BRAND EQUITY IN INTERNET BUSINESS: AN EXPLORATORY STUDY Kai-Lung Hui, Yah-Ting Gwee Department of Information Systems National University of Singapore Singapore [email protected], [email protected] Patrick Y.K. Chau School of Business The University of Hong Kong Hong Kong [email protected] November 2003 Abstract Previous research suggests that brand equity is shaped by various marketing mix elements, such as store image or distribution intensity. However, because the Internet has no physical boundary, some of these traditional marketing variables may not be applicable in shaping the brand equity of online firms. Drawing on theories in economics, marketing and psychology, we develop a model of brand equity formation that pertains to Internet businesses. Inside our model, we explore the significance of several Web-based marketing variables that could affect consumer perceptions of product quality and brand knowledge, both of which are relevant components of brand equity. By performing structural equation and regression analyses on two sets of survey data, we find that product/technology innovation and Website quality contribute to improving an online firm’s brand equity. The effect of advertising intensity on perceived quality is mixed, but advertising always improves the brand knowledge of consumers. These results shed important strategic implications for online firms who are building up new brands on the Internet. Keywords: value-added features, product/technology innovation, Website quality, advertising intensity, perceived quality, brand knowledge, brand equity We gratefully acknowledge the helpful comments from Bob Zmud, Bernard Tan, three anonymous reviewers and the seminar participants at the 2002 International Conference on Information Systems. 1 I. INTRODUCTION The widespread acceptance of the Internet and electronic commerce has given rise to popular brand names such as Amazon, eBay, Hotmail and Yahoo. Although the Internet is frequently characterized as an efficient channel that reduces search and transaction costs, it also intensifies product and price competitions due to the use of intelligent tools, such as comparison-shopping agents or interactive decision aids (Haubl and Trifts 2000). To differentiate from competitors and appeal to online consumers, other than continuously improving products and setting appropriate prices, an online firm often needs to initiate innovative practices, employ advanced technologies, or offer high-quality services that create extra value to consumers (Willcocks and Plant 2001). These efforts may help form unique and favorable brand impressions. A brand is a name and/or symbol like logo, trademark and package design that uniquely identifies products or services of a seller, and differentiates them from those of its competitors (Aaker 1991; Keller 1998; Kotler 1994). Brand is valuable because it influences consumer preferences. A good brand can signal product superiority to consumers, which may subsequently lead to favorable consumer attitude that, in turn, brings in better sales and financial performance for the firm (Aaker and Jacobson 2001; Erdem and Swait 1998). Therefore, a brand may enhance the perceptual value of its products to consumers. Such an extra value (cf. values provided by intrinsic product attributes) due to the brand is commonly called brand equity (Farquhar 1989). For commercial firms, the importance of building and cultivating brand equity is best epitomized by the frequent sales and acquisitions of reputable brands, which typically involve substantial monetary premiums (Keller 1998). Keller (1993) suggests that a brand is built by the creation of firm-associated mental structure in consumers’ memory, which helps consumers organize their knowledge in a way that aids their brand selection strategies and decisions. For instance, being the first online bookstore that carries a wide catalog, Amazon has received substantial recognition in online book retailing. Despite the fierce competition posed by hundreds of other Internet bookstores, Amazon has been persistently the leader in the online book retail industry, with an overwhelming market share lead over all online competitors.1 Similarly, owing to its innovative business concept, Priceline.com 1 See Karen J. Bannan “Book Battle,” Mediaweek, vol. 10, no. 9, February 2000, pp. 72-76; Jim Milliot “Chain sales rose 4.8% in first quarter, to $1.7 billion,” Publishers Weekly, vol. 249, no. 21, May 2002, pp. 9; and the company profiles in Hoovers Online (http://www.hoovers.com). It is instructive to observe that the leading brick- 2 has gained much attention and recognition among consumers, which helps create a reputable brand for its Internet travel-booking service. Clearly, brand may continue to play an important role in electronic commerce (Willcocks and Plant 2001). Prior research has identified various marketing mix elements (such as pricing, promotion, store image, advertising and distribution intensity) that may affect consumer memory perception of brands (Keller 1993; Yoo et al. 2000). However, because Internet stores have no physical boundary and are constantly available, some of these marketing mix elements may not be relevant. For example, store image needs to be redefined for online firms because the “stores” are now replaced by a group of related Webpages. Similarly, in online retailing, distribution intensity