Advances in Consumer Research Volume 10, 1983 Pages 389-393
SHOPPING WITHOUT PURCHASE: AN INVESTIGATION OF CONSUMER BROWSING BEHAVIOR
Peter H. Bloch, Louisiana State University
Marsha L. Richins, Louisiana State University
Browsing behavior is discussed as a significant form of consumer behavior which can occur independently of specific purchase occasions. After defining the term browsing and reviewing related literature, results of an empirical study of browsing are described. A mail survey of 438 adults was conducted and correlates of recreational browsing were examined in two separate situations: (1) browsing in car dealerships and (2) clothing store browsing. The construct was further investigated by the use of multiple discriminant analyses to determine which variables best distinguish between browsers and nonbrowsers. Finally, implications of the research. for retail strategy are discussed.
While most consumer studies still center around buying behavior, a number of authors have called for increased attention to consumer behavior which exists outside the purchase context. For example, Belk (1992) has argued for more emphasis on product acquisition methods other than purchase such as renting or trading, whereas, Jacoby (1978) has pointed to the dearth of research on consumption and disposition of products. It is clear that consumers regularly interact with marketing institutions and have contact with products outside of purchase occasions.
The focus of the present study is upon a different aspect of non-purchase consumer behavior-browsing. Most of us at one time or another have visited a retail store without a particular purchase in mind. Such browsing may be done 'for the fun of it," to see new developments in a product class, or perhaps just to fill time while waiting. Given the pervasiveness of browsing behavior among consumers, it is surprising that this element of consumer behavior has received so little research attention.
For purposes here, browsing is defined as the examination of a store's merchandise for recreational or informational purposes without a current intent to buy. As defined, browsing is both a form of leisure activity and a form of external search behavior. Browsing can provide a consumer with a way to spend a rainy afternoon and in addition it can add to the individual's store of information concerning new product developments, brand differences, or sale prices. Furthermore, the search aspect of browsing may be pleasurable in and of itself. The browser can satisfy his/her curiosity motives and add to feelings of self-esteem through the acquisition of product-specific or marketplace expertise.
One point concerning browsing's non-purchase orientation deserves further clarification. While the intent to purchase is not explicit in browsing, such activity may of course lead to a purchase. Such purchases may be immediate and of the impulse variety (Bellenger, Robertson, and Hirschman 1978; Stern 1962) or they may occur later as a result of browsing-generated information. Although browsing may culminate in a purchase, the intent here is to examine browsing as independent of buying activity.
While virtually no research has investigated browsing, two related areas, consumer search behavior and shopping orientations, have received relatively extensive research attention. A considerable literature exists pertaining to consumer search behavior (Bettman 1979; Newman 1977) and several authors have indicated that retail store visits constitute a very important source of product information (Newman and Lockeman 1975; Newman and Staelin 1972). The search literature, however, has overwhelmingly considered pre-purchase search behavior with the emphasis on those retail visits made with an upcoming purchase in mind. Non-purchase information gathering (including brow-sing) done to satisfy epistemic or curiosity motives (Fennell 1978; Sheth 1979) has been essentially ignored. For example, a consumer may visit a retailer to explore and gather information, not to help solve a particular consumption problem, but for its intrinsic satisfaction. The store's merchandise is interesting to the consumer and learning more about it via browsing is a source of pleasure.
Past work on shopping orientations provides further insight into browsing behavior. While no study has studied browsing in depth, Darden and Ashton (1974) noted that browsing tendency was ineffective in distinguishing among supermarket shopper types. In one paper on shopping orientations, Tauber (1972) posited a number of shopping motives which are unrelated to the purchase process (e.g.j diversion, sensory stimulation, and social contact). In other words, consumers gain satisfaction from shopping itself apart from that available from the products which may be purchased. Recent work by Bellenger and his colleagues (Bellenger and Korgoankar 1980; Bellenger, Robertson, and Greenberg 1977; Korgoankar 1981) supports Tauber's contention by profiling shoppers who enjoy shopping as a leisure time activity. These "recreational shoppers" have been found to be active information seekers, to prefer department stores and closed shopping malls, and spend more time shopping. While there is a recreational component to this research stream, the enjoyment of shopping as described in these studies remains applicable to planned, purchase-directed retail trips as well as to leisure time browsing. The study reported here adds to existing research on search and shopping behavior by focusing on non-purchase browsing activity.
