Journal of Business Research 58 (2005) 526 532

An empirical investigation of Web site characteristics, consumer emotional states and on-line shopping behaviors
Venkatapparao Mummalaneni
School of Business, Virginia State University, Petersburg, VA 23806 USA

Abstract On-line retailing is rapidly emerging as an alternative mode of shopping and is expected to garner a substantial share of the retail market in future. The low order conversion ratio of Web site visitors and the inability of on-line retailers to attract repeat customers are the main hurdles identified for their lack of success at present. Enhancement of Web site features that facilitate the navigation of the consumer through the purchasing stages is suggested as a means of overcoming these hurdles. The present study proposes the stimulus organism response (S-OR) framework of Mehrabian and Russell [Mehrabian A, Russell JA. An approach to environmental psychology. Cambridge (MA): MIT Press, 1974] as a viable model for the investigation of consumer responses to the Web sites of virtual stores and finds substantial support for its propositions. The paper concludes by offering some directions for Web site design as well as further research in this area. D 2003 Elsevier Inc. All rights reserved.
Keywords: Web site characteristics; Emotional states; Shopping behaviors

Internet shopping is altering the retail landscape by reducing the information search costs of consumers (Hoque and Lohse, 1999) and the transaction costs incurred by the retailers. Electronic retailing, or e-tailing, has emerged only recently and is poised for a takeoff as the digital economy evolves further (Department of Commerce, 2000). While the future growth of e-tailing appears certain according to several forecasts (Ernst and Young, 2000, Department of Commerce, 2000), the success of individual Internet retailers is by no means assured. E-tailer shakeout is already in progress (Green, 2000) and the scramble for the identification of successful e-tail models is on. The need for improving the order conversion ratio (proportion of Web site visitors that actually place a purchase order) and attracting repeat customers are the imperatives for success (Reibstein, 2000). Through better Web site design, e-tailers can expect to improve the order conversion rate, which at present hovers around an abysmal 1.8% (Tedeschi, 2000). The environment of the virtual storefronts created through Web page design is not unlike the atmosphere of the brick-and-mortar stores with their emphasis on layout,

* Tel.: +1-804-524-5808, +1-804-706-4837; fax: +1-804 524-6845. E-mail address: vmummala@vsu.edu (V. Mummalaneni). 0148-2963/$ see front matter D 2003 Elsevier Inc. All rights reserved. doi:10.1016/S0148-2963(03)00143-7

merchandise displays, lighting, signage and so on. Analogies have in fact been drawn between real and virtual stores and their features compared. Lohse and Spiller (1999) identified the attributes of real stores such as layout and atmosphere and the corresponding characteristics of on-line retail stores. Whether the retail store is real or virtual, customer interactions with the store environment influence their emotional responses and shopping behaviors (Chang et al., 2002; Novak et al., 2000; Sherman and Smith, 1986; Baker et al., 1992; Donovan et al., 1994; Spies et al., 1997). Underlying much of the retailing research focusing on the influence of store atmosphere on store patronage behaviors is the stimulus organism response (S-O-R) framework developed by Mehrabian and Russell (1974) from an environmental psychology perspective. Originally introduced by Donovan and Rossiter (1982) to the retail store context, this framework suggests that store atmosphere affects emotional states of consumers, which, in turn, influence their shopping behaviors. The present research proposes that the propositions of the S-O-R framework are equally applicable to the context of virtual stores. Based on this framework, hypotheses are developed regarding the relationships between emotional responses of visitors to virtual storefronts and their shopping

V. Mummalaneni / Journal of Business Research 58 (2005) 526532

527

intentions/behaviors. A study is conducted with student subjects and the hypotheses are tested using the data.