is not relevant because the Internet is ubiquitous – consumers can access online stores anytime from anywhere, and most online stores deliver products to buyers in other geographical regions. Therefore, the brand equity of an online firm is possibly shaped by a different set of marketing factors pertaining more to Website characteristics and technological attributes. It is important for firms to recognize and study the emerging factors that help construct their brand equity in the online context. In this research, we explore factors that affect consumer perception of online firms and test a model of brand equity formation that applies to Web-based industries. Specifically, using brand knowledge and perceived quality as separate components of brand equity, we investigate whether Web-based marketing efforts could affect consumers’ brand perceptions. The selected marketing mix factors (value-added features, product/technology innovation, Website quality and advertising intensity) are generic across business sectors, and they form the basic building blocks of many business-to-consumer (B2C) electronic storefronts. Hence they have wide applicability and should be of interest to Website managers who are eager to establish unique brand identities among Internet consumers. We conducted two studies to validate the model and test the theoretical hypotheses about the studied Web-based factors. In the first study, we surveyed a group of university students about their perception of various Websites’ attributes, and used the data to refine measurement items and test the hypotheses. Then, we repeated the survey with another sample of working students, and further examined the detailed properties/contributions of several constructs. The and-mortar bookstore, Barnes and Noble, only has a market share of around 40 percent in the conventional book retail industry, whereas Amazon alone captures close to 80 percent of the online market. 3 studied Websites (online firms) were carefully chosen from three service industries such that they do not have any physical retail outlets. This was to avoid possible confounds due to existing brick-and-mortar operation or a priori brand advantage. We selected Websites from three industries and tested the hypotheses using two groups of subjects with different demographics to ensure external validity and generalizability of our findings. Our results strongly suggest the significance of several Web-based marketing factors in shaping brand equity (through enhancing brand knowledge and perceived quality). Both Website quality and product/technology innovation are positive attributes that improve brand equities of Internet businesses. The effect of advertising intensity is mixed, whereas that of value-added features is not statistically significant. The proposed model explains substantial variations of the studied firms’ brand equities. This confirms our conjecture that Website- or technology-related characteristics are important for understanding online firms’ brand positions. The rest of this article is organized as follows. Section 2 sketches the theoretical model of brand equity, brand knowledge and perceived quality, and explains the rationales behind the testing of the selected Web-based marketing factors. Sections 3 and 4 highlight the objectives of the two studies, and present the research method and empirical results. Section 5 discusses the findings and implications, and concludes the paper. II. THEORY The extant literature has defined brand equity by multiple indicators or dimensions. Specifically, Aaker (1991, 1996) posits that brand equity is shaped by perceived quality, brand awareness, association, loyalty and objective indicators such as proprietary assets (patents, trademarks, etc.) or market share. Based on associative network memory models in cognitive psychology, Keller (1993) suggests that brand knowledge is key to conceptualizing and managing customer-based brand equity. From an information economics perspective, Erdem and Swait (1998) propose that brand equity can be characterized by clarity and credibility of brand signals. Other research has also measured brand equity using financial worth (Simon and Sullivan 1993) or consumer utility that is not accountable by physical attributes of the studied products (Kamakura and Russell 1993; Park and Srinivasan 1994; Swait et al. 1993). It is obvious that multiple dimensions can be used to conceptualize and measure brand equity. These can be broadly classified as either attitudinal or behavioral (Cobb-Walgren et al. 4 1995). Generally, behavioral indicators such as actual brand choice, loyalty or market share may reflect the salience of a brand’s equity, but they are not readily observable at the consumer level in most online environments. Instead, we focus on two attitudinal variables, brand knowledge and perceived quality, that are well conceived to be important in shaping customer-based brand equity (Aaker 1996; Keller 1993; Yoo et al. 2000), and that can be directly probed by consumer surveys. In subsequent discussions, we first conceptualize the relationship between brand equity, brand knowledge and perceived quality, and then explain why certain Web-based marketing variables should affect brand equity of Internet businesses. Brand Equity, Brand Knowledge and Perceived Quality When consumers are exposed to a particular brand, either through