SAJEMS NS Vol 4 (2001) No 1 99 The Measurement of Consumer Satisfaction with Selected Elements of the Total Retail Experience: An Exploratory Study of Fast Food and Supermarket Retailers N S Terblanche Department of Business Management, University of Stell en bosch C Boshoff Department of Business Management, University of Port Elizabeth ABSTRACT Although attempts have been made to identify some of the dimensions of retail shopping experience, these have been largely fragmented and uncoordinated. No attempt has yet been made to combine the efforts of many retailing students into a comprehensive model that accurately describes the total retailing experience. Also, very little is known about the relationship between the individual dimensions of retail shopping and customer satisfaction. This study attempts to reduce this gap in South African retailing literature by, first modelling the total retailing experience and, then, assessing the influence of selected individual retailing dimensions on customer satisfaction. It also investigates whether the impact of these dimensions of the retailing experience differs between fast food restaurants and supermarket retailers. The empirical results suggest a fairly consistent pattern of relationships between fast food restaurants and supermarkets. JEL L 81 1 INTRODUCTION The retail environment worldwide is in a constant state of adjustment and adaptation. Although competition, emanating from new entrants and formats as well as foreign sources, contributes to this constant flux, consumers also exert a major influence on this state of affairs. Present-day consumers not only have more choices than ever before, they also have more information at their disposal than ever before. In addition, increasing pressure of time may decrease the propensity to remain loyal to a retailer. Consumers often also assume that various rights and choices are owed to them (Schriver, 1997). Against this R ep ro du ce d by S ab in et G at ew ay u nd er li ce nc e gr an te d by th e P ub lis he r (d at ed 2 00 9) . 100 SAJEMS NS Vol 4 (2001) No I background, many retailers attempt to differentiate themselves by implementing loyalty schemes to attract and retain customers. Another avenue open to retailers to differentiate themselves, is to deliver exceptional service quality as part of their retailing strategy (Berry, 1986; Hummel & Savitt, 1988; Reichheld & Sasser, 1990). The marketing literature of the past decade contains many pages devoted to service quality, especially to SERVQUAL, the instrument developed to measure service quality (Parasuraman, Zeithaml & Berry, 1988). Most of the studies using or assessing the SERVQUAL instrument were conducted in fairly "pure" services settings like banking, credit card services and securities brokerage. Because a retail store experience involves activities such as browsing, price comparisons, search for merchandise, evaluating product variety and quality, and interaction with store personnel, the SERVQUAL instrument in its original form is not suitable to capture the unique blend of merchandise and service that makes up a retailing experience. Recently Dabholkar, Thorpe and Rentz (1996) proposed an instrument based on SERVQUAL, which they suggest measures service quality in a retailing environment. Apart from the common dimensions that are likely to be shared by pure service environments and retail environments, this instrument also captures additional dimensions of retail service quality unique to the retail environment. The development of the instrument is justified on the grounds that current measures of service quality do not adequately capture customers' perceptions of service quality for retail stores (i.e. stores that offer a mix of goods and services). Support for this view, inter alia, emerged from the findings of Finn and Lamb (1991) that tested SERVQUAL in four different types of retail store and did not find a good fit to the five-factor structure. They therefore concluded that SERVQUAL, without modification, could not be used as a valid measure of service quality in retailing. Yet another alternative available to retailers, with which they can differentiate themselves, is to provide a positive total retail experience (TRE) for their clients. Our contention is that in retailing where a mix of goods and services is offered, the focus should rather be on the management of the controllable elements of the .. total retail experience, as opposed to a perspective restricted to the management of service quality only. When considered in this way, service quality is only a component of the consumer's total retailing experience. 