1. Literature review Two streams of literature provide the necessary background for this study. First, the newly emerging set of studies on on-line shopping behaviors of consumers is summarized. Then, prior research on the influence of store atmosphere on the emotional responses and shopping behaviors of consumers is reviewed briefly. 1.1. Electronic retailing/on-line shopping According a recent Department of Commerce Report (2000), . . .the evolution of digital business is still in an early stage and the large investments being made by the retail industry into information technology infrastructure should bear fruit in the future. Several surveys point to the increasing acceptance of on-line shopping by consumers and the continued growth of e-tailing. A recent report from Ernst and Young (2000) indicates that the number of on-line shoppers jumped from 17 million in 1998 to 39 million in the following year. On-line retail sales of goods stood at US$5.3 billion during the last quarter of 1999 according to the first official measure of e-tail released by the U.S. Bureau of the Census (Department of Commerce, 2000). Specific measures of consumer shopping behavior such as the number of items purchased and the average amount spent by each shopper indicate an upswing as well (Ernst and Young, 2000). Recent studies of electronic shopping have focused on the strategies adopted by e-tailers in order to identify the factors behind retailer success or failure. Of the 10 merchant attributes considered critical to purchasing, those related to Web site design were rated to be fairly important by the nearly 100,000 on-line shoppers surveyed (Reibstein, 2000). Product representation was ranked first and ease of ordering ranked fifth among the 10 attributes. While Web site navigation and looks were ranked last by the shoppers, Reibstein (2000) suggests that it very well might be a reason to not shop there. Drawing data from a survey of shoppers at the 28 stores of a cybermall, Lohse and Spiller (1999) regressed 13 virtual store attributes on the number of unique visitors to the store and sales in dollars. Product lists (especially, those with pictures and descriptive information) showed strong relationships with both traffic and sales, perhaps due to their ability to facilitate navigation and decision making by consumers. Swaminathan et al. (1999) proposed a model with vendor characteristics, security of transaction, concern for privacy and customer characteristics as predictors of a purchase by consumers and tested it with data from a recent GVU WWW User Survey. While their model explains only 13% of the variance in the likelihood of a purchase, convenience

of using Web vendors alone accounts for 7% of the variance. Frequency of on-line purchases was sought to be explained through a model incorporating demographic, shopping orientation, channel knowledge and perceived channel utility variables by Li et al. (1999). Results of hierarchical regression analysis indicate that the addition of channel utility variables raises the amount of explained variance from 26.9% to 28.9%, a 2% increase. Among the channel utility variables, accessibility, measured through Web site characteristics, is particularly influential. Theoretical rationale, as well as empirical evidence for the influence of Web site design on shopping behaviors, has been offered by Hoque and Lohse (1999). Using the information cost perspective, they are able to show that the design of user computer interface affects consumer choice. A recent report from the trade group Shop.org identifies Web page design, site search and checkout process as some of the crucial variables that explain the high rate of shopping cart abandonment by e-tail consumers (Tedeschi, 2000). Poor labeling of products, clutter and inconvenience are some of the characteristics of Web sites identified in the same report. Improvements in site design, navigation and ease of use could lead to significant gains in conversion rates (of visitors into customers) as well as sales (Tedeschi, 2000). Conceptualizing flow as a cognitive state experienced during the navigation of the Web, Novak et al. (2000) hypothesized and tested the relationship between on-line experiences and consumer behaviors. Their model incorporates parts of the S-O-R framework and employs same scale items in measuring variables such as arousal and positive affect (pleasure). Their results show that features of Web site such as the ease of ordering and customer support have an impact on arousal, flow and Web exploratory behaviors. They suggest that the relationship between on-line customer experience and marketing outcome variables deserves to be investigated in future research. 1.2. Store atmosphere and shopping behavior Studies investigating the relationship between store atmosphere and consumer shopping behaviors have relied in the main, on the S-O-R framework proposed by Mehrabian and Russell (1974). According to this framework, emotional states of the individual mediate the influence of the environment on individual behavior. As adapted by Donovan and Rossiter (1982), the model proposed that store environment arouses the basic emotional states of pleasure arousal dominance (PAD) in the individual and the emotional states in turn would affect approach and avoidance behaviors of individuals. The results of the exploratory study conducted by Donovan and Rossiter (1982) indicated that while the influence of pleasure on all of the approach/ avoidance behaviors was statistically significant, arousal had significant influence only on affiliation measures and dominance on none of the measures.