communications or direct experience with its products, they absorb certain information related to the brand and store such information in their memory. Such storage of information leads to the accumulation of brand knowledge. Keller (1993, pp.8) suggests that customer-based brand equity is formed when consumers’ brand knowledge generates differential effect on their response to the marketing of the brand. In this perspective, brand knowledge is conceptualized as nodes in consumer memory that are connected by associative links of various strengths to other memory nodes. The extent of information retrieval from memory depends on how strong the nodes are connected to each other and whether they are accessible by external cues (Raaijmakers and Shiffrin 1981; Ratcliff and McKoon 1988). High brand knowledge implies that the brand nodes are more salient and easily retrievable from memory, which directly increases the probability of consumers choosing the brand. Essentially, a brand adds value to its products if consumers have positive knowledge about it. This view of brand knowledge as a defining characteristic of brand equity is rooted in memory research in cognitive psychology (Keller 1993), and its basic principle is consistent with empirical findings in memory and decision-making studies (Alba et al. 1991). Economic human capital model also supports brand knowledge as a relevant dimension in characterizing brand equity (Ratchford 2001). The human capital model regards consumers as production units, who combine input goods, time and knowledge to produce household activities. The unique feature of this model lies in maximization of consumer utility over activities instead of goods. Because knowledge facilitates activity production, higher knowledge on a particular 5 brand implies a lower price of activities associated with the brand.2 Such a lower price, in turn, directly increases demand for the activities and indirectly raises demand for goods that carry the brand (as compared to rival brands that help produce similar activities, but that are less well known by the consumers). Therefore, a brand adds value to its products when consumers are familiar with it and perceive that they could use the products to efficiently generate activities. This perspective may help explain why consumers occasionally make decisions based on brand familiarity, and why knowledge could inhibit switching behavior (Alba and Hutchinson 1987; Jacoby et al. 1977; Ratchford 2001). To conclude, the cognitive psychological view of brand knowledge focuses on memory organization and salience of brand nodes, while the human capital model stresses the importance of brand knowledge in efficient production of consumer activities. Both theories provide a solid foundation for the instrumental role of consumer knowledge in defining brand equity. Besides brand knowledge, it is useful to observe that every product carries an intrinsic level of quality, which could be perceived differently because of its brand.3 Perceived quality is defined as consumer judgment of a product’s overall excellence or superiority (Zeithaml 1988, pp. 3). This definition highlights two characteristics of perceived quality. First, it is subject to consumer judgment and is distinct from objective quality. Second, it is conceived based on overall product excellence, and therefore it is a high-level abstraction that is more general than specific product attributes. Essentially, perceived quality is one form of overall evaluation that is analogous to attitude (Olshavsky 1985). This conceptualization is coherent with common quality definitions, and it was widely applied in consumer research (see, e.g., Aaker 1991; Erdem and Swait 1998; Holbrook and Corfman 1985; Yoo et al. 2000). Perceived quality affects consumer response, such as product choice or store patronage. Indeed, consumer perception of quality is often more superior than objective quality in predicting preference or behavior (Howard 1977; Jacoby and Olson 1985). Ceteris paribus, high perceived quality differentiates a product from competing offers (Erdem and Swait 1998; Yoo et al. 2000). This might translate into favorable reactions, such as purchasing or having a higher intention to choose a particular brand (Cronin et al. 2000; Zeithaml 1988). Further, high perceived quality 2 The price here represents the marginal cost of producing the activities, which may include time cost, knowledge contribution and the price of (input) goods that carry the brand (Ratchford 2001). 3 For ease of presentation, in subsequent discussion, the word “product” includes tangible products, intangible products (such as digital information), and offline or online services. 