Browsing appears to be a product- or store-specific activity. For instance, a particular consumer may browse in record stores but rarely go into stores selling pianos and organs. Such specificity may be a result of two basic factors. The first is the consumer's interest or involvement in the merchandise carried in the retail store. High levels of involvement generally stimulate a desire to browse in order to see the latest models and keep up with new developments in the product category. Several researchers have posited a positive connection between search behavior and product interest in a prepurchase context (Clarke and Belk 1979; Gronhaug 1975). This relationship also appears relevant to the explanation of browsing behavior.
Apart from consumer interest, selectivity in browsing may be a function of store or product characteristics. Store factors include atmospherics, store accessibility, image, number of salesclerks, and the way merchandise is displayed, among others. For example, a consumer may avoid a particular retail outlet as a place to browse despite strong interest in the products offered because of the possibility of being confronted by salesclerks. The mere presence of a salesperson at the entrance of a store may discourage visits from browsers. The browser who does not in tend to make a purchase may not want to interact with salespeople.
Product characteristics may also influence browsing. Fennell (1978) has suggested that product classes which are more complex or differentiated lend themselves more to consumer exploration and information seeking. Thus, it is not surprising that grocery products as studied by Darden and Ashton (1974) did not attract many recreational browsers.
Since browsing is a type of non-purchase search behavior, and may be due in part to product interest, it is further suggested that browsing is associated with other types of search behavior. For example, the person who frequently browses in camera stores may be expected to read photography magazines. his hypothesis is supported by Bellenger and Korgoankar (1980) who found that recreational shoppers tend to be information seekers as indicated by amount of television viewing and magazine readership. Browsing may be considered as one outlet for a general information hunger concerning a particular product class or the marketplace in general.
Going one step further, it may be that browsing also bears a positive relationship to both product knowledge and opinion leadership. Visiting stores and gaining product-related information from other sources results in heightened knowledge about the product class in question. Being familiar with the product class and aware of recent product trends should also lead to relatively high levels of word-of-mouth activity. Thus, the browser might be considered as a "socially integrated" consumer (Reynolds and Darden 1971), an individual high in both information search and opinion leadership. As suggested by Corey (1971) and Gronhaug (1975), strong product interest motivates both the giving and receiving of relevant information.
The foregoing discussion is summarized in the following set or hypotheses investigated in this study. Browsing is positively related to:
H1: The degree of interest in the class (or classes) of products carried in a store,
H2: The propensity to engage in other forms of search behavior relating to the product class in question,
H3: The degree of knowledge concerning the product class, and
H4: The degree of word-of-mouth activity concerning the product class.
Apart from testing these specific hypotheses, the present study also examined product/store differences and demographic influences on browsing behavior.
Products Under Study
Given the desire to investigate product and store differences in browsing activity, data was collected pertaining to two product classes, automobiles and clothing. These products were chosen because of their obvious familiarity and the potential for variance in consumer interest levels with respect to these goods. Also, these products are sold in very different types or retail outlets under very different shopping situations.
i Data collected through a mail survey of adult residents of the Portland, Oregon SMSA, an area frequently used for test markets because of its representativeness. A sample of 800 names was drawn at random from the city directory. To help generate greater variance in responses, these 800 names were supplemented with those of persons presumed to have high interest in either automobiles or clothing. Thus, surveys were also sent to 350 persons on customer mailing lists provided by women's and men's fashion clothing boutiques and to 350 members of local sports car clubs. As relationships among variables, rather than generalization to specific populations, were of interest, this sample design was deemed appropriate (Calder, Phillips, and Tybout 1980). A postcard reminder was mailed out four days after the S questionnaire.
Of the 1500 questionnaires mailed, 485 were returned, providing a 30 percent response rate. After editing, a usable sample of 438 responses were available for analysis.
Separate measures of browsing were used for automobiles and clothing. Both measures employed a five-point scale ranging from 1 (never) to 5 (very frequently) and centered around recreational browsing done without the intend to buy. For automobiles, respondents were asked how often they "visited car dealers just to look at new models without being in the market for a car." For clothing, the question read, "How often do you browse in clothing stores just for fun or recreation?"
The independent variables were operationalized as follows:
1. Product interest was measured on a five-point scale ranging from 1 (not at all interested) to 1 ( extremely interested).
2. Information seeking was operationalized as the number of product-related magazines read regularly or subscribed to. Pretests were used to generate a list of six automotive magazines and six fashion magazines. [The automotive magazines were: Car and Driver, Road and Track, Hot Rod, Autoweek, Motor Trend, and Car Craft. The fashion magazines were: Vogue, L'Officiel, Mademoiselle, Glamour, Harper's Bazaar and Gentlemen's Quarterly which was the only men's fashion magazine included in the list. Due to lack of other similar publications, a male's readership or Gentlemen's Quarterly counted two points rather than one.]