528

V. Mummalaneni / Journal of Business Research 58 (2005) 526532

The framework had been employed in numerous retailing behavior studies since then, albeit in modified forms. Sherman and Smith (1986) improved upon the Donovan and Rossiter (1982) study by varying the study setting. Instead of student subjects indicating their shopping intentions, their study measured the behaviors of actual shoppers at retail stores. However, the consumer mood measure employed in their study was a summated measure of the three distinct emotional states employed by Donovan and Rossiter (1982). The recent study by Spies et al. (1997) also employed mood state as the mediating variable. The results of the Baker et al. (1992) study confirm the mediating role of pleasure and arousal on consumer willingness to buy. Donovan et al. (1994) conducted a study that affected several conceptual and methodological modifications from the pioneering study of Donovan and Rossiter (1982). While support was found for the mediating role of pleasure, the role of arousal received only partial support. The Sherman et al. (1997) study on the other hand found that the influence of pleasure was limited to the liking for the store and the dollar amount spent in the store, whereas arousal affected behavioral measures such as time, money spent and the number of items purchased during the shopping trip. A recent study of Korean consumers concluded that while store characteristics such as location directly affected store attitudes, consumer emotional responses mediated the effects of others (for instance, store facilities, product assortment and salespersons service) on store attitudes (Yoo et al., 1998). Thus, while inconsistencies abound regarding the specifics, these studies provide broad support for the propositions of the S-O-R framework.

Following the S-O-R framework proposed by Mehrabian and Russell (1974) and the empirical work of Donovan and Rossiter (1982) and Donovan et al. (1994) reviewed earlier, it is hypothesized that H1: The better the design of the Web site and its ambience (a) the greater the pleasure experienced by the on-line shopper. (b) the greater the arousal experienced by the on-line shopper. H2: The influence of design and ambience characteristics of the Web site on (a) satisfaction is mediated by the pleasure and arousal experienced by the on-line shopper. (b) the expressed intention of loyalty to the store is mediated by the pleasure and arousal experienced by the on-line shopper. (c) the amount of time spent at the store is mediated by the pleasure and arousal experienced by the on-line shopper. (d) the number of items purchased at the store is mediated by the pleasure and arousal experienced by the on-line shopper. (e) the (dollar) amount spent is mediated by the pleasure and arousal experienced by the on-line shopper.

3. The study Undergraduate students enrolled in a consumer behavior course during the Fall semester of 2000 collected the data as part of an assignment. The choice of (mostly) student subjects is deemed justifiable in view of their status as early adopters of on-line retailing as well as the practice of using them at the exploratory stages of research (see Donovan and Rossiter, 1982; Baker et al., 1992). Each respondent was asked to select two on-line stores, one each from the categories of apparel and footwear, items frequently purchased by students. The choice was restricted to two stores in each category. J. Crew and Spiegel were the apparel stores and Footlocker and Footaction were the footwear stores that the respondent could choose from. The sequence in which the stores were evaluated was randomized. In order to make the task more realistic, students were asked to prepare shopping lists prior to the shopping trip. The students filled a questionnaire about their experiences at each of the stores.

2. Hypotheses Fig. 1 represents the S-O-R framework in the context of on-line shopping, which is the source of the hypotheses proposed here. Following Donovan et al. (1994) and Sherman et al. (1997), the present study posits pleasure and arousal as the consumer emotional states that mediate the influence of store environment on shopping outcomes and behaviors. Novak et al. (2000) demonstrated the influence of Web site characteristics on the cognitive and emotional states of the consumer and recommended that their influence on marketing outcomes be examined in future research. This study proposes that satisfaction with the shopping experience and the expressed intention of store loyalty are affected by the store environment as well.

Fig. 1. Effects of on-line store environmental characteristics on shopping behaviors.

V. Mummalaneni / Journal of Business Research 58 (2005) 526532 Table 1 Items and reliability coefficients of variables Variable Design factor Items Large Small Roomy Cramped Colorful Drab Comfortable Uncomfortable Well-organized layout Unorganized layout Good displays Bad displays Large selection Limited selection Helpful signage Unhelpful signage Unattractive Attractive Unpleasant Pleasant Dull Bright Lively Unlively Cheerful Depressing Boring Stimulating Happy Unhappy Bored Relaxed Unsatisfied Satisfied Pleased Annoyed Contented Melancholic Despairing Hopeful Frenzied Sluggish Calm Excited Dull Jittery Unaroused Aroused This shopping trip was truly a joy. I continued to shop, not because I had to, but because I wanted to. I enjoyed this shopping trip for its own sake, not just for the items I may have purchased. This shopping trip was not a nice expedition. I was disappointed because I had to go to another store to complete my shopping. I intend to shop at this store in future. I would recommend this store to my friends. I will avoid this store in future. Number of items 8 Cronbachs a .88