6 may act as an internal cue that inspires the superiority of a brand to consumers, and hence may raise consumer utility. These imply that perceived quality could also be used to conceptualize brand equity. Finally, perceived quality may relate to brand associations in consumer memory (Keller 1993). The brand nodes of low quality products are often less frequently activated; prior experiments have found quality to affect the association and recall of memory nodes that are reflective of family-brands (Janiszewski and Van Osselaer 2000). Hence perceived quality may interact with brand knowledge. The shaded parts of Figure 1 sketch the conceptual framework of brand equity that we adopt in this study. <Insert Figure 1 here> Exogenous Web-Based Factors Given that brand knowledge and perceived quality shape brand equity, it is important for online firms to devise an appropriate mix of marketing elements that could (perceptually) enhance their brands and product qualities. Since many Internet firms provide pure online services, traditional marketing mix elements may not be directly applicable. Instead, online firms need to appreciate new strategic variables that could differentiate them from competitors. We group the selected Web-based variables into three dimensions: product, place and promotion.4 Product: value-added features and product/technology innovation Because no physical salesperson is located in the “store” to offer assistance, many online firms provide peripheral tools or services to consumers during navigation, purchase or consumption of their products. We collectively refer such peripheral tools or services as “value-added features”. Common examples include provision of help information, instant product support, alert service, personalized update, and many others. These features help improve consumer experience with online firms, but they are typically not positioned as part of the firms’ products. Rather, they add value through enhancing products to consumers. 4 We excluded the price dimension, which is the other component in the traditional 4P classification of marketing mix instruments (Kotler 1994; Van Waterschoot and Van den Bulte 1992), because our studied firms offer free services to consumers. As mentioned earlier, we restricted our scope to pure online service firms because their brand equities are not susceptible to influences from brick-and-mortar operations. For instance, it is difficult to estimate whether the brand equity of Barnes and Noble comes from its large chain of retail stores or its Web storefront. 7 Prior research in electronic commerce has paid little attention to the effect of value-added features on consumer perception. The literature on service marketing provides insight into this area of study. Generally, consumers are found to appreciate extra service or special treatment by service personnel (Bitner et al. 1990; Ostrom and Iacobucci 1995). Bowen et al. (1989) suggest that augmented product, which includes customer services, could become a source of product advantage for service-oriented firms. Further, for postal service, Goodman et al. (1995) find that consumer perception of peripheral services could spillover to their perception of a firm’s core services. These imply that peripheral services directly affect consumers’ evaluation and quality perception of products. The benefits of peripheral services could potentially be extended to e-businesses. Prior research has found information technology an effective means for providing support and valueadded services to consumers (Haubl and Trifts 2000; Meuter et al. 2000; Nault and Dexter 1995). An online firm could achieve synergies by complementing its products with high-quality valueadded features, which may enhance consumers’ quality perception of the products (Willcocks and Plant 2001). Therefore, we hypothesize: H1: The quality of value-added features is positively related to perceived quality; better value-added features lead to a more favorable quality perception. Besides adding value through peripheral features, an online firm could also improve its corporate image by continuously investing in product or technological innovations. In this study, a firm is said to achieve product/technology innovation if it is quick in launching new products or deploying state-of-the-art technologies.5 Real life examples may include Priceline.com, which pioneered the “name your own price” practice in its online travel-booking service; and eBay, which introduced the proxy bidding technology and feedback mechanism in its Internet auction business. Our investigation of product/technology innovation is motivated by earlier research on market pioneer and innovator advantage. In general, studies have found that early movers enjoy initial and sustain continuous market share advantages. Although such advantages are partly due 5 We consider both product and technology innovation in a single construct because online firms, especially those providing Web-based services, often incorporate new technologies into their “products” and advertise them together. For consumers, the two types of innovations may appear inseparable. 8 to supply-side efficiency,6 Schmalensee (1982) proposes that an early mover with satisfactory product quality could create a perceptual reference point for consumers to make judgments. This implies that a firm that continuously promotes innovative products or technologies may be able to “preempt” a particular market by positioning itself as the de facto standard with ideal quality. Equivalently, innovation activities may improve consumers’ perceived familiarity and quality of a firm’s products. In a similar vein, Robinson and Fornell (1985) suggest that being an innovator or product leader gives consumers a favorable image and higher familiarity on a firm’s products. Carpenter and Nakamoto (1989) demonstrate that an early mover could shift consumer preference toward its products because of naïve learning. Consumers may assign disproportional attribute weights to an early brand, and the brand may become prototypical to