Assessing Consumers' Affective Responses to Retail Environments: a Tale of Two Simulation Techniques

ABSTRACT - Sixty-two student subjects rated either 12 slides of the interiors of various apparel stores (visual stimulus) or 12 in-store background music selections (audio stimulus) using a modified version of the Mehrabian and Russell pleasure/arousal/dominance scale for affective responses. The affect measures were used to predict subjects' perceptions of the visual or the audio stimuli on a number of store image variables (i.e., merchandise selection, prices, service quality, etc.). Results of several multiple regression analyses indicated that the relationships among the affect dimensions and the store image variables depended on whether subjects were in the visual versus the audio stimulus condition. In some cases the signs of the beta coefficients for a given affect dimension were even reversed. Yet, the means and standard deviations for each affect and store image measure did not significantly differ by condition, suggesting the results were due to something other than nonrepresentative samples of audio and visual stimuli.

Citation:

Charles S. Areni, John R. Sparks, and Patrick Dunne (1996) ,"Assessing Consumers' Affective Responses to Retail Environments: a Tale of Two Simulation Techniques", in NA - Advances in Consumer Research Volume 23, eds. Kim P. Corfman and John G. Lynch Jr., Provo, UT : Association for Consumer Research, Pages: 504-509.

Advances in Consumer Research Volume 23, 1996      Pages 504-509

ASSESSING CONSUMERS' AFFECTIVE RESPONSES TO RETAIL ENVIRONMENTS: A TALE OF TWO SIMULATION TECHNIQUES

Charles S. Areni, Texas Tech University

John R. Sparks, University of Dayton

Patrick Dunne, Texas Tech University

ABSTRACT -

Sixty-two student subjects rated either 12 slides of the interiors of various apparel stores (visual stimulus) or 12 in-store background music selections (audio stimulus) using a modified version of the Mehrabian and Russell pleasure/arousal/dominance scale for affective responses. The affect measures were used to predict subjects' perceptions of the visual or the audio stimuli on a number of store image variables (i.e., merchandise selection, prices, service quality, etc.). Results of several multiple regression analyses indicated that the relationships among the affect dimensions and the store image variables depended on whether subjects were in the visual versus the audio stimulus condition. In some cases the signs of the beta coefficients for a given affect dimension were even reversed. Yet, the means and standard deviations for each affect and store image measure did not significantly differ by condition, suggesting the results were due to something other than nonrepresentative samples of audio and visual stimuli.

Much research has been devoted to examining the affective dimensions of a store's image, and their relationships with more specific store perceptions (i.e., prices, merchandise quality, etc.) (see Markin, Lillis, and Narayana 1976; Donovan and Rossiter 1982; Donovan, Rossiter, Marcoolyn, and Nesdale 1995). Consistent with this program of research, Darden and Babin (1994) recently reported systematic relationships among four affect dimensions and perceptions regarding prices, quality, store personnel, and crowding. In particular, they found that higher levels of pleasantness (unpleasantness) were associated with: 1) higher (lower) prices, 2) more (less) helpful store personnel, 3) higher (lower) overall quality, and 4) lower (higher) levels of crowding. Further, activity (sleepiness) was associated with: 1) more (less) helpful personnel, 2) higher (lower) overall quality, and 3) higher (lower) levels of crowding.

Darden and Babin (1994) relied on a memory-task in which subjects recalled the environments of stores with which they were relatively familiar. The study reported below examines the relationships among affective responses and specific store perceptions using two stimulus-based simulation techniques. Specifically, the retail environments of various apparel stores are simulated via: a) slides of actual store interiors, or b) verbal instructions combined with musical selections. The affect-store image relationships emerging in each condition are compared to one another, and to those obtained by Darden and Babin (1994).

