DTD 5

ARTICLE IN PRESS

Journal of Business Research xx (2004) xxx – xxx

Modeling the impact of internet atmospherics on surfer behavior Marie-Odile Richard* HEC-Montreal Business School, Department of Marketing, 3000 Chemin de la Coˆte-Ste-Catherine, Montre´al, QC, Canada H3T 2A7 Received 19 November 2003; accepted 16 July 2004

Abstract This paper examines the role of Internet atmospherics cues on the behavior of surfers and their impact on variables such as site attitudes, site involvement, exploratory behavior, pre-purchase and purchase intentions. Atmospherics cues are central (structure, organization, informativeness, effectiveness and navigational characteristics) and peripheral (entertainment). A conceptual model is developed based on a review of existing findings and tested with a large sample of consumers who responded to a questionnaire after navigating through an existing pharmaceutical web site. Structural equations modeling was used to test 10 major hypotheses. Among the key findings, all atmospherics cues were impacting the other constructs, with the central cues mostly affecting site involvement and exploratory behavior, while entertainment affected site involvement and site attitudes. These findings contribute to the theoretical and managerial understanding of the role of Internet atmospherics on the navigation behavior of visitors. D 2004 Elsevier Inc. All rights reserved. Keywords: Internet; Atmospherics; Structural equations; Model; Healthcare marketing

1. Introduction

2. Conceptual background

The Internet is growing in importance since the product is becoming more information-based and offers the opportunity to separate information about a product from the product itself (Bra¨nnback, 1997). For companies, the focus is shifting from creating websites to strategic aspects involving how to best use this medium. One important objective of firms on the Web remains effective communications with consumers. This emphasizes the importance of developing and testing systematic models of the Web as a communication tool. This paper extends the literature by, first, proposing a new model of information-seeking in online retailing; second, following Eroglu et al. (2001), integrating research on Web site environmental cues.

Turley and Milliman (2000) identified over 60 studies with relationships between store atmospherics and consumer behavior. Atmospherics influence consumer perceptions of retail products (Obermiller and Bitner, 1984), and store approach/avoidance behaviors such as consumers’ store patronage and spending (Donovan and Rossiter, 1982; Donovan et al., 1994). Atmospherics are the bintentional control and structuring of environmental cuesQ or bconscious design of space to create certain buyer effectsQ (Kotler, 1973) Atmospheric cues may be more influential than other marketing inputs at the point of purchase (Baker et al., 1994) and impact purchase decisions more than the product itself (Kotler, 1973). Web atmospherics are the bconscious designing of Web environments to create positive affect and/or cognitions in surfers in order to develop positive consumer responsesQ (Dailey, 2004). For Milliman and Fugate (1993), a Web atmospheric cue is comparable to a brick-and-mortar cue

* Tel.: +1 514 342 2050. E-mail address: [email protected]. 0148-2963/$ - see front matter D 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.jbusres.2004.07.009

JBR-06054; No of Pages 11

ARTICLE IN PRESS 2

M.-O. Richard / Journal of Business Research xx (2004) xxx–xxx

and is described as bany Web interface component within an individual’s perceptual field that stimulates one’s senses.Q Eroglu et al.’s (2001) typology divides Web atmospherics into two groups: (1) high task-relevant cues facilitate and enable the consumer’s shopping goal attainment; (2) low task-relevant cues are inconsequential to the completion of the shopping task. Unfortunately, researchers have not examined the impact of specific cues (e.g., colors, music), but have focused on very general cues lessening the probability of finding theories that explain their influence (Turley and Milliman, 2000). Another theory, the Elaboration Likelihood Model (ELM), suggests that low-involvement subjects process information through the peripheral rather than the central route (Petty et al., 1983), relying more heavily on cues as opposed to detailed and elaborate product specific information. With the Internet, marketers use many cues (e.g., search engines, keywords) to attract and influence consumers (McGaughey and Mason, 1998). Based on this conceptual background, we elaborate our conceptual model. 3. Conceptual model Following Mehrabian and Russell (1974) and Donovan and Rossiter (1982), the model is divided into three parts: Stimuli, Organism and Outcomes. Eroglu et al. (2001) identified high and low task-relevant cues. Most Web atmospherics cues belong to the first category because the concern is to evaluate the impact of information content on the other variables. In the second, figured Website entertainment. These dimensions lead to the processing variables such as approach/avoidance behaviors, which are emotional responses, and exploratory behavior and site involvement categorized as cognitive variables. The main focus of the conceptual model (Fig. 1) is on these variables as applied to the Internet. To complete this model are outcomes such as pre-purchase and purchase intentions.

Next, the model is described and the hypotheses to be tested are developed. 3.1. Internet atmospherics cues Internet atmospherics cues are critical to site effectiveness since visitors decide which Web pages to browse, for how long, and how much information to acquire. For a better understanding of what constitutes high-quality Web content, six important factors are identified from the literature and described in turn. 3.1.1. Navigational characteristics The characteristics of products and sites encountered early in online browsing influence the visitors’ levels of arousal and pleasure, and therefore their responses. Menon and Kahn (2002) show that if starting experiences encountered by surfers in a simulated Internet shopping trip are high in pleasure, there is an influence on site attitudes and surfers engage in more arousing activities such as exploration and tendencies to examine new products. Navigational cues are important in creating or not impeding the experience of surfers (Hoffman and Novak, 1996; Novak et al., 2000), leading to formation of positive site attitudes (Eagly and Chaiken, 1993; Csikszentmihalyi, 1977). Finally, Lynch et al. (2001) show that site quality influences surfers’ probability of buying during the visit and returning to visit. Thus: H1: When consumers surf the Web, navigational characteristics are positively related to: (a) site attitudes, (b) exploratory behavior, and (c) purchase intentions. 3.1.2. Structure Huizingh (2000) reported four different navigational structures: a tree, a tree with a return-to-homepage button, a tree with horizontal links and an extensive network. Most sites have a simple structure. Since there is no prior finding, we surmise that the more complex the structure, the more surfers are involved with the site. Finally, it seems logical that for people who like scrolling and browsing throughout

Fig. 1. Conceptual model.