them. These discussions lead to the following hypotheses. H2a: The degree of product/technology innovation is positively related to perceived quality; faster innovation leads to a more favorable quality perception. H2b: The degree of product/technology innovation is positively related to brand knowledge; faster innovation leads to higher brand knowledge. Place – Website quality In the Internet context, a Website is equivalent to a firm’s retail store.7 We define Website quality as the overall excellence or superiority of a firm’s Website. This includes features such as user interface, design layout and system performance.8 With the increasing exchange and marketing activities over the Internet, designing and maintaining a good Website is becoming important for an online firm to capture higher traffic and sales (Lohse and Spiller 1998). In conventional retailing, research has found that consumers sometimes evaluate products based on store information (Dodds et al. 1991). Drawing on theories related to inference, 6 Lieberman and Montgomery (1988) provide a detailed discussion on advantages enjoyed by early movers. For empirical evidence, see, for example, Dowling and McGee (1994), Henard and Szymanski (2001), Isobe et al. (2000), Robinson (1988, 1990) and Robinson and Fornell (1985). 7 Many pure-play online firms do not maintain physical outlets, and consumers can patronize them only through their corporate Websites. This is particularly the case for firms that offer online services. By nature, some online services, such as search engine or information portal, cannot be provided in a physical setting. 8 For more discussion on Website design and performance assessment, see McKinney et al. (2002), Palmer (2002) and Szymanski and Hise (2000). 9 schema formation and affordance, Baker et al. (2002) posit that consumers respond to design, social and ambient environment cues when evaluating stores and products. Their empirical results confirm that store design consistently affects consumer judgment of merchandise quality. Generally, a Website can be conceptualized in architectural terms that encompass major design and performance dimensions (Kim et al. 2002). Various characteristics of Websites, such as functional excellence or system interface, may assume the role of service speed, store layout and ambience in conventional retail stores. Recent research has found that Website quality affects consumer satisfaction of and loyalty toward online firms (McKinney et al. 2002; Palmer 2002; Srinivasan et al. 2002; Szymanski and Hise 2000). A careful manipulation of Webpages can influence consumer preferences (Mandel and Johnson 2002). Following the theory posited by Baker et al. (2002), Website quality may serve as an extrinsic cue that helps consumers infer the quality of a particular firm’s products. Our next hypothesis is posited as: H3: Website quality is positively related to perceived quality; a higher quality Website leads to a more favorable quality perception. Promotion – advertising intensity Advertising is perhaps one of the most popular means for promoting brands and products to consumers. We define advertising intensity as the extent to which consumers are exposed to advertisements of a particular brand across all marketing channels. On the Internet, the majority of firms increase advertising intensity through pop-up browser windows, banner advertisements or mass emails. Compared to traditional media such as TV or newspaper, advertising over the Internet has a global reach and higher interactivity, and it requires less resource input (i.e., lower expense) from advertisers. It is widely held that advertising contributes to the establishment of brand equity (Aaker and Biel 1993; Cobb-Walgren et al. 1995). Advertising affects consumer behavior through both cognitive and affective responses (Alba et al. 1991; Vakratsas and Ambler 1999). In general, an advertisement conveys brand and/or product information to consumers. Theories in information economics hypothesize that, for search goods, advertising may help consumers identify relevant product attributes and assess objective product quality; for experience goods, the fact that a firm 10 advertises may signal to consumers the superiority of its products (Nelson 1974).9, 10 Empirical research has repeatedly found advertising to affect perceived product quality. A reasonable level of advertising expense or repetition may lead to favorable quality evaluations (Kirmani 1997; Kirmani and Wright 1989). Although advertising may confer favorable quality signals, previous research has also found consumers to be skeptical of claims about experience attributes in advertisements (Ford et al. 1990). Kirmani and Wright (1989) conjecture that consumers might undermine a particular advertisement when its repetition appears to be excessive, or when advertising cost does not seem to be of concern to the advertiser.11 Their study (and, subsequently, Kirmani 1997) further demonstrates that advertising expense and perceived quality exhibit an inverted-U relationship. That is, an unusually high expenditure actually dampens quality perceptions. Further, in an extensive review of the advertising literature, Vakratsas and Ambler (1999) conclude that returns to advertising are diminishing, and consumers’ affective responses may lead to wear-out of heavily repeated advertisements (Alba et al. 1991). For online service firms, their products are generally more experiential in nature.12 Because advertising on the Internet generally incurs lower costs, when an online firm advertises heavily, the two “negative” images, desperation and no pain, put forward by Kirmani and Wright (1989) may counteract the positive effects suggested by the information economics perspective. It is difficult to envisage the net effect of advertising intensity on perceived product quality. As such, we pose our hypothesis in the following exploratory form: H4a: Advertising intensity is related to perceived quality. 