AFFECTIVE RESPONSES AND STORE ATMOSPHERE

The Mehrabian-Russell model describes affective responses to stimuli on three dimensions: pleasure-displeasure, arousal-nonarousal, and dominance-submissiveness (Mehrabian andRussell, 1974; Mehrabian, 1976). Pleasure-displeasure describes happiness or satisfaction with the stimulus; arousal-nonarousal refers to the alertness or excitement evoked by the stimulus; dominance-arousal captures the extent to which the individual controls or is controlled by the stimulus. Donovan and Rossiter (1982) introduced the pleasure/arousal/dominance (PAD) framework to the study of store atmosphere. They used the PAD model to predict: overall store evaluations, browsing behavior, time spent shopping, affiliation with the store, and the tendency to talk with others while shopping. Although pleasure-displeasure and arousal-nonarousal were useful predictors, dominance-submissiveness exhibited limited explanatory value, due possibly to the questionable internal reliability of the scale. Indeed, many of Mehrabian and Russell's tests of the model revealed questionable predictive validity of the dominance-submissiveness dimension. [Darden and Babin (1994) developed a new scale for measuring affective responses to store environments based on extensive research. Their results indicated that pleasant versus unpleasant and activity versus sleepiness are relatively independent, unipolar dimensions in contrast to the bipolar conceptualization of Donovan and Rossiter (1982). Nevertheless, Darden and Babin's pleasant, unpleasant, activity, sleepiness dimensions are similar enough to Donovan and Rossiter's pleasure-displeasure, arousal-nonarousal dimension to allow for comparisons.] Owens (1992), in fact, has even recommended that the dominance-submissiveness dimension be dropped in favor of a shopping-specific dimension.

Given that musical selections are used as stimuli in the experiment reported below, it is important to note that studies of musical perceptions also identify pleasure-displeasure and arousal-nonarousal as dimensions on which music is assessed. Wedin (1972), for example, had subjects rate forty music selections from various genres on 125 unipolar adjective scales. The results of a principal components analysis indicated three distinct dimensions labeled: intensity-softness, pleasantness-unpleasantness, and solemnity-triviality. Wedin describes the pleasantness-unpleasantness dimension as corresponding to: "a fundamental quality of emotion that is well-known from several studies" (p. 250). Moreover, the intensity-softness dimension bears a remarkable resemblance to Mehrabian and Russell's arousal-nonarousal continuum. The adjectives most strongly associated with this dimension were: energy, intensity, activity, softness, relaxation, tenderness, and intimacy. [The last two adjectives, tenderness and intimacy, seem the least consistent with an arousal-nonarousal interpretation. However, research has shown that individuals desire low levels of arousal when interacting with a romantic partner (Butler and Biner, 1987; Biner, Butler, Fischer, and Westergren, 1989; Areni and Kim, 1994). The association of these terms with the others may reflect this preference.]

Wedin's third dimension, solemnity-triviality, clearly reflects a departure from the model of Mehrabian and Russell. It was most strongly associated with the adjectives: solemnity, dignity, grandiosity, popularity, and triviality. Consistent with Owen's (1992) recommendation, the research reported below drops the dominance-submissiveness dimension of the PAD model in favor of a distinction more relevant to categorizing apparel stores. Given that apparel stores can be categorized as featuring: formal attire, business attire, sporting apparel, casual wear, and so on, Wedin's (1972) third dimension seemed ideal for capturing affective responses to the environments of apparel stores.

RESEARCH HYPOTHESES

Eroglu, Ellen, and Machleit (1992) discuss numerous techniques for simulating retail environments ranging in realism from verbal descriptions to actual store interiors, and in the extent to which experimental subjects rely on memories of previous in-store experiences. Researchers using more than one simulation technique have reported generally consistent results across methods (Hui and Bateson, 1990), but many of the simulation methods identified by Eroglu et al. have yet to be compared. Moreover, many dimensions of retail environments are still unexplored. The study reported below examines the relationships among affective responses to the environments of apparel stores and perceptions regarding: 1) the quality of service, 2) merchandise quality, 3) price levels, 4) merchandise selection, 5) the fashion value of merchandise, and 6) the pleasantness of the shopping experience. However, apparel store environments are simulated either by presenting slides of store interiors or by providing verbal instructions and playing musical selections as "background music." There are reasons to suspect that the affect-store image relationships will vary between the two conditions. First, individuals tend to adopt different strategies for processing visual and aural information (Posner, Nissen, and Klein, 1976). Moreover, patrons are likely to rely on different sources of information to form different store perceptions (Mazursky and Jacoby, 1986). Hence, both the null and the alternative hypotheses presented below seem plausible and worthy of examination.