529

Ambience

.84

Pleasure

.87

Arousal

.71

analysis of the store image items yielded design and ambience factors among others. Design factor made of eight items has a reliability coefficient of .88, whereas the ambience factor with six items (a=.84) showed an acceptable level of reliability. Emotional states were measured using scales originally developed by Mehrabian and Russell (1974) and employed in retail atmosphere studies. Following Sherman et al. (1997), semantic differential items with seven-point scales were used to measure pleasure (six items) and arousal (four items) dimensions (see Table 1). Shopping behaviors relating to time spent at the store, number of items purchased and money spent were each measured with a single item. These are quantitative measures indicating time in minutes, number of items and the dollar value of purchases made or intended. Satisfaction was measured using seven-point Likert scales adapted from those developed by Babin et al. (1994). Intended loyalty to the store was measured with the help of three items. The items measuring store characteristics, pleasure, arousal, satisfaction and loyalty variables and their reliabilities are shown in Table 1. Following standard practices, negatively worded items were reversed and variable scores were obtained by summing across the multiple items measuring the variables.

Satisfaction

.75

4. Analysis and results A thumbnail sketch of the sample can be created from the demographic data provided by the study participants. The 130 participants provided 260 evaluations of e-tailers, with 250 usable responses. Nearly all participants are full-time students, with a significant percentage of them working part time. The age of participants ranges from 21 to 55, with an average of 23.4 years. The sample is skewed in terms of gender, with over 64% being female. Over 86% of the sample is African American, nearly 5% Hispanic and about 7% Caucasian, with the others constituting the remainder of the sample population. Hypothesis 1 posits the influence of Web site design and ambient factors on pleasure and arousal experienced by the on-line shopper. This hypothesis is tested through regression analysis. The results of regression analyses for the two dependent variables are shown in Table 2.
Table 2 Regression results: influence of Web site characteristics on emotional states Predictor variable Dependent variable: pleasure (b coefficients) .519 ** .226 ** .477 117.06 (.000) Dependent variable: arousal (b coefficients) .570 ** .012 ** .329 62.81 (.000)

Intended loyalty

.77

Since this was a required assignment for the course, the students were told that it was not necessary to actually purchase any items; instead, they could indicate what they would have liked to purchase from the store. This is consistent with the practice in the relevant literature (Donovan and Rossiter, 1982; Baker et al., 1992), of employing shopping intention measures in place of actual purchases. 3.1. Measurement of variables Characteristics of the store environment were measured using the items employed by Sherman et al. (1997). Factor

Ambience Design Adjusted R2 Model F (significance)

** Statistically significant at .01 level.

530

V. Mummalaneni / Journal of Business Research 58 (2005) 526532

Table 3 Regression results: influence of Web site characteristics on shopping behaviors Predictor variable Ambience Design Adjusted R2 Model F (significance) Satisfaction .432** .105 .253 43.54 (.000) Intended loyalty .440** .199** .348 68.101 (.000) Time spent .075 .034 .002 1.30 (.275) Number of items .218 ** .047 .056 8.39 (.000) US$ Amount spent .093 .064 .013 2.624 (.075)

** Statistically significant at .01 level.

These results indicate substantial support for Hypothesis 1. Both of the models are statistically significant and the Web site environmental characteristics explain over 47% and 32% of the variance in pleasure and arousal, respectively. However, while the influence of both ambience and design factors on pleasure is statistically significant, the influence of design factor is not significant in the case of arousal. Hypothesis 2 states that the effects of Web site characteristics on shopping outcomes and behaviors are mediated by pleasure and arousal experienced by on-line shoppers. To test the mediational effects of the consumer emotional states, the three-step procedure suggested by Baron and Kenny (1986) is followed here. Three regression models need to be tested to establish mediational effects. First, the emotional responses (pleasure and arousal) are regressed on the Web site environmental characteristics of ambience and design. Next, shopping outcomes and behaviors are regressed on the Web site characteristics. Finally, shopping outcomes and behaviors are regressed on a full set of predictor variables including both Web site characteristics and shopper emotional states. Mediational effects are demonstrated according to Baron and Kenny (1986) if the effects of Web site characteristics are statistically significant in the first two models but not in the third. If the effects of emotional states on shopping behaviors are statistically significant, the mediating role of the shopper emotional states is established conclusively. The results of the three regression models are presented in Tables 2, 3 and 4, respectively. Since seemingly related variables such as ambience and arousal are being employed as predictors in the regression equation, multicollinearity is a concern. Multicollinearity has been ruled out since the VIF values are far below the threshold value of 10 (Hair et al., 1998). Results presented in Table 3 show partial support for the influence of Web site characteristics on shopping behaviors. The influence of Web site characteristics is statistically