ARTICLE IN PRESS M.-O. Richard / Journal of Business Research xx (2004) xxx–xxx

3

various sites, the structure of the site is important and positively influences their exploratory behavior and purchase intentions. Thus:

precise knowledge of available information and are not sure whether their requirements can be met or how these requirements may be reached. Thus:

H2: When consumers surf the Web, structure is positively related to: (a) site involvement, (b) exploratory behavior, and (c) purchase intentions.

H4: When consumers surf the Web, informativeness is: (a) positively related to site attitudes, and (b) negatively to exploratory behavior.

3.1.3. Effectiveness of information content Specific product information is most often available in a site. The perception of site content can be measured by how informative it is, if it provides detailed and specific information on products or other relevant topics (Huizingh, 2000). But one study found contrary results: the information content of Web pages, per se, does not appear to attract visitors (Dholakia and Rego, 1998). Site attitudes and purchase intentions. In ELM, effectiveness of information content is a central cue which explains site attitudes, brand attitudes and is positively related to purchase intentions (De Pelsmacker et al., 1998). Site involvement. Researchers noted that the Internet contained more information than other media, and is a highly involving medium (Novak et al., 2000). Thus, the measure of information content is considered as an indicator of site involvement (Yoon, 2000; Okasaki and Rivas, 2002). For utilitarian motives, surfers are involved in a site because of the information related to a product/service (Park and Young, 1986). Exploratory behavior. Without prior findings, we surmise that when surfers find an interesting topic, effectiveness of information content induces them to scroll and browse to get the most complete and appropriate information about it. Thus:

3.1.5. Organization Organization induces surfers to follow the central route, according to ELM. It evaluates how well a website presents itself and how it tour-guides its surfers (Chen and Wells, 1999). A low score suggests that it poorly leads surfers to their destination. Poor organization is caused by too many links, layers or animations, causing surfers to develop lower site attitudes and involvement (Chen and Wells, 1999). Thus:

H3: When consumers surf the Web, effectiveness of information content is positively related to: (a) site attitudes, (b) site involvement, (c) exploratory behavior, and (d) purchase intentions. 3.1.4. Informativeness Informativeness focuses on the site as an interactive provider. bIntelligent, resourceful, knowledgeableQ are adjectives that are used (Maddox, 1998; De Pelsmacker et al., 1998). Site attitudes. For Chen and Wells (1999) ad attitude (A ad) is a useful indicator of site value, but the unidimensionality of A ad does not provide a complete explanation of consumers’ ad ratings (Pashupati, 1997). Chen and Wells (1999) and Chen et al. (2002) developed three scales (informativeness, organization and entertainment) that better correlate with and explain site attitudes. In the ELM, informativeness is a central cue that impacts attitudes, which influence purchase intentions (De Pelsmacker et al., 1998). Exploratory behavior. Without prior findings, we surmise that informativeness influences surfers’ browsing and scrolling. Browsing is performed when surfers do not have a

H5: When consumers surf the Web, organization is positively related to: (a) site attitudes, and (b) site involvement. 3.1.6. Entertainment For McQuail (1983), the value of entertainment brests in its capability to fulfill audience needs for escapism, diversion, aesthetic enjoyment or emotional release.Q People scoring Web ads high in value develop favorable attitudes and high involvement with the information content of the site (Larkin, 1979). In ELM, peripheral cues may be attractive sources (here, entertainment value), humor, and visuals (Cho, 1999). For Stern (1990) and De Pelsmacker et al. (1998), consumers who consider an ad as entertaining, positively evaluate the brand and have intentions to purchase it. We surmise that entertainment has an effect on site attitudes and encourages more site exploration. Finally, site involvement based on valueexpressive motives leads to affect because the site appeals emotionally or aesthetically to surfers (Park and Young, 1986). Thus: H6: When consumers surf the Web, entertainment is positively related to: (a) site attitudes, (b) site involvement, (c) exploratory behavior, and (d) purchase intentions. 3.2. Processing variables 3.2.1. Site attitudes For Stevenson, Bruner and Kumar (2000) battitude toward the Web siteQ is useful in understanding its impact on site value. For Jee and Lee (2002), Web sites reflect characteristics of traditional ads. Thus, site attitudes should lead to consequences identical to those found in attitude research (Lutz et al., 1983; Homer, 1990). Brown and Stayman (1992) found that ad attitudes influence brand attitudes and purchase intentions. Similarly, site attitudes have a positive impact on ad attitudes, brand attitudes and purchase intentions (Bruner and Kumar,

ARTICLE IN PRESS 4

M.-O. Richard / Journal of Business Research xx (2004) xxx–xxx

2000; Dodds, 1991). In ELM, the two routes to persuasion are the central route that requires bhigh effort scrutiny of attitude-relevant information,Q and the peripheral route influenced by contextual factors and attitude change is determined by both processes (Petty and Cacioppo, 1981, 1986). Thus:

likely purchase online than others (Kwak et al., 2002). Therefore:

H7: When consumers surf the Web, their site attitudes are positively related to: (a) involvement in purchase decisions, and (b) purchase intentions.