2 THE TOTAL RETAIL EXPERIENCE: A BRIEF OVERVIEW OF THE LITERATURE Berman and Evans (1998: 19) defme total retail experience as all the elements that encourage or inhibit consumers in their contact with the retailer. The total R ep ro du ce d by S ab in et G at ew ay u nd er li ce nc e gr an te d by th e P ub lis he r (d at ed 2 00 9) . SAJEMS NS Vol 4 (2001) No I 101 retail experience is influenced by two groups of components. The first group consists of the non-controllable components. These include aspects such as adequacy of street parking, timing of deliveries from suppliers and taxes, all of which are not controllable by the retailer. The controllable components, in tum, consist of a variety of elements that the retailer can indeed control, like the number of salespeople on the floor, variety of brands carried and stock on hand. In this study the emphasis is on the controllable elements of the total retail experience. The retail literature suggests that the dimensions of the controllable components may be grouped under seven dimensions. A brief description of the controllable components of the total retail experience follows below. 2.1 Personal interaction as a dimension of the total retail experience Loyal customers are the backbone of any retailer's business (Schriver, 1997: 20- 21; Reicheld, 1993: 68). It is thus in the interest of a retailer to enter into and maintain long-term relationships with customers. The personal interaction element of the total retail experience offers numerous opportunities for creating and fostering long-term bonds with customers. Some of the typical activities that are suitable for personal interaction are selling and fmalising transactions, providing information, reducing perceived risk, providing accompanying services and handling complaints (Davidson, Sweeney & Stampfl, 1988: 143; Speer, 1996: 13-14). . 2.2 Tangibles as a dimension of the total retail experience The tangibles that are elements of the total retail experience, are also components of the overall image that a retailer wishes to project to consumers. Tangibles include such aspects as physical facilities, transaction documentation, and appearance of personnel and materials that enhance the service delivery and communication of a retailer (parasuraman, Berry & Zeitharnl, 1991 : 41 ; Dabholkar, Thorpe & Rentz, 1996: 6-7). The tangible physical environment and atmosphere can assume a variety of roles in the marketing and management of a retailer (Bitner, 1992: 67). Insights and evidence from the environmental psychology literature support the notion that physical surroundings can influence the attitudes as well as the behaviour of consumers in a retail shop (Donovan & Rossiter, 1982; Darden, Erdem & Darden, 1983; Ridgway, Dawson . & Bloch, 1990). Tangible store attributes can also be seen as the "means" by which a consumer is able to achieve a desired "end", like a satisfYing total retail experience (Kerin, Jain &Howard, 1992: 381). R ep ro du ce d by S ab in et G at ew ay u nd er li ce nc e gr an te d by th e P ub lis he r (d at ed 2 00 9) . 102 SAJEMS NS Vol 4 (2001) No 1 2.3 Perceived product quality as a dimension oftbe total retail experience Perceived product quality has been identified as a detenninant of a consumer's perceived value of a product (Dodds, Monroe & Grewal, 1991; Kerin, Jain & Howard, 1992: 382) and will therefore have an influence on the TRE of a consumer. Various researchers have also reported on the importance of the quality of products (merchandise) in consumers' perception and evaluation of retail shops (Dabholkar, Thorpe & Rentz, 1994: 4). In a more recent study Sirohi, Mclaughlin and Wittink (1998: 240) operationalised product quality for a supermarket with fifteen indicators. For some products their quality level is of great importance to the consumer, and some retailers position themselves mainly on this basis. Retailers selling high quality, high priced, well-known luxury brands are examples of this. 2.4 Product variety and assortment as a dimension of tbe total retail experience Product variety and assortment are components of the conventional retail marketing mix (Hasty & Reardon, 1997: Il). Products are probably the most complex expectation customers have of retail shops, because they expect to find a variety or a selection of different kinds of products consistent with personal shopping intentions and preferences (Davidson, Sweeney & Stampfl, 1988: 141). The range of food and other product lines that supennarkets, for instance, offer, illustrates the complexity