significant in case of shopper satisfaction, intention of loyalty and the number of items purchased, but not in the case of time or money spent by the shopper. Even among these three outcome variables, the influence of design factor is statistically significant only in case of loyalty, whereas the influence of ambience is significant in all models. The amount of variance explained varies greatly across the five shopping behaviors. While the greatest amount of variance is explained in case of loyalty (35%), hardly any is explained in case of time spent at the Web site. The results presented in Table 4 indicate only partial support for the hypothesis. Regression models are statistically significant (at the a=.05 level) in all cases except the amount spent on purchases. The amount of explained variance ranges from a miniscule 2% in case of the dollar amount spent on purchases to a respectable 41% in case of satisfaction and an even higher 46% in case of loyalty. The effects of Web site characteristics are no longer statistically significant with the lone exception of store ambience on shoppers intended loyalty. Even in this case, the effect size has decreased from the second set of regression models estimated. The effects of pleasure and arousal on the shopping variables are not uniform, however. Pleasure proves to be a significant source of influence in a majority of cases, including satisfaction, loyalty and the number of items purchased; the contributions of arousal are significant only in case of satisfaction and time spent. In the case of time spent at the store, arousal is a statistically significant predictor, but not pleasure. The more aroused consumers are by the store environment, the more time they spend at the store. Pleasure apparently does not influence the amount of time spent at the store Web site. In regard to purchase variables, the regression model is statistically significant in the case of number of items purchased, but not the amount spent. The amount spent by the study participants on purchases is not affected by their

Table 4 Regression results: influence of Web site characteristics and consumer emotional states on shopping behaviors Predictor variable Ambience Design PLEASURE AROUSAL Adjusted R2 Model F (significance) Satisfaction .102 .007 .416** .199** .408 44.34 (.000) Intended loyalty .177* .105 .407** .091 .455 53.60 (.000) Time spent .363 .035 .016 .210* .023 2.45 (.047) Number of items .088 .006 .237* .014 .081 6.46 (.000) US$ Amount spent .008 .041 .099 .087 .020 2.273 (.000)

* Statistically significant at .05 level. ** Statistically significant at .01 level.

V. Mummalaneni / Journal of Business Research 58 (2005) 526532

531

emotional responses of pleasure or arousal or the characteristics of the Web site. The number of items purchased, however, is affected by pleasure but not by arousal. When consumers derive pleasure out of their interaction with the virtual store environment, they are likely to purchase more items. Based on the criteria suggested by Baron and Kenny (1986), it may be concluded that among the three shopping outcome variables directly affected by store environmental characteristics, their influence is perfectly mediated by the shopper emotional states in case of satisfaction and the number of items purchased, but only partially mediated in case of loyalty, for the influence of ambience is still statistically significant in the third regression model (Table 4).

5. Discussion Though its origins are recent, on-line retailing is growing rapidly and by all estimates will account for a substantial proportion of retail sales in future. With the exception of a few firms such as E-Bay, on-line retailers are yet to succeed in finding ways of attracting and retaining customers. It is proposed here that Web site design is an important element of the retail strategy that would help e-tailers find success. The results of the present study indicate the usefulness of the S-O-R framework in understanding the relationships among Web site characteristics, emotional responses of shoppers and their purchasing behaviors. The picture that emerges from the results of this study is a complex one. While broadly validating the S-O-R framework, the results are mixed regarding the mediating role of emotional responses on shopping behaviors. However, such results are not uncommon in this stream of research. For instance, the research by Dawson et al. (1990) found only partial support for the mediating role of pleasure and arousal. Further, the amount of variance explained in retail preference and choice by Dawson et al. (1990) is comparable to the R2 values reported here. From a retailer perspective, it is disappointing to find that the time and dollar amount spent at the store Web site are not influenced by ambience or design factors. While ambience had an influence on the number of items purchased, it appears that this influence is mediated by the pleasure variable in the final model. Retailers must direct their efforts to the generation of the pleasure response in order to ensure advantageous marketing outcomes. Stickiness of the Web site (i.e., its ability to keep the consumer at the site) is emerging as a crucial factor in the success of on-line stores. Retailers must simply do a better job of designing their Web sites to hold shopper interest. Time spent at the store Web site is influenced by arousal but not pleasure according to our results. Considering their recent origins, virtual stores presumably present a novel environment to the shopper