3.3. Outcomes

3.2.2. Exploratory behavior Exploratory behavior has bthe sole function of changing the stimulus fieldQ (Berlyne, 1963). Studies suggest a twofactor conceptualization: exploratory acquisition of products, and exploratory information seeking (Baumgartner and Steenkamp, 1996). Browsing is performed when surfers are unsure of available information, whether or how their requirements may be reached; and is general or purposeful. bPurposiveQ browsing occurs when surfers have specific requirements, whereas bgeneralQ browsing allows surfers to fine-tune knowledge of their requirements or to keep themselves up-to-date on the latest changes in a specific field (Rowley, 2000). Shoppers’ exploratory behavior, characterized by information-search or exploration through purchasing, positively influences their site attitudes. The more they explore the possibilities offered by the Web, the more they fine-tune their requirements and have a positive idea of the site visited, triggering approach behavior toward the site. Thus: H8: When consumers surf the Web, exploratory behavior is positively related to: (a) site attitudes, (b) site involvement and (c) involvement in purchase decisions. 3.2.3. Site involvement Involvement is important in audience processing of traditional advertising (Petty and Cacioppo, 1986) and is very important in Web advertising (Raman and Leckenby, 1998; Cho, 1999). Day et al. (1995) refer to involvement as a motivational state influenced by ba person’s perception of the object’s relevance based on inherent needs, values and interests.Q However, with the internet, the relevant variable is site involvement, which taps a behavioral response, not a personality dimension. Highly site-involved surfers are prone to search for information in sites, and to explore new stimuli because of a higher need for environmental stimulation (Balabanis and Reynolds, 2001). For Yoo and Stout (2001), visitors with a high level of site involvement have more intentions to interact with a site, leading to extensive search and trials of interactive functions. Finally, highly site-involved people search for more information before purchasing, process relevant information in-depth, and use more criteria in their decisions than others (Leong, 1993; Maheswaran and Meyers-Levy, 1990). Highly site-involved customers more

H9: When consumers surf the Web, site involvement is positively related to: (a) site attitudes, (b) involvement in purchase decisions, and (c) purchase intentions.

3.3.1. Involvement in purchase decisions Involvement in purchase decisions precedes purchase intentions. Gore et al. (1994) reported that people highly involved in purchase decisions must recognize the problem, search actively for information, evaluate the alternatives by spending time to search for the best choice and purchase, whereas people lowly involved in purchase decisions neither do extensive information search nor evaluations of alternatives. Customers engage in ongoing information collection without specific needs (Bloch et al., 1986): value-added information is interesting and helpful. Access to information can improve consumer decisions, be an incentive for people to shop online (Jarvenpaa and Todd, 1997) and have an effect on online decisions (Ranganathan and Ganapathy, 2002). Thus: H10: When consumers surf the Web, involvement in purchase decisions is positively related to purchase intentions. 3.3.2. Purchase intentions In hierarchy-of-effects models, purchase is the ultimate stage and it takes place long after exposure to ad messages. However, with the Internet, purchase might take place either at the same time as exposure to ad messages or within a short time because shoppers can request information instantly and directly via the Internet (Cho, 1999). The final model that will be used to test the 10 hypotheses is provided in Fig. 2.

4. Data and methodology The model is tested with a pharmaceutical site. Health care is one sector where the Internet has become an invaluable communication tool. Healthcare ranks as the fourth most popular topic on the Web; among women and seniors, health sites are the second most popular destination (Bellman et al., 1999). The data were collected from the homepage of an OTC drug from one of the largest, pharmaceutical companies in North America. A Web-based methodology minimizes biases of ad testing methods and is based on an experimental method that is clear and unobtrusive to respondents. Answers that reflected their viewing behavior on the site after they did their surfing activities were measured. It was designed to test the effectiveness of environmental cues on

ARTICLE IN PRESS M.-O. Richard / Journal of Business Research xx (2004) xxx–xxx

purchase behavior and implemented on the Web itself. Participants were Web surfers engaged in their own activities. This method does not suffer from lack of external validity as content is viewed in its actual form, by viewers, and within the appropriate site environment. Thus, the methodology provided instantaneous measures of site environmental cues effectiveness. The questionnaire was a structured, non-disguised instrument, with closed-ended questions measuring respondents’ agreement/disagreement on a five-point Likert scale, except for site involvement which used a five-point semantic differential scale. It contained the variables in Fig. 2. Having been studied previously, a pool of existing items was available. The appropriate measures for each concept were selected from the literature and adapted. bNavigational characteristicsQ (CHPS; Bell and Tang, 1998) had 11 items. bOrganizationQ (ORG), bEntertainmentQ (ENT) and bInformativenessQ (INFO) of the site (Chen and Wells, 1999), contained 2, 4 and 3 items. bStructureQ (STR) and bEffectiveness of the information contentQ (EFF) of the site (Bell and Tang, 1998) had 5 and 10 items. bSite AttitudesQ (ATTI; Eighmey, 1997) had 10 items. bExploratory behaviorQ (EXPB; Novak and Hoffman, 1997, 2000) had 8 items. bSite involvementQ (SINV; Muehling et al., 1990) had 8 items. bInvolvement in purchase decisionsQ (PPURI; Gore et al., 1994) for nonprescription drugs had 7 items. Purchase intentions (PURI) had one indicator.