of the product variety decision. Consumers not only expect to find a variety of product lines, but also an assortment of different colours, brands, styles, models and sizes for each line. Product variety and the depth of assortment will to a large extent be influenced by the image the retailer wishes to project. For example, a supermarket's product variety and assortment, is associated with the customer services and facilities of the supermarket (Bishop, 1984; Doyle, 1984). Kerin, Jain and Howard (1992: 385) also reported that previous research and focus groups found overall variety and assortment to be indicators of the shopping experience. The importance of assortment and variety in retailing is also confinned by the fact that one of the items in the . Index of Consumer Sentiment Toward Marketing deals with the adequacy of the selection of products offered by retail shops (Gaski & Etzel, 1986: 79). 2.5 Store environment as a dimension of tbe total retail experience Store environment includes all the elements that contribute towards a pleasant shopping atmosphere. These elements are shop layout, aisles that make it easy to shop, store cleanliness, well-spaced product displays and attractive decor. The retail literature suggests that customers value the convenience that physical aspects, such as store layout, contribute to shopping (Gutman & Alden, 1985; R ep ro du ce d by S ab in et G at ew ay u nd er li ce nc e gr an te d by th e P ub lis he r (d at ed 2 00 9) . SAJEMS NS Vol 4 (2001) No 1 103 Hummel & Savitt, 1988; Mazursky & Jacoby, 1985; Oliver, 1981). Dabholkar et 01. also found empirical support for the contribution of shop layout to shopping convenience (1996: 7). Dabholkar et 01. (1996: 6) again treated store appearance and store convenience as two sub-dimensions of the physical aspects in their study. The marketing literature supports the notion that consumers use the physical environment to form attributions about service failures, and base purchase decisions on inferences they make from various elements of the physical environment (Bitner, 1990; Ward, Bames & Bitner, 1992; Bloom & Reve, 1990; Zeithaml, 1988). 2.6 Product prices as a dimension of the total retail experience Gaski and Etzel (I986) used price as a component in their Index of Consumer Sentiment Toward Marketing. This stresses the importance of prices in consumer decision-making. In the absence of other cues, consumers tend to use price as an indicator of product quality (Davidson, Sweeney & Stampfl, 1988: 143). Price is a generally controllable extrinsic cue of a product that has an influence on the way shoppers perceive product quality (Sirohi et 01., 1998: 226). Price as an extrinsic cue is, for example, most relevant in a supermarket, as consumers do not regard the time and effort spent in evaluating intrinsic cues as significant (Sirohi et 01., 1998: 227). Price is therefore one of the extrinsic cues that has received a great deal of attention in research (Dodds, Monroe & Grewal, 1991; Mazursky & Jacoby, 1985; Rao & Monroe, 1989). Kerin, Jain & Howard (1991: 383) suggest that because of the variety of extrinsic cues available in a supermarket (e.g. Cleanliness, assortment and variety), price and quality perceptions could co-vary and that consumers "get what they pay for". 2.7 Policy as a dimension of the total retail experience Policy as a dimension of the total retail experience captures those elements that are influenced by a shop's responsiveness to the customer's needs (Dabholkar et 01., 1996: 7). Typical policy elements would apply to areas of shop operations such as the return or exchange of purchases, shopping hours, payment options available and a system or process to deal with customer enquiries and/or complaints (Westbrook, 1981; Mazursky & Jacoby, 1985; Dickson & Maclachlin, 1990; Dickson & Albaum, 1977). The way in which a retailer responds to a service failure is essential for the maintenance of customer loyalty (Berry & Parasuraman, 1991) as well as a positive TRE because a service failure and the subsequent recovery efforts create strong memories for customers (Zeithaml & Bitner, 1996: 126). R ep ro du ce d by S ab in et G at ew ay u nd er li ce nc e gr an te d by th e P ub lis he r (d at ed 2 00 9) . 