MEASURING GROCERY STORES SERVICE QUALITY IN INDONESIA: A RETAIL SERVICE QUALITY SCALE APPROACH

The growing number of modern grocery stores in Indonesia is a challenge for each grocery store to maintain and increase their number of consumers. The success of maintaining and improving service quality will affect long-term profitability and business sustainability. Therefore, in this study, we examined consumer perceptions of service quality in one of modern grocery stores in Indonesia. Data were collected from 387 consumers of grocery stores in Jakarta, Bogor, Depok, Bekasi, Cibubur, and Subang. Structural Equation Modeling (SEM) through Maximum likelihood and Bayesian estimation was employed to analyze the data. The finding indicated that the five indicators of the retail service quality scale consisting of physical aspects, reliability, personal interactions, problem solving and policies provided valid multi-item instruments in measuring consumer perceptions of service quality in grocery stores.


Introduction
Lifestyle changes and demands of a more efficient and fast shopping process lead to changes in consumer shopping transactions from traditional markets to modern markets.Modern grocery stores are preferred, especially by upper middle class people and living in urban areas.This is because consumers do not only want quality products at competitive prices, but also the quality and convenience of the shopping process.In addition, the provision and arrangement of various products in one place make the efficiency of shopping time can be achieved (Torlak, Uzkurt, & Özmen, 2010); (Canada, 2011); (Terano, binti Yahya, Mohamed, & bin Saimin, 2015).The higher consumer preference for shopping in modern grocery stores makes this business competition even higher (Sirohi, McLaughlin, & Wittink, 1998).In addition, the liberalization of the government since 1998 through the signing of a letter of intent with the IMF provides a great opportunity for foreign investors to open a modern retail business in Indonesia (KPPU, 2007).This has led to the growing brand of modern grocery stores in Indonesia.The growing number of modern grocery stores is a challenge for each grocery store to maintain and increase their number of consumers.The success of maintaining and increasing the number of consumers will affect the level of corporate profits and longterm business sustainability that will contribute to the national income and also contribute to absorbing the national workforce (Soliha, 2008).The fundamental strategy that modern grocery stores applied to maintain and increase their customers is by improving service quality (Hummel & Savitt, 1988).According to (Vazquez, Rodrı ǵuez-Del Bosque, Dı áz, & Ruiz, 2001), service quality is the consumer's perception of the overall service quality they receive and adjust the effort to achieve the subjective service quality with the cost incurred by the service provider.In this study, examined consumer perceptions of service quality in one of the modern grocery stores in Indonesia.Prior research on service quality has been widely practiced, but there is still little evidence in term of service quality in modern grocery stores (Lewis & Thomas, 1990); (Vazquez et al., 2001); (Siu & Chow, 2004); (Torlak et al., 2010).In addition, the differences in the method of service quality analysis, creating different contribution of each study to the literature.

Service quality in modern grocery stores
Various strategies are carried out by modern grocery stores to maintain and to enhance consumer perceptions of their service qualities.Some of them offer a large and diverse range of products, while others offer a variety of pricing policies (Levy, Weitz, & Grewal, 1998), attractive and convenient store interior conditions to increase consumer time around (Andreu, Bigné, Chumpitaz, & Swaen, 2006) and provide shop assistants who are ready to help and promote certain products (Torlak et al., 2010).(Siu & Chow, 2004) analyzed service quality in Japanese grocery retailing in Hong Kong by using a retail service quality scale consisting of physical aspects, reliability, personal interactions, problem solving and policies, following (Dabholkar, Thorpe, & Rentz, 1996).The findings indicated that personal interaction and trustworthiness was the most influential service quality indicator.Moreover, (Vazquez et al., 2001); (Ahmad, Ihtiyar, & Omar, 2014); (Ibrahim et al., 2013) also analyzed service quality by using the same indicators.According to the prior studies, the indicators of the retail service quality scale were more extensive in capturing service quality in retailing stores compared to SERVQUAL indicators used in companies that offer services in general (Vazquez et al., 2001).The first indicator of retail service quality scale proposed by (Dabholkar et al., 1996) was physical aspects.This indicator embodied a broader understanding than tangible indicators in SERVQUAL which consists of physical appearance, store layout, design, and product shelf arrangement.The second indicator was reliability.This indicator consisted of the ability of grocery stores to keep their promises and to set services rightly.The third indicator was personal interaction.This indicator embodied the ability of personnel to be courteous, to be helpful and to treat customers in a good manner.The fourth indicator was problem solving.This indicator embodied the ability of employees to handle returns and exchanges and to resolve customers' problems and complaints.The last indicator was policies.These indicators indicated the specific policies on retail stores on products quality, parking, operation hours, and method of payment, pricing strategy and brand setting (Dabholkar et al., 1996); (Vazquez et al., 2001); (Abu, 2004); (Torlak et al., 2010).All of those indicators cannot be measured when we analyze service quality in retail stores by using SERVQUAL.More details about the indicators of the retail service quality scale used in this study are denoted in Appendix A1.