H_o: The relationships among affective responses and store image dimensions do not depend on the method used to simulate retail environments.

H_a: The relationships among affective responses and store image dimensions depend on the method used to simulate retail environments.

An additional research question relevant to the issues presented above is whether the dimensionality of affective responses to retail environments is contingent upon the simulation method. Although no hypotheses are offerred, this issue is pursued below.

METHOD

Subjects and Procedure

Sixty-two undergraduate business students were divided into two groups, corresponding to the "store interior" treatment and the "in-store music" treatment. In the store interior treatment, subjects (n=26) viewed twelve slides of the interiors of various apparel stores. The stores were selected so as to be unfamiliar to the subjects. In addition, all references to store names, manufacturers' brands etc. were omitted from the slides. After viewing each slide, subjects responded to the pleasure/arousal/seriousness scale (PAS), adapted from Mehrabian and Russell's (1974) pleasure/arousal/dominance scale and Wedin's (1972) seriousness/triviality scale. Additionally, subjects rated the stores on six image measures. This procedure was repeated for subjects (n=36) in the in-store music treatment except, instead of slides, subjects listened to twelve two-minute instrumental music selections. Subjects were instructed to imagine themselves shopping for apparel and to base their responses "on the type of clothing store you might associate with the background music being played." Each subject in both groups provided 12 observations.

Stimuli

The twelve slides of the various retail store interiors were selected from a large pool of slides used in the preparation of a retailing textbook. The primary objective in slide selection was heterogeneity. That is, the final group of twelve slides was intended to represent as heterogeneous a group of retail clothing stores as possible. With a similar objective, music was selected from a large commercial production music library. A variety of musical genre's was included in the final twelve, including pop, rock, country, blues, classical, easy listening, jazz, soul, and so on.

Measures

The performance of the three PAS subscales was assessed using common factor analysis. Items that did not load as predicted or loaded on more than one factor were deleted. The remaining items exhibited satisfactory unidimensionality and internal consistency. The pleasure dimension (a=.92) was composed of three semantic differential items (pleasing/annoying, satisfying/disappointing, hopeful/despairing); the arousal dimension (a=.84) was measured using three items (exciting/calming, frenzied/sluggish, stimulating/relaxing); the seriousness dimension (a=.91) consisted of two items (serious/light-hearted, dramatic/playful). All items were measured on ten-point scales. The twelve pleasure, arousal and seriousness scores obtained from each subject were calculated as the mean of the appropriate scale items. Means and standard deviations for these measures are presented in Table 1.

The six store image items were also measured on ten-point scales. These items were intended to capture a variety of specific impressions (poor service/excellent service, low quality merchandise/high quality merchandise, unpleasant shopping experience/pleasant shopping experience, low prices/high prices, poor merchandise selection/wide merchandise selection, outdated merchandise/up-to-date merchandise). Means and standard deviations for these items also appear in Table 1. Important for purposes of interpreting the results reported below, there were no significant differences between means or standard deviations for any of the affect or store image variables.

Analysis and Results

Analysis of variance was used to assess whether the relationships among the PAS measures and the store image variables were contingent upon the stimulus condition. Significant (p<.05) arousal by treatment interactions were found for three of the six store image measures (a fourth approached significance; see Table 2). One significant seriousness by treatment interaction also resulted. Overall interaction effects on the six store image items were estimated using multivariate analysis of variance. The results indicated significant (p<.05) overall arousal by treatment and seriousness by treatment interactions.