104 SAJEMS NS Vol 4 (2001) No I 3 OBJECTIVES OF THE STUDY The main objective of this study was to compare an "overall moder' based on data collected from 2063 customers of four retail sectors, namely clothing, hardware. supermarkets and fast food, with two identical models based on only the data from the fast food sub-sample (635 cases) and the supermarket sub- sample (665 cases). This comparison tried to assess the stability of the model. It was believed that if the three models returned similar fit indices in relatively diverse retail environments such as supermarkets and fast food, this would represent strong evidence of generalisability. The following propositions were considered: PROPOSITION 1: The model fit indices of the total sample, the fast food sample and the supermarket sample will not differ substantially. A second objective was to assess whether the three dimensions of the total retail experience modelled in this study as predictors of customer satisfaction would impact on customer satisfaction to a similar extent. The following proposition was considered: PROPOSITION 2: The path coefficients measuring the influence of the three dimensions of the total retailing experience (personal interaction. physical cues and product variety and assortment) on customer satisfaction, will be the same for the total sample, the fast food sample and the supermarket sample. The theoretical model that was tested empirically is illustrated in Figure I. R ep ro du ce d by S ab in et G at ew ay u nd er li ce nc e gr an te d by th e P ub lis he r (d at ed 2 00 9) . SAJEMS NS Vol 4 (2001) No 1 105 Figure 1 The theoretical model R ep ro du ce d by S ab in et G at ew ay u nd er li ce nc e gr an te d by th e P ub lis he r (d at ed 2 00 9) . 106 SAJEMS NS Vol 4 (2001) No I 4 METHOD 4.1 Generation of the questionnaire items Despite the criticism levelled against SERVQUAL (Buttle, 1995), the items used for the measurement of the various dimensions of service quality, are also regarded as suitable to measure some of the controllable elements of the total retail experience. For example, physical facilities, equipment, materials, business hours and various aspects of staff behaviour are covered by the items of SERVQUAL. The items of four of the SERVQUAL dimensions, namely responsiveness, reliability, empathy and assurance, were used in this study to measure the personal interaction component of the total retail experience (Kerin, Jain & Howard, 1992; Zeithaml, 1988; Dabholkar, Thorpe & Rentz, 1996; Sirohi, Mclaughlin & Wittink, 1998; Baker, Grewal & Parasuraman, 1994; Ghosh & McLafferty, 1987). SERVQUAL has proved to be an adequate measure of the reliability, responsiveness, empathy and assurance dimensions of service delivery. All of these dimensions are of prime importance in the interaction between customers and shop personnel (Parasuraman, Berry & Zeithaml, 1991: 41). The product variety/assortment and product quality dimensions were measured with self-generated items and published scales. The three product quality items suggested by Finn and Kayande (1997) were supplemented with two self- generated items. Product variety and assortment were measured with a five-item instrument. The tangible components involved in service delivery were measured with the original four items of SERVQUAL. All items were linked to a 7-point Likert-type scale. Satisfaction was measured by means of the following four items: Will you continue to buy from XXX in the future? How do you rate the overall service of XXX ? How often have you complained about service at XXX ? How often have you returned unsatisfactory products to XXX ? 4.2 Sample and data collection The procedure followed in this study was a combination of judgement, convenience and random sampling. Four retail industries were selected on a judgmental basis, namely fast food, clothing, supermarkets, and hardware stores. These industries were selected because they offer a combination of both service and physical products, and their market offer cannot be described as a "pure" service, independent of any other service component. Two organisations in each industry were then selected on a convenience basis and invited to participate in R ep ro du ce d by S ab in et G at ew ay u nd er li ce nc e gr an te d by th e P ub lis he r (d at ed 2 00 9) . SAJEMS NS Vol 4 (2001) No 1 107 the study. Individual respondents (customers) to be interviewed were selected on a simple random basis. Personal interviews, using a structured questionnaire, were conducted with customers of firms in these four industries, on the retailer's premises following a shopping visit. The interviews were conducted over a period of a month to include all the various types of clients in terms of time frequenting the retailers as well as cash and account customers. For example, it was necessary to conduct interviews at night-time with clients from the fast food retailers. A total of2 063 questionnaires were completed. 