9 In the information economics perspective, the value of advertising stems from the predisposition that consumers lack price or quality information, and they engage in different nature and extents of search activities. For more discussion, see Stigler (1961) and Nelson (1970, 1974). 10 In his discussion of experience goods, Nelson (1974) posits that a heavily advertised brand is likely to be more efficient, and it may offer a lower price per unit of utility to consumers. Therefore, higher advertising intensity may (indirectly) suggest to consumers that a brand is a better buy. 11 Kirmani and Wright (1989) coin them as “desperation” and “no pain” effects. The underlying thought is that no pain advertising dampens the quality signal proposed by the information economics literature, whereas desperation gives a reverse signal that the firm may actually lack confidence in its products. 12 For instance, consumers can hardly know the performance of a search engine before they actually conduct some searches; the connection speed, functionality and user-friendliness of a browser-based email service is generally not obvious to consumers prior to usage. 11 Finally, advertising directly confers brand and product information to consumers, and it serves to increase the awareness of and knowledge on a particular brand (Keller 1987; Nelson 1974; Stigler 1961; Vakratsas and Ambler 1999). Research in memory and consumer knowledge suggests that repeatedly showing an advertisement enhances a brand’s salience in consumers’ memory (Alba et al. 1991; Alba and Hutchinson 1987). Therefore, our last hypothesis is: H4b: Advertising intensity is positively related to brand knowledge; more exposure to advertisements leads to higher brand knowledge. Figure 1 depicts the full research model and the hypothesized relationships. We conducted two studies using different samples of subjects to validate and test the model. III. STUDY ONE The objective of this study is to establish the measurements of the various constructs modeled in Figure 1 and test the hypotheses about their relationships. Previous brand valuation research has employed survey-based, expert judgment and financial methods to measure brand equity.13 We adopt a self-report survey approach because our hypotheses are formulated at the consumer level. That is, we use the subjective preference that consumers attach to the tested brands as a surrogate measure of brand equity. The expert judgment approach is not appropriate because it involves arbitrary measures and nonstandard brand dimensions (Aaker and Jacobson 2001). We do not use financial measures because many online firms are not publicly traded; even for those that are listed, their share prices are probably too volatile for affirming their actual market values.14 Subjects We solicited subjects from a pool of undergraduate and postgraduate students who enrolled in a telecommunications course in a large university. All participants were volunteers, but they were given monetary incentives to complete the survey. We received altogether 402 responses and, after dropping 10 incomplete responses, we have a total of 392 valid data points. 13 For an in-depth discussion on the strengths and weaknesses of the various valuation methods, see Simon and Sullivan (1993) and Aaker and Jacobson (2001). 14 The rapid boom and crash of many dot.com’s share prices are sufficient to show the inaptness of the financial approach to measure brand values of online firms. 12 Stimuli To ensure that the subjects are aware of the studied brands, we restricted our stimuli to brands that provide two types of Web-based services, search engine and browser-based email, which are popular among most Internet users. To identify popular brands within the local context, we invited another 30 undergraduate students from the same university to suggest three popular brands in each service category. Frequency counts suggest that AltaVista, Google and Yahoo are the most popular search engines among the subjects, whereas Hotmail, Lycos Mail and Yahoo Mail are the most popular browser-based email service providers.15 In our survey, we measured the seven latent constructs for each of the above brands. We deliberately chose multiple brands and industries to induce higher variations in the subjects’ responses. These variations are useful for identifying the structural model and testing the research hypotheses. Measures We used multiple items to measure each of the studied constructs. The survey instruments of advertising intensity, perceived quality, brand knowledge and brand equity were adapted and modified based on items that were used in previous research (Aaker 1996; Brucks 1985; CobbWalgren 1995; Park et al. 1994; Yoo et al. 2000).16 By following similar styles and focusing on the conceptual meanings of the