3. Self-perceived product knowledge relative to "most other people" was assessed using a rive-point scale ranging from 1 (very little knowledge to 5 (a great deal of knowledge."
4. Word-of-mouth activity was operationalized as responses to the question, "How much information about (cars/clothing styles and fashions) do you give to your friends?" Again, a five-point scale was used with endpoints 1 (very little information) and 5 (a great deal of information).
In addition to these variables used in hypothesis testing, demographic data were also collected.
To examine browsing behavior, several types of analysis were conducted. First. Pearson correlations were computed to examine the extent of relationship between browsing and each of the four independent variables relevant to the hypothesis tests. The relationships of certain demographic variables to browsing behavior were also investigated. Spearman rank order correlations were computed for browsing and age, education, and income. T-tests were used to identify possible connections between browsing and sex, marital status, whether the respondent was employed full time and whether the respondent had small children living at home. Finally, multiple discriminant analysis was used to test the power of each of the hypothesized predictors and demographic variables to discriminate between browsers and non-browsers on a dichotomized transformation of the dependent variable. All analyses were performed twice, once for each of the two products under study.
Before discussing the results of the hypothesis tests, the distribution of scores on the two browsing measures were examined for product/store differences. The two response distributions showed clear product differences, with browsing being much more common in the case of clothing (X=2.50, s=1.36) than cars (X=1.73, s=1.64). One explanation for the low level of browsing in car dealerships is the nature of the retail outlet involved versus that typical for clothing stores. Car dealers are often geographically dispersed without the ease of spontaneous access typical of clothing stores which are often located in the enclosed shopping malls preferred by recreational shoppers (Bellenger and Korgoankar 1980). The potential automobile browser may also dislike confrontations with salespersons.
One could argue that since automobiles are very expensive and infrequently purchased the nature of the product class may account for this low level of browsing. However, the large attendance at regional auto shows and the noticeable attention paid to cars displayed in shopping mall walkways attest to widespread consumer interest in the recreational examination of automobiles outside of the dealership environment. It will take further research to isolate the relative effects of product characteristics and the retail environment on browsing activity.
Moving beyond possible product or store influence, Table 1 presents results of correlation analyses used to examine the effect of hypothesized variables on browsing behavior. In both product categories, all four of the research hypotheses were supported. Thus, interest in the product class, information search in the product class, product knowledge, and word-of-mouth activity concerning the product class are all positively related to browsing behavior. Results were somewhat stronger in terms of variance explained for the clothing case,
CORRELATIONS BETWEEN INDEPENDENT VARIABLES AND BROWSING BEHAVIOR
In order to further illuminate the nature of browsing behavior, demographic predictors were also analyzed and results are displayed in Table 2. For automobiles, moderate relationships with browsing were observed for age (rho=-.11) and income (rho=.14). In addition, males are more likely than females to browse in car dealerships, and individuals employed full time are more likely to browse than those not so employed. For the clothing case, the browser tends to be young, female, single, and a person without young children living at home. These demographic findings for both automobiles and clothing are not surprising given the observed relationship of product interest to browsing behavior.
RELATIONSHIPS BETWEEN DEMOGRAPHIC VARIABLES AND BROWSING BEHAVIOR
A final set of analyses was conducted to determine the combined relationship of product interest, information seeking, product knowledge, and word-of-mouth with browsing behavior. These four variables were simultaneously entered in a discriminant analyses for each of the product classes. Before conducting these analyses, however, the two dependent variables were dichotomized in order to produce groups of browsers and.non-browsers. Given the distribution of scores on the browsing measures and a desire to examine extreme cases, browsers for each of the two product classes were defined as those subjects who indicated positions 3, 4, or 5 on the scales (resulting in 77 car browsers and 189 clothing browsers); non-browsers included those who indicated that they never browsed by marketing position 1 (resulting in 244 car non-browsers and 134 clothing nonbrowsers). Sample members indicating position 2 on the dependent variables were excluded from the discriminant analyses.
Discriminant analysis results for the automobile case are displayed in Table 3. The function which emerged from the analysis was significant (p < .001) and accounted for 35 percent of the variance in the dependent variable as determined by the statistic I2 (Peterson and Mahajan 1976), which is analogous to R2 in multiple regression. The function correctly predicted group membership in 81 percent of the cases, considerably better than chance. However, this percentage of correct classifications is likely somewhat inflated since the same observations used in producing the discriminant function were used to test its classificatory power (Morrison 1969). Because the focus of the research was on the analysis of group differences, rather than classification; this procedure was deemed acceptable. Results shown in Table 3 are in concert with those obtained in the correlation analysis, lending further support for the study hypotheses.