4.3 Scale purification The first step in the data analysis process was to assess how many dimensions the data contained. For this purpose, a Maximum Likelihood Exploratory Factor Analysis was done, specifYing a Direct Quartimin oblique rotation (Jennrich & Sampson 1966) of the original factor matrix. Several different factor solutions were considered. The most interpretable factor structure (factor loading exceeding 0.04 and no crossloadings) to emerge, was a 3-factor solution. All the factors in the 3-factor solution had Eigen values above 1.00 and a sufficient number of items loading on them to a significant (0.40) extent (Hair, Anderson, Tatham & Black, 1998). Table I shows the sorted rotated factor loadings matrix. The three factors that emerged were named personal interaction (measured by 3 assurance, 4 empathy, 2 reliability and 3 responsiveness items of SERVQUAL), physical cues (measured by I reliability and 2 tangibles items of SERVQUAL and 4 quality items) and variety (represented by 5 variety items). The items that did not load to a significant extent on a separate factor were excluded from subsequent statistical analysis. The remaining 24 items were then subjected to a reliability analysis to assess the internal consistency of the instrument. All three factors, as well as the overall instrument returned Cronbach alpha coefficients above the 0.7 level suggested by Peterson (1994). The reliability results are summarised in Table 2. The theoretical model depicted in Figure 2 was then subjected to empirical assessment. The structural equation modelling capabilities of the computer programme RAMONA (Browne & Mels, 1992) were used for this purpose. R ep ro du ce d by S ab in et G at ew ay u nd er li ce nc e gr an te d by th e P ub lis he r (d at ed 2 00 9) . 108 SAmMS NS Vol 4 (2001) No 1 Table 1 Sorted Rotated Factor Loadings Factor 1 Factor 2 Factor 3 personal physical product variety interaction cues and assortment PERS 1 0.859 -0.024 -0.047 PERS2 0.826 -0.080 0.022 PERS3 0.798 -0.022 -0.032 PERS4 0.758 0.050 -0.067 PERS5 0.751 -0.048 0.038 PERS6 0.740 -0.006 0.041 PERS7 0.731 0.035 -0.016 PERS8 0.662 0.117 -0.044 PERS9 t- 0.626 -0.024 0.227 0.614 0.110 I§ PERS 11 0.578 0.082 O. PERS 12 0.458 0.281 O. PHYS 1 -0.087 0.826 PHYS2 -0.058 0.810 O. PHYS3 0.024 0.770 0.039 PHYS4 0.157 0.539 -0.014 PHYS5 0.075 0.448 0.155 PHYS6 0.272 0.444 -0.042 PHYS7 0.148 0.400 0.117 VARlI -0.007 0.086 0.805 VARl2 0.042 -0.113 0.793 VARl3 -0.046 0.099 0.660 VARl4 0.038 0.118 0.659 VARl5 0.146 0.166 0.434 EIOEN 6.196 2.940 2.486 Table 2 Reliability Analysis Results: Total sample Dimensions oc Personal interaction .937 Physical cues .863 Variety .857 Consumer satisfaction .725 Overall .948 R ep ro du ce d by S ab in et G at ew ay u nd er li ce nc e gr an te d by th e P ub lis he r (d at ed 2 00 9) . SAJEMS NS Vol 4 (2001) No 1 109 Table 3 Fit Indices Total Sample Fast Food Supermarket Chi-square (df) 2243.49 (249) 1473.8 (344) 1240.5 (344) RMSEA I 0.062 0.072 0.063 GFI 0.900 0.846 0.882 AGFI 0.880 0.818 0.861 ECVI 1.566 2.520 2.056 Parsimonious Fit Indices PNFI 0.826 0.761 0.818 PGFI 0.816 0.770 0.803 NFl 0.908 0.837 0.899 5 FINDINGS When comparing different models, the parsimonious fit indices produced by most SEM programmes ought to be used (Hair Anderson Tatham & Black, 1998: 658). Hair et al. (1998) suggest that when comparing two models, the Parsimonious Normed Fit Index (PNFI) is particularly usefuL Differences between 0.06 to 0.09 are regarded as "indicative of substantial model differences". When comparing the PNFI of the fast food sample (model) with the supennarket model the difference is 0.057, and when comparing the total model with the supermarket model the difference is only 0.008 both less than the lower bound of the range for substantial model difference proposed by Hair et al. (1998). When comparing the total model with the fast food model, the difference is 0.065 - just inside the range proposed by Hair et al. The total model is thus not quite as good a predictor of customer satisfaction for a fast food restaurant as it is for a supennarket. It appears as if the model fares slightly better in retail environments where physical products are dominant in the market offering as opposed to services (the fast food industry). However, given the relatively small PNFI differences, Proposition 1 is accepted. The empirical results of the three empirical models (the total sample, and the fast food and supennarket samples) are depicted in Figures 2, 3 and 4. In the total sample (Figure 2) personal interaction (0.585; p