constructs, we constructed new items for value-added features, Website quality and product/technology innovation. Altogether, our initial survey contains 50 items that measure the seven constructs. All items were framed using 7-point Likert scale. The items were then customized for each of the selected brands, which resulted in six versions of survey. Except brand name and service category, all items were identically phrased across the versions. We further conducted two rounds of pilot studies to assess and purify the items. In each round, 24 upper-division undergraduate students (who did not participate in the main survey or the brand selection task) from the same university were invited to complete the pilot survey. Feedbacks were then collected regarding the layout, structure and clarity of the 15 AltaVista, Google and Yahoo were named 70 times (out of 90) in the search engine category, while Hotmail, Lycos Mail and Yahoo Mail were named 72 times (out of 90) in the browser-based email category. These figures indicate that the identified brands are popular and representative. 16 We used subjective measures instead of objective measures for brand knowledge. Prior research has found that subjective and objective knowledge are highly correlated, and subjective knowledge also affects information search and consumer behavior (Brucks 1985; Park et al. 1994). Subjective knowledge includes consumers’ self-confidence in their knowledge which, according to the human capital model (Ratchford 2001), is useful for predicting brand preferences. 13 items. We also performed preliminary reliability analyses to detect problematic items. After the two rounds of pilot studies, we discarded, rephrased and added some items to arrive at the final survey, which contains a total of 48 items, with at least six items per construct.17 Procedure We organized the survey into four sections. The first section contains items that measure the studied marketing factors, namely value-added features, product/technology innovation, Website quality and advertising intensity. The second section consists of items that measure perceived quality and brand knowledge. The third section contains items for measuring brand equity. In the last section, we ask for demographic data about the subjects. Except for demographics, the question orders are randomized within each respective section. We organized the questions into distinct, ordered tiers to minimize possible interference between the marketing mix elements and the dependent constructs. The six versions of the survey were then randomly distributed to the subjects. 64 subjects indicated that they had not tried the assigned brand before filling in the survey. Because brand name may play an unusually large influence in these cases (especially on perceived quality; see Dodds et al. 1991 and Rao and Monroe 1989), we discard these 64 data points and focus on the remaining 328 responses. Except Lycos Mail, the 328 responses are well distributed across the studied brands – 59 for AltaVista, 66 for Yahoo, 68 for Google, 56 for Hotmail, 20 for Lycos Mail and 59 for Yahoo Mail. Among the 328 subjects, 45.7 percent of them are female, and the majority of them are between 18 to 25 years old. Most of the subjects have extensive Internet experience – 65.2 percent of them spend more than 12 hours per week to surf the Internet.18 Data Analysis We performed confirmatory factor analysis (CFA) to assess the measurements of the variables and test the hypotheses.19 The two-step recommendation by Anderson and Gerbing (1988) was 17 For brevity, we only report 27 retained items in Table 1, after the measurement model was re-specified (see below). The full set of 48 items is available from the authors upon request. 18 While students may not represent the majority of e-business consumers, the use of student subjects is considered acceptable here, because they are frequent users of both search engine and browser-based email services. In Study Two, we extend the analysis to a sample of working adults and use brands from another online industry as stimuli to further assess the proposed hypotheses. 19 LISREL 8.51 was used in all analyses. 14 closely followed. First, we examined whether the measurement model tallies with the subjects’ responses and, if necessary, re-specified the model to ensure that the constructs are adequately measured. We then tested the hypotheses by estimating a structural model using the validated items. Note that when examining the CFA model fits and performing model comparisons, we used a more conservative significance level of 0.01 for all χ2 statistics. This is because the χ2 statistic is sensitive to sample size, and it would probably reject a model when the sample size is larger than 200, irrespective of true model fit (Hair et al. 1998). Measurement model We begin by estimating a 7-construct measurement model that includes all possible correlations between the constructs (this is sometimes called a “saturated” model; see Anderson and Gerbing 1988). Based on this model, we assess the conceptual distinctiveness of the seven constructs by performing 21 (=7×6/2) pair-wise comparisons using χ2 difference tests. All the χ2 differences, obtained by subtracting the overall χ2 of the 7-construct model (the hypothetical measurement model) from the respective 6-construct models (obtained by merging two constructs at a time), are