5. Results 5.1. Exploratory factor analysis (EFA) EFA determines how observed variables are linked to their underlying factors. Since constructs could have items loading

5

on more than one factor, the minimal number of factors were identified. Analysis of the scales resulted in deleting items presenting poor psychometric properties or changes in Cronbach alphas. After deletion, each construct was unidimensional and factorially distinct, all items used to operationalize a construct loaded on one factor. The percentage of variance varied between 53.6% for PPURI and 88.4% for ORG (Table 1). EFA provided 10 factors with eigenvalues greater than 1.0, with each item having a loading greater than 0.4. For Nunnally (1967), acceptable Cronbach’s a start at 0.60. All Cronbach’s coefficients were equal (EXPB: a=0.69) or greater than 0.70, indicating good reliability. 5.2. Confirmatory factor analysis (CFA) CFA confirmed the measurement model (Byrne, 1994). The Lagrange Multiplier test identified a few covariances, which were taken into account. The 10-factor structure was confirmed with a first-order CFA. Estimation of the CFA model generated a v 2, comparative fit indices (CFI) and standardized root mean-square error of approximation (RMSEA) of 611.91 (df=446, v 2/df=1.372), 0.976 and 0.038. According to Hu and Bentler’s (1999), the model demonstrated a good fit (Baumgartner and Homburg, 1996). For Fornell and Larcker (1981), convergent validity is established if average variance extracted (AVE) accounts for 0.50 or more of total variance. In Table 2, except for involvement in purchase decisions, AVE varies from 0.53 (exploratory behavior) to 0.77 (organization). Convergent validity was confirmed for all constructs, except involvement in purchase decisions. Discriminant validity is the degree to which measures of different constructs are unique from each other. It is established if AVE is larger than the squared correlation coefficients between factors (Fornell and Larcker, 1981). This criterion was met across all pairs of factors.

Fig. 2. Model of the impact of internet atmospherics on surfer behavior.

ARTICLE IN PRESS 6

M.-O. Richard / Journal of Business Research xx (2004) xxx–xxx

Table 1 Exploratory factor analysis Constructs

Items

% of variance

Factor loadings

Cronbach alpha

Navigational cues

It is easy to use. Navigational problems are limited. There are good search agents to find information. Easy keywords to find information are used. The structure is well-organized. It allows a great overview of its structure. The structure is straightforward. Confusing sitea Irritating sitea Exciting site Imaginative site Entertaining site Informative site Useful site Resourceful site Information is accurate. Information is up-to-date. Product information is complete. The web site makes it very easy for me to build a relationship with the company. Surfing this Web site is an excellent way for me to spend my time. I was smiling while I was exploring this Web site. I was part of a like-minded group of people while using this Web site. This Web site was a playful experience. When I hear about a new Web site, I’m always eager to check it out. I like to browse the Web and find out about the latest sites. Unimportant/Important. . . to me Not worth/Worth. . . remembering Irrelevant/Relevant. . . to my needs Not worth/Worth. . . paying attention to It takes a very long time to decide before buying drugs. I get as much information as possible before purchasing a drug. I always compare product characteristics among brands of a specific drug. Before looking at this site, I was interested in reading about the needed drug. Definitively not willing/Extremely willing. . . to buy

67.2

0.734 0.777 0.864 0.809 0.800 0.818 0.775 0.886 0.886 0.769 0.758 0.825 0.866 0.858 0.721 0.719 0.846 0.692 0.691 0.690 0.762 0.744 0.787 0.826 0.810 0.782 0.811 0.814 0.810 0.709 0.715 0.734 0.849

0.83

Structure

Organization Entertainment

Informativeness

Information content effectiveness Site attitudes

Exploratory behavior Site involvement

Involvement in purchase decisions Purchase intentions a

76.2

88.4 76.2

80.5

69.3

63.3

76.6 72.7

53.6

0.84

0.87 0.84

0.88

0.78

0.85

0.69 0.87

0.71

Reversed scale.

5.3. Full structural model The results show strong support for the full structural model with CFI=0.976 and RMSEA=0.039. The average off-diagonal value of the standardized residual matrix was 0.046. As per Hu and Bentler’s (1999), the fit of this model is judged acceptable.

Following Byrne (1994), we tested the significance of individual parameters. The results of factor loadings, and the test statistics indicate that all loadings are significant. Finally, we analyzed the path coefficients representing the hypothesized relationships between the various constructs. Table 3 and Fig. 2 provide the standardized values of the regression coefficients and relate the paths to the original ten

Table 2 Convergent and discriminant validity ORG ENT STR INFO EFF CHPS SINV PPURI ATTI EXPB

ORG

ENT

STR

INFO

EFF

CHPS

SINV

PPURI

ATTI

EXPB

0.77 0.00 0.20 0.09 0.04 0.05 0.04 0.01 0.00 0.01

0.64 0.08 0.19 0.12 0.04 0.14 0.06 0.38 0.15

0.65 0.22 0.22 0.15 0.05 0.03 0.06 0.02

0.72 0.29 0.12 0.09 0.01 0.05 0.03

0.54 0.10 0.10 0.10 0.14 0.19

0.57 0.03 0.02 0.02 0.06

0.66 0.22 0.24 0.09

0.39 0.13 0.07

0.54 0.25

0.53

Diagonal entries show index of the average variance extracted. Entries below diagonal represent squared correlation coefficients.

ARTICLE IN PRESS M.-O. Richard / Journal of Business Research xx (2004) xxx–xxx Table 3 Regression coefficients’ estimates H CHPSYATTI CHPSYEXPB CHPSYPURI STRYSINV STRYEXPB STRYPURI EFFYATTI EFFYSINV EFFYEXPB EFFYPURI INFOYATTI INFOYEXPB ORGYSINV ORGYATTI ENTYATTI ENTYSINV ENTYEXPB ENTYPURI ATTIYPPURI ATTIYPURI EXPBYATTI EXPBYSINV EXPBYPPURI SINVYATTI SINVYPPURI SINVYPURI PPURIYPURI

H1a H1b H1c H2a H2b H2c H3a H3b H3c H3d H4a H4b H5a H5b H6a H6b H6c H6d H7a H7b H8a H8b H8c H9a H9b H9c H10

Parameter estimate NS ** ** NS NS ** NS * (one-way) **** ** * ** ** NS **** **** *** ** (one-way) NS NS ** NS NS **** **** **** *

Standard Test Standardized error statistic estimate 0.107 0.161 0.146 0.155 0.134 0.121 0.115 0.172 0.168 0.136 0.094 0.132 0.107 0.062 0.097 0.127 0.116 0.125 0.082 0.128 0.083 0.122 0.072 0.050 0.053 0.079 0.150