DISCRIMINANT ANALYSIS RESULTS
To determine whether demographic variables would significantly add to the discriminatory power provided by the four predictors, a second discriminant analysis was conducted with demographic variables added on a stepwise basis. [Ordinally scaled variables (age, income, education) were treated as interval data in performing the discriminant analyses.] In the automobile case, no demographic variable was able to further distinguish browsers from non-browsers at a .05 significance level based on the change in Rao's V.
Discriminant analysis results pertaining to clothing store browsing using the four hypothesized predictors are also shown in Table 3. Again, the discriminant function was significant at the .001 level. This function,provided 41 percent explained variance as measured by I2, a figure somewhat better than obtained in the automobile case. As for classificatory power, this function correctly predicted group membership in 81 percent of the cases, the same figure observed for automobiles.
In the second clothing discriminant analysis with demographics added on a stepwise basis, one variable, marital status, did prove to be an effective discriminator of clothing browsers versus non-browsers using the change in Rao's V criterion . Inclusion of marital status did not increase the classificatory power of the discriminant function above the 81 percent mark and only slightly increased the variance accounted for (I2=.42). Other demographic characteristics significant in the univariate tests for automobiles and clothing in Table 9 did not enter the stepwise analyses due to colinearity with other predictor variables.
Results of both the univariate and multivariate analyses reported above provide strong support for the research hypotheses. Browsing is related to product interest and knowledge, as well as to word-of-mouth activity and information search concerning the product class. There also appears to be demographic influences on browsing; however, these influences may operate through the mediating variable of product interest.
CONCLUSIONS AND IMPLICATIONS
The present results have indicated that significant numbers of people do browse in retail outlets without an upcoming purchase in mind. Browsers were found to have higher interest in and knowledge of the product class concerned than do non-browsers. In addition, browsers tend to be opinion leaders in the classic sense for the product class in question. That is, browsers disseminate product information and have greater exposure to relevant mass media (Corey 1971; Summers 1970).
As defined in this research, browsing is conducted without a prerequisite purchase intention. One might then ask what the managerial- relevance of the browser segment is if these people are not visiting stores in order to buy. The relevance of these consumers derives from two sources. First, browsers do at times make purchases. In some cases, however, these purchases are temporally removed from the browsing visit. Information obtained while looking around a store may be retained for retrieval at some later time culminating in a purchase. These delayed-action purchases are in addition to any impulse purchases which may occur at the time of the browsing visit.
Additional research is needed to determine whether these later purchases, fueled by browsing-derived information, occur in the same stores where the browsing was done. If not, one retailer through its helpful staff or informative displays may be generating purchases for competitors who are short on service and information, but who offer lower prices. In fact, the delayed-action purchases may not occur in a retail store at all--rather, a discount mail order operation may reap the benefit from a local store's hospitable browsing environment.
A second reason why browsing is significant to retailers is the influence browsers may have over other consumers. Opinion leadership, which is related to browsing, has long been recognized as an important force in the acceptance of new products and in the dissemination of product information (Glock and Nicosia 1963; Robertson 1971). As such, retailers need to consider the impact of browsers in generating sales among other consumer segments. In-store promotion and information may not only be useful in creating impulse buying by browsers and other recreational shoppers as suggested by Bellenger and Korgoankar (1980), but also in providing product information which can be disseminated to other less wellCinformed consumers.
The question might ultimately concern how far a retailer should go in encouraging browsing. The retailer must balance the benefits of impulse buying and word-of-mouth advertising with the risks of losing sales to discounters as described above. Since loss of sales to discounters may occur whether browsing is encouraged or not, the best policy may be to encourage browsing for the other potential benefits it reaps. This approach must be tempered, of course, if such encouragement adds significant costs, for instance, if selling space is limited and browsers cause sufficient crowding to turn away more purchase-oriented customers.
The study results suggest that consumers' propensity to browse is subject to product and/or store influences. Stores may beckon potential browsers as a result of the r atmospherics, product assortment, ease of access, or the attitudes of their sales staff toward browsers. For example, automobile dealers who wish to increase showroom traffic in order to generate word-of-mouth activity may have to alter the way browsers are treated during their showroom visits. Less pressure and a more relaxed manner on the part of sales personnel may make a browser reel more welcome. Because the consumer's perception of dealership's sales tactics will be difficult to change, auto retailers may attempt to encourage browsing away from the dealership. Thus, they might benefit from increased reliance on displays of new models in shopping malls or airports, less threatening environments than showroom floors stocked with overly-inquisitive or aggressive salesmen.