Data and measurement
To analyze service quality in grocery stores, service quality indicators (Dabholkar et al., 1996) were adapted.The indicators consisted of physical aspects, reliability, personal interactions, problem solving and policies.A face to face interview was conducted with 387 consumers of grocery stores in Jakarta, Bogor, Depok, Bekasi, Cibubur, and Subang.The questionnaire items were measured by using a Likert scale.The data were analyzed by using Structural Equation Modeling (SEM) method and processed by using Amos 24 statistical package through Maximum likelihood and Bayesian estimation technique.

Characteristics of respondents
The result of descriptive analysis in table 1 indicated that the majority of local grocery stores consumers was female (53.23%) and the rest was male (46.77%).Then, judging from the frequency of shopping, as many as 97% of respondents said they had carried out shopping more than 3 times and 92% of respondents said they had been shopping monthly.Thus, it can be concluded that almost all respondents in this study were consumers who were loyal to the grocery stores.The result of parameter estimation using maximum likelihood The results of descriptive statistics on the indicators used in the model were denoted in table 2. Furthermore, the assumption of univariate normality of the data indicated that, critical skewness value (c.r) of all indicators was between the values of -2,58 ≤ c.r ≤ 2.58.however, the coefficient of multivariate kurtosis was 84.280 ( > 2.58 ).Thus, it could be concluded that based on multivariate normality assumption, the data were not normally distributed (Ghozali, 2008).
The evaluation of maximum likelihood estimation on goodness of fit model indicated that Chi-square value = 292.904,DF = 109, RMSEA = 0.066, GFI = 0.919, PGFI = 0.654, CFI = 0.922, PCFI = 0, 739, and TLI = 0.903.Therefore, the overall indicators indicated that the model has reached conformity (Arbuckle, 2009;Browne & Cudeck, 1993;Ghozali, 2008;Moss, 2009).All indicator coefficients were positive and significant at alpha 5% (Table 3).All standardized loadings were > 0.50 and C.R. value > 2.00 indicated that all indicators of latent variables employed in the model were valid and significant at alpha 1%.The main indicator of the physical aspects was Q4, i.e. facilities and services make consumers comfortable to shop (std.loading = 0.825).The main indicator of reliability was Q8, i.e.The price written on the label was consistent with the cashier's payment (std.loading = 0.821).The main indicator of personal interaction was Q18 (std.loading = 0.650), i.e. store employees worked on the best service.The main indicator of problem solving was Q20 (std.loading = 0.754), i.e. cashiers in the store had a reliable abilities.The main indicator of policies was Q23 (std.loading = 0.551), i.e. adequate parking space available in store.The relationship between each latent variable was displayed in figure 2.

Figure 1. The relationship between the indicators of the variable
Since the multivariate normality assumption was not confirmed in the data, then the estimation was continued by employing the Bayesian estimation technique.This technique is one alternative that can be used on data that is not normally distributed (Ghozali, 2008).