The specific nature of these significant interactions were investigated further using multiple regression (see Table 2). Responses to the six store image measures were regressed against pleasure, arousal and seriousness scores for the two groups. In general, the models explained a respectable amount of variance and all were significant (p<.01). Consistent with Donovan and Rossiter (1982), the pleasure component of the PAS scale explained most of the variance in the six store image measures. The significant stimulus interactions are more clearly illustrated in the signs and magnitudes of the standardized regression coefficients presented in Table 2. Generally, they are reflected in the significance of the standardized regression coefficient for one group and not the other; however, in one instance (arousal ¦ ST2), both arousal coefficients are significant but in opposite directions.

TABLE 1

MEANS AND (STANDARD DEVIATIONS) BY TREATMENT

TABLE 2

REGRESSION RESULTS: STANDARDIZED REGRESSION COEFFICIENTS

Subjects associated higher levels of pleasure with: 1) better service, 2) higher quality merchandise, 3) more pleasant shopping experiences, 4) higher prices, 5) wider selections, and 6) more up-to-date merchandise. Higher levels of seriousness were associated with: 1) better service, 2) higher quality merchandise, 3) higher prices, 4) more up-to-date merchandise, and to a lesser extent, 5) more pleasant shopping experiences and 6) wider selections.

The most interesting results, however, concern the arousal dimension. For subjects in the in-store condition higher levels of arousal were associated with: 1) poorer service, 2) lower quality merchandise, 3) more unpleasant shopping experiences, but surprisingly, 4) more up-to-date merchandise. On the other hand, subjects in the store interior condition associated higher levels of arousal with: 1) higher quality merchandise, 2) higher prices, and 3) more up-to-date merchandise. The discrepancy in the relationship between arousal and merchandise quality is interesting given that the means and standard deviations of the two measures do not differ from one another by stimulus condition at the a=.10 level of significance. Moreover, arousal exhibits consistent relationships with other merchandise perceptions across the two stimulus conditions.

DISCUSSION

Notwithstanding the deliberate selection of stimuli to maximize heterogeneity of impressions, a possible explanation for the reversal of the signs of the beta coefficients between treatments is that the samples of store interiors and background music selections were not representative with respect to the pleasure/arousal/seriousness (PAS) dimensions. Specifically, research has shown an inverted-U relationship between arousal and shopping preferences (Raju, 1980). If the sample of store interiors (background music selections) were systematically below (above) the optimal level of arousal, then one would expect the signs of the beta coefficients to differ. That is, more arousal would be more favorable for the store interior stimuli (i.e., positive betas), whereas more arousal would be less favorable for musical selections (i.e., negative betas).

One strong point against this interpretation is that means and standard deviations for the affect and store image measures were not significantly different by stimulus condition. Moreover, the signs of the betas for each PAS dimension varied across store image variables within each treatment. In other words, for some image dimensions the store interiors were rated as too low in arousal (i.e., positive betas), whereas for others the interiors were too high in arousal (i.e., negative betas). For example, the betas for the arousal dimension were positive for service quality, merchandise quality, and pleasantness of shopping experience, but negative for stylishness of merchandise (see Table 2). This explanation suggests that something other than a biased sampling of stimuli produced the results reported above.

An alternative explanation for these findings is that the norms subjects have for background music and store interiors are different with respect to seriousness, and, more importantly, arousal. Although the means and standard deviations for the affect and store image measures did not differ between stimulus conditions, perhaps subjects expectations did vary. In particular, retail patrons may expect in-store background music to be somewhat soothing and relaxing relative to the design of store interiors, which they anticipate to be more stimulating. Hence, while actual ratings did not differ between stimulus conditions, ratings relative to expectations did. As a result, the musical selections were seen as generally too arousing, whereas the store interiors were typically not arousing enough.

