pharmaceutical area have dif-ferent importance in determining the quality of service perceived by the user · Hypothesis 2: The dimensions of SERVQUAL in the pharmaceutical area have inter-dependence relationships, and together determine the quality of service perceived by the user Previous research suggests that only measuring the expectations and perceptions of the service are not sufficient to be able to determine consumer satisfaction, so the importance of individual attributes must be identified so that practices as well as management resources can be oriented and properly assigned (Abu-Salim et al., 2017). In this context, the same author has shown that men tend to prefer to classify marketing-related stimuli to a more general level of abstraction compared to women. Women tend to focus more on differentiation points, so they will need to use a greater number of descriptors when evaluating products and services. The above, for example, makes women prefer a higher level of personalization in banking services compared to men. In other context, Hwang et al. (2019) found that women's expectations were significantly higher compared to men's, in four of the five dimensions of fast food service quality. The four dimensions with the highest expectations were: tangibility, reliability, sensitivity and empathy, while security expectations were very similar. Gender is a basic cultural difference related to the division of the sexual role in society that has often been overlooked in relation to the evaluation of services. Other study as Ojekalu et al. (2019), found that the gender and education of the occupants significantly influenced the perception of service quality. Women rated the quality of service of managers better compared to men. However, the age of the occupants did not significantly influence the perception of service quality. Other demographics variables such as age and region of residence have not shown consistently robust relationships with the quality of service perceived by users, see Fatima et al. (2018), Boah et al. (2020) and Ojekalu et al. (2019)). In the above context, Wajda and Hu (2004) have shown that men tend to prefer to classify marketing stimuli to a more general level of abstraction than women. Women will tend to fo-cus on differentiation points and will use a greater number of descriptors in their evaluation of products and services. This arguments make us propose the following hypothesis for the area of pharmaceutical services: · Hypothesis 3: The relationships of interdependence of SER-VQUAL dimensions in the pharmaceutical area and the quality of service perceived by the user, are different depending on the gender of the same The central strategy of relationship management of the organization is the ability of that orga-nization to develop and improve Long-term relationships with users and to satisfy current users. The main focus of such organizations is on user satisfaction and loyalty, that is, to retain their customers / users and generate repeated orders / services (Ozuem et al., 2017).
Patient satisfaction is important because it leads to a higher rate of patient retention and cus-tomer loyalty (Mamalis et al., 2020). A good deal of research has explored a variety of health care service quality dimensions that may influence patient satisfaction, such as continuity of care, physician expertise, the concern shown by the physician Journal of Technology Management & Innovation © Universidad Alberto Hurtado, Facultad de Economía y Negocios. 84 and other medical staff, and phys-ical facilities, see Wan et al. (2020), Xie and Or (2017), Zaid et al. (2020), however, to our best knowledge, we have not found studies on the relationships between the loyalty of users of a ser-vice, measured by indicators such as the rate of visit of a service and dimensions of perceived quality of service.
Study of Zaid et al. (2020), show that patient satisfaction varies according to the number of outpatient visits he has had. Regarding how this finding relates to SERVQUAL dimensions, the author finds that patients with a low visit rate value the quality of service based mainly on the dimensions of reliability and safety, while the group of patients with the most visit rate, They make based on the dimension of empathy.
This arguments make us propose the following hypothesis for the area of pharmaceutical services: · Hypothesis 4: Rate of visit of pharmaceutical service is a determinant in the rela-tionships of interdependence of SERVQUAL dimensions and the quality of service perceived by the user