In conclusion, the present study has attempted to provide explicit recognition of an under-researched aspect of consumer behavior. The understanding of consumer behavior will be broadened as researchers continue to move outside the confines of the purchase to examine the wide variety of non-purchase consumer behavior, of which browsing is just one part.
Belk, Russell (1982), "Acquiring, Possessing, and Collecting: Fundamental Processes in Consumer Behavior," Proceedings 1982 Marketing Theory Conference (forthcoming).
Bellenger, Danny E., Robertson, Dan H., and Greenburg, Barnett A. (1977), "Shopping Center Patronage Motives," Journal of Retailing, 53, 29-38.
Bellenger, Danny E., Robertson, Dan H., and Hirschman, Elizabeth C. (1978), "Impulse Buying Varies by Product," Journal of Advertising Research, 18, 15-18.
Bellenger, Danny E., and Korgoankar, Pradeep. (1980), "Profiling the Recreational Shopper," Journal of Retailing, 56, 77-92.
Bettman, James R. (1979), An Information Processing View or Consumer Choice, New York: Addison-Wesley.
Calder, Bobby J. Phillips, Lynn U., and Tybout, Alice M. (1980) "The Design, Conduct, and Application of Consumer Research: Theory vs. Effects Oriented Studies," in Marketing in the 80's, eds. R. Bagozzi et al., Chicago: American Marketing Association, 307-311.
Clarke, Keith and Belk, Russell (1979), "The Effects of Product Involvement and Task Definition on Anticipated Consumer Effort," in Advances in Consumer Research, Vol. 6, ed. ". Wilkie, Ann Arbor: Association for Consumer Research, 313-318.
Corey, Lawrence G. (1971), "People Who Claim to be Opinion Leaders: Identifying Their Characteristics by Self-Report," Journal of Marketing, 35, 48-53.
Darden, William R. and Ashton, Dub (1974), "Psychographic Profiles of Patronage Preference Groups," Journal of Retailing, 50, 99-112.
Fennell, Geraldine (1978), "Consumers' Perceptions of the Product-Use Situation," Journal of Marketing, 42 38-47.
Glock, Charles Y., and Nicosia, Francesco, M. (1963), "Sociology and the Study of Consumers," Journal of Advertising Research, 3, 20-24.
Gronhaug, Kjell (1975), "How New Car Buyers Use Advertising," Journal of Advertising Research, 15, 49-53.
Jacoby, Jacob (1978), "Consumer Research: A State of the Art Review," Journal of Marketing, 42, 87-96.
Korgoankar, Pradeep. (1981), "Shopping Orientations of Catalog Showroom Patrons," Journal or Retailing, 52, 78-90.
Morrison, D. G. (1969), "On the Interpretation of Discriminant Analysis," Journal of Marketing Research, 6, 156-163.
Newman, Joseph W. (1977), "Consumer External Search: Amount and Determinants," in Consumer and Industrial Buying Behavior, eds. A. Woodside and J. Sheth, New York: North Holland, 79-94.
Newman, Joseph W., and Lockeman, Bradley D. (1975), "Measuring Prepurchase Information Seeking," Journal of Consumer Research, 2, 216-222.
Newman, Joseph W., and Staelin, Richard (1973), "Information Sources of Durable Goods," Journal of Advertising Research, 13, 19-29.
Peterson, Robert A., and .Mahajan, Vijay (1976), "Practical Significance and Partitioning Variance in Discriminant Analysis," Decision Sciences, 7, 649-658.
Reynolds, Fred D., and Darden, William R. (1971), Mutually Adaptive Effects of Interpersonal Communication," Journal of Marketing Research, 8, 449-454.
Robertson, Thomas S. (1971), Innovative Behavior and Communication, New York: Holt, Rinehart, and Winston.
Sheth, Jagdish N. (1979), The Surpluses and Shortages in Consumer Behavior Theory and Research," working paper #573, University of Illinois, College of Commerce and Business Administration.
Stern, Eawkins, "The Significance of Impulse Buying Today," Journal of Marketing, 26, 59-62.
Summers, John O. (1970), "The Identity of Women's Clothing Fashion Opinion Leaders," Journal of Marketing Research, 7, 178-185.
Tauber, Edward H. (1972), "Why do People Shop?" Journal of Marketing, 30, 50-52.