Estimation of parameters using Bayesian confirmatory factor analysis (CFA)
The estimation results using a Bayesian SEM technique were converging on an iteration of the sample number of 63,000 (500 + 62,500).The indicators of Q5, Q11, Q18, Q20, and Q24 did not appear on the output because the model was given constraint 1 (Ghozali, 2008).All indicators were significant at alpha 5%.This could be inferred from the 95% credible internal lower bound and 95% credible internal upper bound that did not contain value of 0 (Table 4).The validity of each dimension was demonstrated through the posterior distribution of first and last of each loading factor, where we expect the distribution of loading factor did not contain the value 0 (Figure 2 through figure 13).The relationship of each indicator with the latent variable is denoted in table 5.The relationship formed between each indicator with the latent variable was the direct relationship only, without any indirect relationship.The effect of Q19 on problem solving was 0.640, while the effect of Q20 was 0.761.The effect of Q20 was greater than the effect of indicator Q19.This result was consistent with the results from maximum likelihood estimation.Furthermore, the greatest effect of the indicator of the personal interaction was given by Q18 (0.657).In reliability, Q10 had the highest effect (0.819).In physical aspects, Q4 had the highest effect (0.825), and finally in the policies, Q23 had the highest effect (0.547).All inter-dimensional relationships with each latent variable were consistent with the maximum likelihood estimation results.

Discussion
The empirical findings in this study indicated the importance of each indicator in giving influence to the latent variables.In the physical aspects, the most important indicator affecting consumer perceptions of grocery stores physical condition was the ability of facilities and store services in providing convenient shopping for consumers (Q4).In the reliability, the most important indicator influencing consumer perceptions on the reliability of grocery stores was the products sold were always fit in size (Q10) and the price written on the label is consistent with the cashier's payment (Q8).These two indicators indicated a huge consumer interest in capabilities of stores in maintaining consumer confidence in the size and price of their products.In the personal interaction, indicator of store employees worked on the best service was dimension which most influence consumer perception of personal interaction (Q18).In the problem solving, the indicator of cashiers in the store had a reliable ability was the indicator that most affected consumer perceptions of the ability of problem solving in grocery stores (Q20).Finally, in the policy variables, the indicator of adequate parking space available in stores was the indicator that most affected the consumer's perspective on grocery store specific policies (Q23).Providing a large and convenient parking space would influence consumers' desire to shop at grocery stores because the majority of consumers were middle to upper income groups who brought their own vehicles when shopping.Overall, all of the above results are in accordance with the retail service quality scale as proposed by (Dabholkar et al., 1996) and proven empirically by (Abu, 2004); (Torlak et al., 2010); (Vazquez et al., 2001); (Ahmad et al., 2014).The implications for store managers are to maintain and improve services and facilities in stores that make consumers comfortable for long shopping.This convenience is positively related to consumers' desire to increase the amount of goods and services into their shopping lists (Glanz & Yaroch, 2004).In addition, the fit product size, the exact weight of fresh products and the appropriateness of the price on the shelf with the price at the checkout is a must-pay task for the store managers.To maintain this issue, the need for strict training and supervision of store employees who handle these tasks is a requisite.In addition, all store employees must also be prepared to always pay attention to personal relationships with consumers.Since cashiers become a resource that is considered important in overcoming consumer problems, the placement of human resources to be placed as a cashier requires serious attention by managers.In addition, the cashier needs to be trained to handle problems during long queues, as well as the ability to complete consumer payment transactions efficiently.Finally, store managers need to add locations for parking.This can be an important consideration as well when the store wants to open branches elsewhere.The availability of a large parking lot is one dimension that ensures customer satisfaction.

Conclusions and suggestion for future studies
The empirical results from this study indicate that the five indicators of the retail service quality scale provide a valid multi-item instrument in measuring consumer perceptions of service quality in grocery stores.However, some suggestions for further research are differentiating consumer perceptions based on demographic and store location differences.In addition, adding questionable items in the questionnaire consisting of consumer perceptions of service quality and the actual performance of service quality is suggested to measure the importance of each latent variable of a retail service quality scale.

Figure
Figure 2. Posterior distribution of Q4