Methodology
In a structural equation model, each unobservable construct (C) is related to a set of observable items or variables ({O1, O2, O3, ..., n }, through a linear combination of fix coefficients ({f1, f2, f3, ..., fn } added to their respective errors ({ϵ 1 , ϵ 2 , ϵ 3 , , ϵ n }. Mathematically this relation is expressed by: By assuming multivariate normality of the variables, the expected value of the term error is null, then (E(ϵ) = 0). By generating a matrix of standardized data (Z), the factor model can be expressed as a function of Z, as (1): with f as a factorial load vector, ϵ (0, ψ), where ψ = E(ϵ j ϵ j j ) for each individual (j) of the sample, and FS j is factorial score for j-th individual in the sample. Then, the factorial score of individual j, j = 1, 2, · · · , m is estimated as (2): for more detail see Thompson and Daniel (1996).
Since the service quality construct is built based on the difference between perceptions and expectations of various dimensions, we have chosen to consider its factorial score as an observ-able variable for each individual that is part of the sample under study. This approach can be consulted in Harrison-Walker (2002).
In this work, we adapt the items of classic SERVQUAL instrument of Parasuraman et al. (1988) to the pharmaceutical environment. Following Harrison-Walker (2002), we applied fac-torial analysis to define the dimensions more relevant in this area. In this way, and after pilot validation of the understanding of each item, we built the instrument shown in Table 1, which measures perceptions and expectations of the quality of service in pharmacy users. The subsequent factorial analysis of this group of questions helped us to define the three dimensions of perceptions and expectations of quality of service to study, namely: Tangibility, Fiability and Empathy. This reduction of the classical dimensions in this area is in agreement with the results of previous studies, see Harrison-Walker (2002). We validate all the latent constructs involved for validity and reliability. For this based on its results of confirmatory factorial analysis (CFA), we prove the construct validity using fitness indexes. Then, under a method of variance extraction with varimax rotation to ensure that each extracted factor covered the maximum of variance shared with the original variables that compose it. The criteria that gives account of the indicated areas are: own values > 1 for each factor, whose sum collects an inertia higher than 70%, with the minimum correlation between factors, confirming the statistical significance of factorial loads with magnitudes greater than 0.7, and discarding loads < 0.5. Each question was evaluated with an ordinal measurement scale of 1 to 5, with N/A = Not applicable/ Do not know. By taking a pilot sample of size 10 individuals, the reliability of the constructs indicated above for perceptions and expectations were estimated separately, examining the Cronbach's alpha statistic.
Subsequently, we study the correlations between the factorial scores of the constructs of per-ception for each dimension of tangibility (PT), fiability (PF) and empathy (PE), and the factorial score of service quality (Q), as well as the various interrelationships among them. We calcu-late average variance expected (AVE) to show the convergent validity. We also tested the dis-criminant validity by building the discriminant validity index summary. Finally, we show the reliability when calculating the composite realibity (CR) indicator. The values for AVE and CR should be above 0.5 and 0.6 respectively, see Kashif et al. (2016). Then we propose structural models to relate constructs and execute the estimation of parameters. Based on the results of the structural models, we test hypotheses and describe the results of the interrelationships between the constructs. We select the best model of interrelationships according to goodness of fit criteria around the data.
The reliability of each structural construct was analyzed by the Composite Realibity (CR) measure > 0.7, which was calculated from (3): (3) where f i and φ i are factorial loads and error variance of each factor to respect items i = 1,... , n of the construct to be evaluated, respectively. The discriminant validity is characterized by Aver-age Variance Extracted (AVE) given by (4): whose value must be greater than the square of the correlations product between each pair of constructs. To evaluate the fit of the proposed structural model, the real data matrix was verified with the hypothesized test by means of the (degrees of freedom) distribution test <3, the DFs were calculated according to 5: where n is the number of unobservable variables in the model and k, the number of unknown parameters to be estimated. Other adjustment measures of the model were evaluated, such as root mean square residue (RMSEA). After confirming the discriminant and convergent validity of the structural model, it was possible to use the scores of the sum of factors in each construct, in order to perform a regression analysis that allows us to validate the associations (sign and magnitude) between constructs. The processing of the structural equations was carried out with the AMOS Graphics, SPSS and R software.

Size sample and surveys application
In order to calculate the sample size of the survey application, we originally considered a margin of error of 5%, a 95% level of reliability, with a 50% prevalence ratio of the estimated variables and with a confidence interval of + -1.96 SD (standard deviation). These indicators were achieved with a total of 200 respondents. The study considered public and private pharmacies, including chains and independent located in diverse locations of Mexico. The data set can be consulted at Rojas and Coluccio (2021).

Results of real-world case study of Mexican pharmaceutical area
Tables 2, 3, 4, 5 and 6 show a characterization of the sample in type of pharmacy, location, gender, visits per month and age of the respondents, respectively.

Masculine 84
Female 116 Total 100 Table 5: Characterization of the sample. Number of visits per month of the respondents. We study the structural model to establish the inter-relationships between dimensions of per-ceptions of SERVQUAL in pharmaceutical area showed in Fig. 1. We considered a method of variance extraction with varimax rotation to ensure that each extracted factor covered the max-imum of variance shared with the original variables that compose it. In this scheme the latent variables that correspond to construct of Tangibility (T), Fiability (F) and Empathy (E), are represented by an oval, whereas the observable variables (PT1, PT2,...,PF1, PF2,...,PE1, PE2,...) that correspond to the questions of each construct are symbolized by rectangles. As we have indi-cated, in this study the factorial score of the quality of service (Q) corresponds to an observable variable, as well as the visit rate (Rate Visit in the figure 1). In this model the relations between the unidirectional variables are represented by an arrow. The measurement model was validated through a CFA in the AMOS Graphics and SPSS software. The criteria that gives account of the indicated areas are: own values > 1 for each factor, whose sum collects an inertia higher than 70%, with the minimum correlation between factors, confirming the statistical significance of factorial loads with magnitudes greater than 0.7, and discarding loads < 0.5. The CFA adjustment indicators were: Comparative Fit Index (CFI)= 0.92, Tucker Lewis index (TLI)= 0.87, RMSEA= 0.094, χ 2 (df=117)= 324.09. In addition, the indicators of average variance extracted (AVE) and Compound reliability (CR) are greater than 0.5 and 0.7 respectively, as shown in Table 8, granting a good adjustment of the model.