especially the first time shopper. Our data show that for a majority of respondents, this is the first encounter with the specific Web storefront. This, combined with the newness of the e-shopping concept, must make the environment a novel one that requires a greater amount of consumers time to explore. While consumers generally found the store environments pleasant, pleasure does not affect the amount of time devoted to the exploration of the Web site. Novak et al. (2000) concluded from their study of customer experiences in on-line environments that with length of Web usage, enjoyment as well as exploratory behaviors decline. It is possible that among experienced users of Web, the relationship between pleasure and time spent would be stronger than the one observed here. The number of items purchased is affected by pleasure but not arousal. A plausible explanation for these results lies in the utilitarian and hedonic shopping values that consumers bring to a shopping expedition (Babin et al., 1994). Those with utilitarian motives would be pleased to complete their shopping task, whereas those with hedonic motives would be focused on the exploration of the Web site itself, but not the purchasing task. If the Web site is not found to be interesting, the latter are certainly dissatisfied. The dollar amount spent on purchases is not influenced by any of the independent or mediator variables in this study. This result is inconsistent with findings of previous studies (for example, Sherman et al., 1997). A plausible explanation for the absence of the relationship lies in the design of the study and the characteristics of the shoppers. Recall that the participants were students who were told that they are required to visit the store but need not purchase any products if they did not wish to. That is, their shopping expedition was involuntary, but they had a choice in regard to the decision to purchase or not. Budgetary constraints faced by full-time students may also explain why certain purchases were not made. The results are unequivocal in regard to the influence of pleasure and arousal on consumer satisfaction with the shopping trip. Both variables exhibit statistically significant and strong effects on satisfaction. They are also perfect mediators of the Web site characteristicssatisfaction relationship. Since satisfaction is widely regarded as an affective response, it is not surprising that the two emotional responses are strongly associated with it. As Yoo et al. (1998) have suggested previously, the need for managing the emotional state of the on-line shoppers through better Web site design cannot be overemphasized. Any improvements in this area are likely to yield rich dividends in terms of shopper conversion ratios and enhanced store loyalty.

6. Limitations and future research Data were collected from university students enrolled in a consumer behavior course. This raises a question regard-

532

V. Mummalaneni / Journal of Business Research 58 (2005) 526532

ing the generalizability of the results to the population of eshoppers. While certain characteristics of the students match the profile of Internet shoppers that is emerging from various studies, they are not representative of this population. The Internet shopper population is described as being young, highly educated and affluent and is skewed toward males and Caucasians (Reibstein, 2000; Ernst and Young, 2000). Our sample matches this profile in terms of education, but not the other characteristics. While students are certainly an attractive market for e-tailers in product categories such as music and footwear, they may not constitute the market for several other products at this stage in the evolution of e-tailing. The artificial setting of the study also casts a cloud of suspicion on the generalizability of its findings. Many students, however, reported that they enjoyed the task and some have actually made purchases, thereby easing the concern about the artificiality of the study to a certain extent. The results of this research demonstrate the effects of virtual store environments on shopper emotions and purchase behaviors. Further studies are needed to validate the findings of the present study and extend the propositions of the S-O-R framework to a variety of virtual store contexts. Studies that experimentally manipulate the environment of the virtual store would clearly capture the effects of store atmosphere on consumer emotions and behaviors. Winer (1999) has recently identified external validity (or lack thereof) as one of the most worrisome issues facing consumer researchers. Future studies testing the S-O-R framework in the Web store context must employ varied settings (specific Web sites) and samples of populationdiffering from the present one on such characteristics as age, income and ethnicity. In keeping with the suggestion made by Winer (1999) to pair marketing science studies with consumer behavior studies as a means of extending the generalizability of results, an intriguing idea would be to use click stream data to test the S-O-R framework. The flow concept tested by Novak et al. (2000) in the on-line environment offers an excellent opportunity for theory building and testing. While flow is defined by them in cognitive terms, its relationship to emotional responses and marketing outcomes can be modeled and tested in future studies. Finally, the issue of digital divide has been raised repeatedly as a matter of concern (for instance, Department of Commerce, 2000). Lack of access to Internet may prevent the minorities and less affluent from fully enjoying the benefits of Internet, including e-shopping. The present study employed a sample of (primarily) African American students. Comparative studies that explore the differences in the use of Internet shopping by different racial groups will certainly advance the frontiers of retailing research.