0.682 1.689 2.072 0.198 0.918 1.958 1.319 1.524 3.592 2.110 1.648 1.946 2.117 0.173 4.992 3.540 3.179 1.762 1.106 0.195 2.486 1.047 1.057 3.659 3.651 6.111 1.635

0.046 0.152 0.126 0.018 0.095 0.135 0.126 0.151 0.448 0.158 0.149 0.221 0.172 0.012 0.464 0.299 0.317 0.139 0.121 0.017 0.231 0.099 0.113 0.262 0.376 0.460 0.121

* Significant at pb0.1. ** Significant at pb0.05. *** Significant at pb0.01. **** Significant at pb0.001.

hypotheses. Nine out of twenty seven subsections of hypotheses have non-significant paths. All the others are significant with t-tests varying from 1.52 to 6.11.

7

pb0.1). H3c and H3d are supported: we found significant effects on exploratory behavior ( pb0.001) and purchase intentions ( pb0.05). H4a and H4b are partly supported with significant negative links between informativeness and site attitudes (t= 1.65, pb0.1) and site involvement (t= 1.95, pb0.05). Concerning organization of the Web site, H5a was supported as we found a significant link with site involvement ( pb0.05), whereas H5b was not. Hypothesis 6 is supported. There is a significant relationship between entertainment and site attitudes (H6a: pb0.001), site involvement (H6b: pb0.001), and exploratory behavior (H6c: pb0.01). There is a significant link with purchase intentions (H6d) with a one sided-test (t=1.76, pb0.05). For Hypothesis 7, contrary to previous findings, we found no relationships between site attitudes and involvement in purchase decisions (H7a) or purchase intentions (H7b). Support for Hypotheses 8 is mixed. Surfers’ exploratory behavior is significantly linked to site attitudes (H8a: pb0.05), whereas there is no significant link with site involvement (H8b) and involvement in purchase decisions (H8c). Hypothesis 9 related to site involvement is supported. There are significant (all pb 0.001) links with site attitudes (H9a), involvement in purchase decisions (H9b), and purchase intentions (H9c). For H10, surfers’ involvement in purchase decisions is negatively related to purchase intentions (t= 1.64, pb0.1). Overall, the results are encouraging, with full or partial support for several sets of hypotheses. This study was exploratory as no specific information was available on many paths. However, two-thirds of the proposed individual relationships were supported, and the model fitted the data very well.

6. Test of hypotheses 7. Interpretation and discussion of findings We briefly discuss support for individual hypotheses. The results are mostly supported for the set of Hypotheses 1 related to navigational characteristics. There is a significantly positive relationship with purchase intentions (H1c: pb0.05). It is significantly related to exploratory behavior (H1b), when a one-sided t-test is used (t=1.69, pb0.05). H1a, related to site attitudes, is not supported. We found partial support for the set of Hypotheses 2 related to site structure. H2a and H2b are not supported (no link with site involvement or exploratory behavior). However, a positive path with purchase intentions was found (H2c: pb0.05). The set of Hypotheses 3 related to information content effectiveness is mostly supported. H3a, its effect on site attitudes, is not supported. H3b, its effect on site involvement is marginally significant with a one sided-test (t=1.52,

The most interesting findings relate to effectiveness of information content. This central cue impacts site involvement (Yoon, 2000), exploratory behavior and purchase intentions (Okasaki and Rivas, 2002). Indirectly, it is positively related to site attitudes, with exploratory behavior mediating this relationship. First, we infer that when information content is effective, surfers are more involved in the site. The more it is adequate to visitors, the more they are engaged in in-depth information search. Surfers exerted cognitive efforts and used the central route, according to ELM. They became keener to search for information (exploratory behavior), affecting positively their attitudes, without being involved or to intend and purchase the OTC drug: they may be labeled information seekers. Site involvement had an impact on pre-purchase and purchase intentions. But of more interest is that site

ARTICLE IN PRESS 8

M.-O. Richard / Journal of Business Research xx (2004) xxx–xxx

involvement is highly related to approach attitudes. Normally, highly involved surfers are attracted by product aspects (information content effectiveness), whereas lowly involved ones focus on peripheral stimuli (entertainment) or some of the site’s characteristics (organization). We infer that highly involved surfers develop positive site attitudes leading to behavior such as repeat visits to collect up-to-date information. Obviously, this carries significant theoretical and managerial implications. Site involvement has a positive impact on involvement in purchase decisions. The more involved surfers are, the more they search for information before purchase, process relevant information in depth and use more criteria in their purchase decisions than others. Moreover, internetinvolved consumers more likely purchase online than those with low-levels of site involvement. Informativeness had a negative relationship with site attitudes. Even though there is updated, accurate and/or complete information, surfers who considered the site as informative, useful, and/or resourceful liked it less than those who did not, inducing the development of avoidance attitudes. Central cues such as effectiveness of information content and informativeness impact exploratory behavior. When the information presented suited visitors and is accurate, complete and up-to-date (content), they scroll and browse in order to gather more information. However, when the site is useful or resourceful to them (informativeness), this relationship is negative, which means that, as expected, informativeness reduced the extent of browsing. According to ELM, these variables make surfers follow the central route, exerting some cognitive efforts. We found that efficient navigational characteristics help develop exploratory behavior (Menon and Kahn, 2002). When the site is easy to use, with few navigational problems and good search agents, surfers spend more time investigating the site, developing positive site attitudes (Baronas and Louis, 1988; Eagly and Chaiken, 1993). Although structure is a central cue, it is not related to behavioral variables, which might due to the variance being explained by other central cues. The simple or complex level of structure does not influence browsing or scrolling, contrary to our hypothesis. Surfers do not exert additional cognitive effort to collect information because the structure is more complex. However, it is a central cue which could participate in the development of purchase intentions. Finally, when the site is confusing or irritating, in addition to other central cues, surfers become less involved in its visit because the organization does help them satisfy their information needs, they are likely to develop avoidance behavior and leave the site. However, surfers who did not give much attention to contents are attracted by visuals, giving them the opportunity to pursue the visit because of the entertainment value of the site. Visitors, who are affected by the visuals, elaborate affective reaction and follow the peripheral route. When a site is entertaining, surfers develop more arousal