Number of visits per month Count
The weight regression estimates, standard error (s.e) and p-values of structural model are shown in Table 9. The intercept estimates, standard error (s.e) and p-values of structural model are shown in Table 10.
We evaluated the interrelationships between the perception constructs of tangibility (T), fi-ability (F) and empathy (E) with the factorial score of service quality (Q) and the rate of visit (Visit Rate) to the pharmacy in Table 11.
When working the two segments of gender separately, the results differ with the results ob-tained previously. When working with the female segment the same procedure indicated in the previous results, we can note that it has good adjustment rates but with a RMSEA a little higher than the allowed (CFI= 0.90, TLI= 0.79, RMSEA= 0.114; χ 2 (df=117)= 292.122). The weight regression estimates and the statistical significance are shown in table 12. The intercepts and the statistical significance are shown in table 13. As can be seen in table 14, the interrelationships between dimensions of perception of SERVQUAL and factorial scores of quality service and rate of visit to the pharmacy in this segment, have changes.

Discussion
This study proposes and tests a service quality model based on SER-VQUAL dimensions in the Mexican pharmaceutical area. Specifically, we develop a theoretical framework that is based on interrelationships between the different dimensions of SERVQUAL, assuming that they are not independent of each other. This way of understanding SER-VQUAL dimensions can provide relevant information to understand better how individual perceptions affect the service quality in the pharmaceutical area.
The study of the service quality and the different dimensions of SER-VQUAL were carried out in different types of Mexican pharmacies such as public, private, self-service, and hospitals. With these data, a SEM was made. In this model, the dimensions were treated as latent variables that interacted in the model. In this sense, our results provide us with valuable information that corroborates our study hypotheses, where the different dimensions interact with each other.
According to the results obtained using SEM, we can point out that the tangibility dimension predicts the fiability dimension. This relationship may be because clients, when they perceive an infrastructure and equipment by the pharmacy, are more likely to evaluate the service positively. Furthermore, the empathy dimension is predicted by the reliability dimension. This relationship can be explained by the fact that evaluating the service delivered in a better way can be related to more personalized and individualized service, responding directly to the client's needs. Finally, support was also found on the relationship between service quality and the dimension of empathy. Thus, all these relationships in the proposed study were significant and positive with a p < 0.01. These results indicate that the classical look where the dimensions are independent may need to be revised. In light of the results, we can see a dependency that can explain the clients' quality service evaluation. In this sense, future research should continue to focus on this interdependence of dimensions and thus be able to understand not only in the pharmaceutical area how these relationships can change and which ones are more important according to the service area that is analyzed.
Finally, our data provide relevant information on the relationship between the visit rate and gender and the perception of quality in the pharmaceutical area regarding the other proposed relationships. In this sense, our results show us that the visit rate is not related to the perceived quality. These results are contrary to what has generally been found in the literature since cus-tomers perceive a better service quality, repeat the service, and reward the organization for the service delivered, see Chow et al. (2007) . This relationship may not occur in the pharmaceu-tical sector because customers turn to pharmacies, not for pleasure but rather forcibly to buy a drug where this drug is the same in all pharmacies. Thus, this choice could generally be based on proximity to where you live or the price of the medicine, but not on the service's perceived quality.       Another interesting finding from this study is that the relationships described seem to change with the gender of the respondents. Previous SERVQUAL studies corroborate this finding, see Mensah and Mensah (2018). According to the results obtained, we can see that the intercept of the service quality factor scores is more negative for the female segment, which generally indicates that the difference between the perception and expectations of service quality is worse evaluated in this segment. The above indicates that this genre is more demanding in its evalua-tions of quality of service in the pharmaceutical area. Also note that the relationships between tangibility and fiability are lower in women, however, once this barrier is overcome, the relation-ships between reliability and empathy, and later between empathy and the factor scores of the quality service are higher in this gender, with respect to of men.