References
Babin BJ, Darden WR, Griffin M. Work and/or fun: measuring hedonic and utilitarian shopping value. J Consum Res 1994;20(March):644 56. Baker J, Levy M, Grewal D. An experimental approach to making retail store environmental decisions. J Retail 1992;68(Winter):445 60. Baron RM, Kenny DA. The moderator-mediator variable distinction in social psychological research: conceptual, strategic and statistical considerations. J Pers Soc Psychol 1986;51:6(Dec.):1173 82. Chang, JE, Simpson, TW, Rangaswamy, A, Tekchadaney, JR. A good website can convey the wrong brand image! A preliminary report. Working paper. EBRC, Pennsylvania State University; 2002. Dawson S, Bloch PH, Ridgway NM. Shopping motives, emotional states and retail outcomes. J Retail 1990;66(Winter):408 27. Department of Commerce. Digital economy 2000, 3rd report. Available: http://www.ecommerce.gov/ede/; 2000. Donovan RJ, Rossiter JR. Store atmosphere: an environmental psychology approach. J Retail 1982;58(Spring):34 57. Donovan RJ, Rossiter JR, Marcoolyn G, Nesdale A. Store atmosphere and purchasing behavior. J Retail 1994;70(3):283 94. Ernst & Young. Global on-line retailingan Ernst and Young special report. 2000. Green H. Shakeout e-tailers. Bus Week e biz 2000;(May 15):101 8. Hair JF, Anderson RE, Tatham RL, Black WC. Multivariate data analysis. Upper Saddle River (NJ): Prentice-Hall; 1998. Hoque AY, Lohse GL. An information search cost perspective for designing interfaces for electronic commerce. J Mark Res 1999;36(August): 387 94. Li H, Kuo C, Russell MG. The impact of perceived channel utilities, shopping orientations, and demographics on the consumers on-line buying behavior. J Comput-Mediat Commun 1999;5(2). http://www.ascusc.org/ jcmc/vol5/issue2/hairong.html. Lohse GL, Spiller P. Internet retail store design: how the user interface influences traffic and sales. J Comput-Mediat Commun 1999;5(2). http://www.ascusc.org/jcmc/vol5/issue2/lohse.htm. Mehrabian A, Russell JA. An approach to environmental psychology. Cambridge (MA): MIT Press; 1974. Novak TP, Hoffman DL, Yung Y-F. Measuring customer experience in online environments: a structural modeling approach. Mark Sci 2000; 19(1):22 42. Reibstein, DJ. Who is buying on the Internet, 1999? Working paper. Wharton School, University of Pennsylvania; 2000. Sherman E, Smith RB. Mood states of shoppers and store image: promising interactions and possible behavioral effects. In: Anderson P, editor. Adv Consum Res, vol. 14. ACR; 1986. p. 251 4. Sherman E, Mathur A, Smith RB. Store environment and consumer purchase behavior: mediating role of consumer emotions. Psychol Mark 1997;14(4):361 78. Spies K, Hesse F, Loesch K. Store atmosphere, mood and purchasing behavior. Int J Res Mark 1997;14:1 17. Swaminathan V, Lepkowska-White E, Rao BP. Browsers or buyers in cyberspace? an investigation of factors influencing electronic exchange. J Comput-Mediat Commun 1999;5(2). http://www.ascusc.org/jcmc/vol5/ issue2/swaminathan.htm. Tedeschi B. Easier-to-use sites would help e-tailers close more sales. New York: New York Times; 2000. June 12. Winer RS. Experimentation in the 21st Century: the importance of external validity. J Acad Mark Sci 1999;27(3):349 58. Yoo C, Park J, Macinnis DJ. Effects of store characteristics and in-store emotional experiences on store attitude. J Bus Res 1998;42:253 63.