and/or pleasure early in online browsing, becoming more involved in the site (Park and Young, 1986) and keener to search for information (exploratory behavior), affecting positively their attitudes. In this situation, we consider these visitors as browsers. Their behavior is experiential, they ask for recreational activity and nonlinear and rather ongoing research (Novak et al., 2003), contrary to informationseekers. Moreover, entertainment is directly linked to purchase intentions, confirming the impulse buying found by Novak et al. (2003). Whether or not surfers look for information, both goaldirected and experiential behaviors are involving (Schlosser, 2003). When the site matches the surfers’ goals, they engage in cognitive elaboration and develop favorable attitudes (Schlosser, 2003). Our results do not support prior findings that site attitudes have a positive effect on involvement in purchase decisions (Shim et al., 2001). Also, we did not find a relationship between site attitudes and purchase intentions, whereas prior research did (Stevenson et al., 2000; Notani, 1997). We infer that surfers’ goals were purely for entertainment or information search rather than intending to purchase. Without the benefit of prior findings, we did not find significant links between surfers’ exploratory behavior and both site involvement and involvement in purchase decisions. However, we found a significant path between exploratory behavior and site attitudes. The more surfers scroll and browse, the more they like information about the topic, producing approach behavior. The path between involvement in purchase decisions and purchase intentions is marginally significant and negative. It is reasonable for a visitor to be involved in the search for information, and after collecting more information postponing their intentions to buy the product.

8. Theoretical and managerial implications Our goal was to examine the impact of Internet atmospherics on surfer behavior. These findings provide behavioral scientists a better understanding of Internet surfer behavior. The theoretical implications take several forms. First, it confirmed several relationships discussed in prior literature, such as: (1) navigational cues positively influencing purchase intentions (Lynch et al., 2001); before developing intentions, surfers navigate through the site or find more information via other channels; (2) a direct link between site involvement and purchase intentions (Kwak et al., 2002); (3) site involvement and attitudes display a significant relationship (Yoo and Stout, 2001). Also, Bruner and Kumar (2000) found that site attitudes impact purchase intentions. But Kwak et al. (2002) found that online ad attitude did not impact Internet purchase. Our results supported Kwak et al. (2002), with site attitudes not

ARTICLE IN PRESS M.-O. Richard / Journal of Business Research xx (2004) xxx–xxx

affecting involvement in purchase decisions and purchase intentions. De Pelsmacker et al. (1998) reported a significant link between effectiveness of information content and site attitudes, whereas, in our study, it is mediated by exploratory behavior. More importantly, we tested relationships that were never studied previously: impact of structure, effectiveness of information content, and informativeness on exploratory behavior; structure and involvement in purchase decisions on purchase intentions; exploratory behavior on site involvement and involvement in purchase decisions; and attitude on involvement in purchase decisions. Finally, our study tested the ELM (Petty and Cacioppo, 1986) in an Internet context. All high task-relevant atmospherics cues were found to have an impact, with the main flow going to exploratory behavior, which affected site attitudes, while in parallel they also impacted purchase intentions. This complex pattern indicates the use of the central route by information seekers. The low taskrelevant cue, entertainment, had the strongest impact on site attitudes (directly and through site involvement), reflecting the use of the peripheral route by browsers/ entertainment seekers; interestingly, entertainment had a secondary impact on exploratory behavior and purchase intentions, reflecting the need for some entertainment by information seekers. This research provides marketers with insights into variables that influence consumers’ purchase intentions for an OTC drug when they use the Internet to seek and collect information. The findings indicate what types of navigational cues, which kinds of structure and effective information content are more likely to involve surfers in seeking product-related information from the Internet. Given the early positioning of information search in the decision process, if marketers can identify which consumer segments rely on navigational characteristics, information content and structure of the site or on entertainment, how to decrease the difficulty of navigating their site, and how to build a site with a structure and information convenient and appealing to their needs, they can better tailor their communication strategies. Their behavior could positively change, with more exploratory behavior, approach behavior and site involvement, ending in purchase. Marketers should study emotions of surfers when they first visit a website as these may affect their behavior. When a marketer builds sites for tasks such as purchasing, registering, doing something that asks for an immediate reaction, deeper browsing or exploration not required or not desired, the web site could be designed with arousing stimuli. If the site is arousing, consumers are more likely to complete their tasks and less likely to search for other stimulation or browse other sites. If marketers want their visitors to stay longer, browse and explore different links, they might create pleasing, enjoyable stimuli to encourage browsing and impulse buying (Menon and Kahn, 2002).

9

9. Limitations and future research The study is not without limitations. Empirical surveys on the Internet may be questioned on external validity, more so since respondents surfed one site and were students with Internet experience. The addition of emotions might have helped explain some paths found non-significant or through indirect paths. Some other areas for future research can be suggested: measure and add attitude toward the brand to have a link with involvement in purchase decisions and purchase intentions; add trust and experience as they influence attitudes and purchase intentions; use metrics for evaluating environmental cues such as stickiness (average time per visit, frequency and recency) (Bhat et al., 2002). Knowing how consumers consider an OTC drug (search or experience product) will help in further testing ELM in a web context. Moreover, this model could be applied to other products necessitating information seeking. Finally, longitudinal studies could trace the evolution and adaptation of consumer behavior when technological developments are brought into navigational characteristics, adding visual and audio capabilities and improving the quality of the information found on the Web.