A second alternative explanation for the findings is that the instructions used in the music treatment were not sufficient for evoking store schema necessary for making meaningful assessments of store image. One additional result supports this contention. In a third experimental condition ("music only"), 53 subjects were asked to rate the same twelve musical selections on the PAS dimensions, but the instructions mentioned nothing about shopping or in-store background music, and the store image items were omitted from the questionnaire. Thus, perceptions of the musical selections in this condition were more stimulus-driven. If the instructions to consider the musical selections in terms of in-store background music and store image were successful in evoking store schema (i.e., "top-down" processing), one might expect results similar to those obtained using the visual stimuli. This, of course, was not the case. If, on the other hand, subjects processing of the "in-store background music" selections was more "bottom-up," [The distinction between "top-down" and "bottom-up" information processing stems from the discovery that perception is influenced by physical characteristics of the perceived stimulus as well as the perceiver's mental representations of the world. Bottom-up processes begin with the sensory input of the stimulus (i.e., musical properties) and proceed to more abstract mental representations, whereas top-down processes begin with mental representations (i.e., store schemas) and proceed to specific sensory input (Glass and Holyoak, 1986). In essence, subjects in the visual stimuli condition may have "matched" each slide to existing store category representations in memory (see Ward, Biner, and Barnes, 1992). Subjects in the music condition, on the other hand, may have engaged in more detailed processing of stimulus characteristics (i.e., temp, timbre, modality, etc.) before forming impressions (see Kellaris and Kent, 1992a, 1992b).] then the responses would be more similar to those obtained in the music condition making no mention of a retail context.

A comparison of factor analyses of the PAS dimensions by treatment lends credence to this interpretation. As shown in Table 3, the results were quite similar in the "music-only" and "in-store music" conditions. In both cases, a three factor solution emerged with each item generally loading on the "correct" factor. The factor solution emerging in the visual stimulus ("store interior") condition was considerably different. Although a three factor solution emerged, the loadings of the individual items were markedly distinct. The items regarding arousal-nonarousal and pleasure-displeasure loaded on two factors rather than one; more importantly, these two "distinct" aspects of affective responses loaded on the same two factors! Indeed, this result is probably related to the treatment by arousal interactions reported in Table 3.

It is important to note that the instructions and measures used in this study are not identical to those used by Darden and Babin (1994). Nevertheless, there is a great deal of consistency between their results and the relationships emerging in the visual stimulus condition of the study reported above. If Darden and Babin's pleasant and unpleasant dimensions are comparable to the pleasure dimension discussed above, then pleasure/pleasantness is associated with higher prices, better service, and higher quality in both studies. Moreover, if arousal is equated with Darden and Babin's activity and sleepiness dimensions, then both studies detected an association between higher arousal, and higher prices and higher quality. Thus, the "in-store music" condition seems to produce the most discrepant results. As suggested by the comparison of the results for the "in-store music" and "music only" conditions, the verbal instructions may not have been sufficient for evoking specific mental representations of store interiors.

TABLE 3

ROTATED FACTOR PATTERN: ALL ITEMS BY ALL GROUPS

CONCLUSION

The research reported above demonstrated that the relationships among affective responses and specific store perceptions depended on whether store environments were simulated via slides or a combination of verbal instructions and musical selections. In general, the musical stimuli were perceived as being too arousing for use as in-store background music, whereas the visual stimuli were typically not arousing enough. Several interpretions of this result are plausible and seem worthy of future research. At a broader level, this research suggests that the multiple bases for simulating in-store environments may not always produce equivalent findings with respect to perceptions and preferences. Although Hui and Bateson (1990) reported similar effects of customer density on perceptions of control and crowding when the former was manipulated via slides versus videotapes, it is not clear that such correspondence will emerge across the simulations methods identified by Eroglu et al. (1992). In particular, a comparison of results obtained using verbal descriptions of store interiors to those based upon more realistic settings (i.e., videotapes and actual stores) would seem a fruitful avenue for future research.