Acknowledgements The author gratefully acknowledges the technical assistance of Isabelle Miodek and Zhiyong Yang and the helpful comments of Mitch Griffin and two anonymous reviewers. An earlier version of this paper won the 2004 Jane K. Fenyo Award of the Academy of Marketing Science.

References Baker J, Grewal D, Parasuraman S. The influence of store environment on quality inferences and store image. J Acad Mark Sci 1994;22:328 – 39 [Autumn]. Balabanis G, Reynolds NL. Consumer attitudes towards multi-channels retailers’ web sites: the role of involvement, brand attitude, internet knowledge and visit duration. J Bus Strategies 2001;18(2):105 – 32. Baronas AK, Louis MR. Restoring a sense of control during implementation. MIS Q 1988;12:111 – 23. Baumgartner H, Homburg C. Applications of structural equation modeling in marketing and consumer research: a review. Int J Res Mark 1996;13:139 – 61. Baumgartner H, Steenkamp JBEM. Exploratory consumer behavior: conceptualization and measurement. Int J Res Mark 1996;13(2):121 – 37. Bell H, Tang NKH. The effectiveness of commercial internet web sites: a user’s perspective. Internet Res 1998;8(3):219 – 28. Bellman S, Lohse GL, Johnson EJ. Predictors of online buying behavior. Comm of ACM 1999;42(12):32 – 8. Berlyne DE. Motivational problems raised by exploratory and epistemic behavior. In: Koch S, editor. Psychology: A Study of Science, vol. 5. New York7 McGraw-Hill; 1963. p. 284 – 364. Bhat S, Bevans M, Sengupta SL. Measuring users’ web activity to evaluate and enhance advertising effectiveness. J Advert 2002;31(3):97 – 106. Bloch PH, Sherrell DL, Ridgeway NM. Consumer search: an extended framework. J Consum Res 1986;13(1):119 – 26.

ARTICLE IN PRESS 10

M.-O. Richard / Journal of Business Research xx (2004) xxx–xxx

Br7nnback M. Is the internet changing the dominant logic of marketing? Eur Manag J 1997;15(6):698 – 707. Brown SP, Stayman DM. Antecedents and consequences of attitudes toward the ad: a meta analysis. J Consum Res 1992;19:34 – 51 [June]. Bruner II GC, Kumar A. Web commercials and advertising hierarchy-ofeffects. J Advert Res 2000;40(1/2):35 – 42. Byrne BM. Structural equation modeling with EQS and EQS/Windows: basic concepts, applications, and programming. Sage; 1994. Chen Q, Wells WD. Attitude toward the site. J Advert Res 1999;39(5): 27 – 37. Chen Q, Clifford SJ, Wells WD. Attitude toward the site II: new information. J Advert Res 2002;42(2):33 – 45. Cho CH. How advertising works on the world wide web: modified elaboration likelihood model. J Curr Issues Res Advert 1999;21(1): 33 – 49. Csikszentmihalyi M. Beyond boredom and anxiety. In: Csikszentmihalyi M, editor. San Francisco (CA)7 Jossey-Bass; 1977. Dailey L. Navigational web atmospherics: explaining the influence of restrictive navigation cues. J Bus Res 2004;57(7):795 – 803. Day E, Stafford MR, Camacho A. Research note: opportunities for involvement research—a scale-development approach. J Advert 1995;24(3):69 – 75. De Pelsmacker P, Dedock B, Geuens M. A study of 100 likeable TV commercials: advertising characteristics and the attitude towards the ad. Mark Res Today 1998;27(4):166 – 79. Dholakia UM, Rego LL. What makes commercial web pages popular? Eur J Mark 1998;32(7/8):724 – 36. Dodds WB. In search of value: how price and store name information influence buyers’ product perception. J Serv Mark 1991;5(3):27 – 36. Donovan RJ, Rossiter JR. Store atmosphere: an environmental psychology approach. J Retail 1982;58(1):34 – 57. Donovan RJ, Rossiter JR, Marcoolyn G, Nesdale A. Store atmosphere and purchase behavior. J Retail 1994;70:283 – 94. Eagly AH, Chaiken S. The psychology of attitudes. Fort Worth (TX)7 Harcourt Brace Jovanovich; 1993. Eighmey J. Profiling user responses to commercial web sites. J Advert Res 1997;37(3):59 – 66. Eroglu SA, Machleit KA, Davis LM. Atmospherics qualities of online retailing: a conceptual model and implications. J Bus Res 2001;50: 177 – 84. Fornell C, Larcker DF. Evaluating structural equation models with unobservable variables and measurement error. J Mark Res 1981;18: 39 – 50 [February]. Gore P, Madhavan S, McClung G, Riley D. Consumer involvement in non prescription medicine purchase decisions. J Health Care Mark 1994; 14(2):16 – 20. Hoffman DL, Novak TP. Marketing in hypermedia computer-mediated environments: conceptual foundations. J Mark 1996;60(3):50 – 68. Homer PM. The mediating role of attitude toward the ad: some additional evidence. J Mark Res 1990;27:78 – 86 [February]. Hu L, Bentler PM. Cutoff criteria for fit indexes in covariance structure analysis: conventional criteria versus new alternatives. Struct Equ Modeling 1999;6(1):1 – 55. Huizingh EK. The content and design of web sites: an empirical study. Inf Manage 2000;37(3):123 – 34. Jarvenpaa S, Todd P. Consumer reactions to electronic shopping on the world wide web. Int J Electron Commer 1997;1(2):59 – 88. Jee J, Lee WN. Antecedents and consequences of perceived interactivity: an exploratory study. J Interact Advert 2002;3(1):1 – 16 [jiad.org/ vol3/no1/jee]. Kotler P. Atmosphere as a marketing tool. J Retail 1973;49:48 – 64 [Winter]. Kwak H, Fox RJ, Zinkhan GM. What products can be successfully promoted and sold via the internet? J Advert Res 2002;42(1):23 – 38. Larkin EF. Consumer perceptions of the media and their advertising content. J Advert 1979;8(2):5 – 7.

Leong SM. Consumer decision making for common, repeat-purchase products: a dual replication. J Consum Psychol 1993;2(2):193 – 208. Lutz RJ, MacKenzie SB, Belch GE. Attitude toward the ad as a mediator of advertising effectiveness: determinants and consequences. In: Bagozzi RP, Tybout AM, editors. Advances in Consumer Research, vol. 10. Ann Arbor (MI)7 Association for Consumer Research; 1983. p. 532 – 9. Lynch PD, Kent RJ, Srinivasan SS. The global internet shopper: evidence from shopping tasks in twelve countries. J Advert Res 2001;41(3):15 – 23. Maddox K. Surveys show increase in online usage, shopping. Advert. Age 1998;69(43):ps6, 2p. Maheswaran D, Meyers-Levy J. The influence of message framing and involvement. J Mark Res 1990;27:361 – 7 [August]. McGaughey RE, Mason KH. The Internet as a marketing tool. J Mark Theory Pract 1998;6(3):1 – 11. McQuail D. Mass communication theory: an introduction. London7 Sage; 1983. Mehrabian A, Russell JA. The basic emotional impact of environments. Percept Mot Skills 1974;38:283 – 301. Menon S, Kahn B. Cross-category effects of induced arousal and pleasure on the internet shopping experience. J Retail 2002;78(1): 31 – 40. Milliman RE, Fugate DL. Atmospherics as an emerging influence in the design of exchange environments. J Market Manag 1993;3:66 – 74 [Spring/Summer]. Muehling DD, Stoltman JJ, Grossbart SL. The impact of comparative advertising on levels of message involvement. J Advert 1990;19(4): 41 – 50. Notani AS. Perceptions of affordability: their role in predicting purchase intent and purchase. J Economic Psychol 1997;18(5):525 – 46. Novak TP, Hoffman DL. New metrics for new media: toward the development of web measurements standards. World Wide Web J 1997;2(1):213 – 46. Novak TP, Hoffman DL, Yung YF. Modeling the flow construct in online environments: a structural modeling approach. Mark Sci 2000;19(1): 22 – 42. Novak TP, Hoffman DL, Duhachek A. The nature of flow experiences on the web. J Consum Psychol 2003;13(1/2):3 – 16. Nunnally JC. Psychometric theory. New York7 McGraw-Hill; 1967. Obermiller C, Bitner MJ. Store atmosphere: a peripheral cue for product evaluation. In: Stewart DC, editor. APA Conference Proceedings, vols. 52–53. Washington7 American Psychological Association; 1984. Okasaki S, Rivas JA. A content analysis of multinational’s web communication strategies: cross-cultural research framework and pretesting. Internet Res 2002;12(5):380 – 90. Park CW, Young SM. Consumer response to television commercials: the impact of involvement and background music on brand attitude formation. J Mark Res 1986;23:11 – 24 [February]. Pashupati K. The dimensions of attitude toward the ad: a re-exploration. In: Macklin B Carle, editor. Proceedings of the American Academy of Advertising. Cincinnati, OH7 U of Cincinnati; 1997. p. 107. Petty RE, Cacioppo JT. Attitudes and persuasion: classic and contemporary approaches. Dubuque (IA)7 William Brown; 1981. Petty RE, Cacioppo JT. The elaboration likelihood model of persuasion. In: Berkowitz L, editor. Advances in Experimental Social Psychology, vol. 22. New York7 Academic Press; 1986. p. 123 – 205. Petty RE, Cacioppo JT, Schumann D. Central and peripheral routes to advertising effectiveness: the moderating role of involvement. J Consum Res 1983;10(2):135 – 46. Raman NV, Leckenby JD. Factors affecting consumers’ web ad visits. Eur J Mark 1998;32(7/8):737 – 48. Ranganathan C, Ganapathy S. Key dimensions of business-to-consumer web sites. Inf Manage 2002;39(6):457 – 65. Rowley J. Product search in e-shopping: a review and research propositions. J Consum Mark 2000;17(1):20 – 35. Schlosser AE. Experiencing products in the virtual world: the role of goal and imagery in influencing attitudes versus purchase intentions. J Consum Res 2003;30(2):184 – 98.

ARTICLE IN PRESS M.-O. Richard / Journal of Business Research xx (2004) xxx–xxx Shim S, Eastlick MA, Lotz SL, Warrington P. An online prepurchase intentions model: the role of intention to search. J Retail 2001;77(3): 397 – 416. Stern BL. Pleasure and persuasion in advertising: rhetorical irony as a humor technique. J Current Issues Res Advert 1990;12:25 – 42. Stevenson JS, Bruner II GC, Kumar A. Web page background and viewer attitudes. J Advert Res 2000;40(1/2):29 – 34. Turley LW, Milliman RE. Atmospherics effects on shopping behavior: a review of the experimental evidence. J Bus Res 2000;49:193 – 211.

11

Yoo CY, Stout PA. Factors affecting users’ interactivity with the web site and the consequences of users’ interactivity. In: Taylor CR, editor. Proceedings. Villanova (PA)7 American Academy of Advertising; 2001. p. 53 – 61. Yoon D. Use of endorsers in internet advertising: a content analysis of top 100 American advertisers web pages. In: Shaver MA, editor. Proceedings. East Lansing (MI)7 American Academy of Advertising; 2000.