REFERENCES

Areni, Charles S. and David Kim (1994), "The Influence of In-store Lighting on Consumers' Examination of Merchandise in a Wine Store," International Journal of Research in Marketing, 11, 117-125.

Biner, Paul M., Darrell L. Butler, Ann R. Fischer, and Amy J.Westergren (1989), "An Arousal Optimization Model of Lighting Level Preferences: An Interaction of Social Situation and Task Demands," Environment and Behavior, 21 (1): 3-16.

Butler, Darrell L. and Paul M. Biner (1987), "Preferred LightingLevels: Variability Among Settings, Behaviors, and Individuals," Environment and Behavior, 19 (6), 695-721.

Darden, William R. and Barry J. Babin (1994), "Exploring the Concept of Affective Quality: Expanding the Concept of Retail Personality," Journal of Business Research, 29, 101-109.

Donovan, Robert J. and John R. Rossiter (1982), "Store Atmosphere:An Environmental Psychology Approach," Journal of Retailing, 58(Spring), 34-57.

Donovan, Robert J., John R. Rossiter, Gilan Marcoolyn, and Andrew Nesdale (1995). "Store Atmosphere and Purchasing Behavior," working paper, University of Western Australia.

Eroglu, Sevgin, Pam Scholder Ellen, and Karen A. Machleit (1992), "Environmental Cues in Retailing: Suggestions For a Research Agenda," Proceedings of the 1991 Symposium on Patronage Behavior and Retail Strategy: Cutting Edge II, 51-60.

Glass, Arnold L. and Keith J. Holyoak (1986), Cognition, New York: Random House.

Hui, Michael K.M. and John E.G. Bateson (1990), "Testing a Theory of Crowding in the Service Environment," in Advances in Consumer Research, (eds.) M.E. Goldberg, G. Gorn, and R.W. Pollay, Vol. 17, Provo, UT: Association for Consumer Research, 866-873.

Kellaris, James J. and Robert J. Kent (1992a), "The Influence of Music on Consumers' Temporal Perceptions: Does Time Fly When You're Having Fun?," Journal of Consumer Psychology, 1, 365-376.

Kellaris, James J. and Robert J. Kent (1992b), "Consumers' Affective Responses to Music as a Function of Objective Stimulus Properties and Subjective Mediators," working paper, University of Cincinnati.

Markin, Rom J., Charles M. Lillis, and Chem L. Narayana (1976). "Social-Psychological Significance of Store Space," Journal of Retailing, 52(1), 43-54.

Mazursky, David and Jacob Jacoby (1986), "Exploring the Developmentof Store Images," Journal of Retailing, 62, 145-165.

Mehrabian, Albert (1976), Public Places and Private Spaces, New York, Basic Books.

Mehrabian, Albert and James A. Russell (1974), An Approach to Environmental Psychology, Cambridge, MA: M.I.T. Press.

Owens, Jan P. (1992), "Store Atmosphere: An Environmental Psychology Approach Revisited," Proceedings of the 1991 Symposiumon Patronage Behavior and Retail Strategic Planning: Cutting Edge II, 37-50.

Posner, Michael I., Mary Jo Nissen, and Raymond M. Klein (1976),"Visual Dominance: An Information Processing Account of Its Origins and Significance," Psychological Review, 83, 157-171.

Raju, P. S. (1980), "Optimum Stimulation Level: Its Relationship to Personality, Demographics, and Exploratory Behavior," Journal of Consumer Research, 7, 272-282.

Ward, James C., Mary Jo Bitner, and John Barnes (1992), "Measuring the Prototypicality and Meaning of Retail Environments," Journal of Retailing, 68, 194-220.

Wedin, Lage (1972), "A Multidimensional Study of Perceptual-Emotional Qualities in Music," Scandinavian Journal of Psychology, 13, 241-257.

Yalch, Richard and Eric Spangenberg (1990), "Effects of Store Music on Shopping Behavior," The Journal of Services Marketing, 4, 31-39.

----------------------------------------