significant with p-values substantially less than 0.01. This implies that the 48 items measure seven distinct constructs. The initial measurement model revealed a modest fit to the data, with a goodness-of-fit index (GFI) of 0.70. To detect problematic items, we examined the standardized residual matrix of the CFA model (Anderson and Gerbing 1988; Hair et al. 1998). Hair et al. (1998) suggest that a measurement model should have less than five percent of standardized residuals exceeding 2.58. Based on this criterion, we performed successive iterations of estimations and dropped 21 items that either loaded considerably onto other constructs, or did not correlate well with other items of the own-constructs. The final measurement model consists of 27 items that measure the seven latent constructs. Each construct is measured by a minimum of three items. Table 1 presents the retained items and their standardized loadings on the respective constructs. <Insert Table 1 here> The re-specified measurement model has a χ2 of 433.28 (p < 0.01), GFI of 0.91, root mean squared residual (RMSR) of 0.05, adjusted goodness-of-fit index (AGFI) of 0.89, nonnormed fit index (NNFI) of 0.97, and parsimonious goodness-of-fit index (PGFI) of 0.73. These values are considered acceptable (Hair et al. 1998; Sharma 1996). 15 To assess convergent validity, we computed the composite reliability and Cronbach alpha of the retained items, and the average variance extracted (AVE) per construct. The results of these computations are reported in Table 1 (under the corresponding constructs). Except valueadded features, all values exceed commonly recommended thresholds (Fornell and Larker 1981; Nunnally 1978). The variance extracted for value-added features is marginal, but it is close to the recommended threshold (0.49 vs. 0.50). All items have standardized loadings exceeding 0.50, and the loadings are statistically significant. We examined discriminant validity of the final (re-specified) model by inspecting interitem and construct correlations (Fornell and Larcker 1981). The variance extracted for each construct is larger than its shared variance with the other latent constructs. This indicates high correlations between items that measure the same constructs, but not across different constructs. Table 2 reports the shared variance among the constructs. <Insert Table 2 here> Structural model The conceptual model can be represented by the following structural equations, η1 0 η = β 2 12 η 3 β13 0 0 β 23 ξ 0 η1 γ 11 γ 21 γ 31 γ 41 1 ζ 1 ξ 0 η 2 + 0 γ 22 0 γ 42 2 + ζ 2 , ξ 0 η 3 0 0 0 0 3 ζ 3 ξ 4 (1) where η’s represent the endogenous constructs (perceived quality, brand knowledge and brand equity), ξ’s represent the exogenous constructs (the four Web-based marketing constructs) and ζ’s denote random errors. β’s and γ’s are path coefficients that are empirically estimated from the structural model. The relationships between the variables and parameters are graphically depicted in Figure 1. The measurement corresponding to the exogenous and endogenous constructs can be represented by the following equations (in matrix notations). x = Λ xξ + δ y = Λ yη + ε , (2) where x and y denote items that measure the exogenous and endogenous constructs respectively, Λx and Λy are parameter matrices containing the standardized loadings, and δ and ε are random 16 measurement errors. The estimated values in Λx and Λy and their statistical significance in the final model are similar to those reported in Table 1. To determine the best-fit structural model, we followed Anderson and Gerbing’s (1988, Figure 1) decision-tree framework to compare alternative specifications. Figure 2 illustrates the sequential χ2 difference tests (SCDTs) that we conducted. Note that we present only the tested branches in Figure 2. <Insert Figure 2 here> Essentially, the hypothesized model was compared to a series of competing models. We omitted the path between perceived quality and brand knowledge, which is the only relationship between the two components of brand equity, to derive the next most likely constrained model, Mc. We constructed the next most likely unconstrained models, Mu1 to Mu4, by adding a direct path from each marketing variable to brand equity. This could test for possible direct effects of the marketing variables on brand equity. By performing all the tests that are shown in Figure 2, the proposed structural model, Mt, which is equivalent to equation (1), was eventually accepted. Table 3 reports a summary of all the tests in the SCDTs. The final structure model has a χ2 of 439.29 (p < 0.01), GFI of 0.91, RMSR of 0.06, AGFI of 0.89, NNFI of 0.97, and PGFI of 0.74. These values are similar to those reported for the measurement model. <Insert Table 3 here> The explanatory power of the structural model is evaluated by inspecting the equationby-equation R2 of the endogenous constructs. Overall, the model can account for 88.24 and 17.79 percent of variations of perceived quality and brand knowledge respectively. 57.77 percent of variations in brand equity could be explained by the proposed factors. This greatly exceeds 10 percent, which indicates substantial explanatory power of our research model (Falk and Miller 1992). Three of the proposed Web-based marketing constructs are relevant for online firms, and they contribute to the studied firms’ brand equities. Table 4 reports all hypothesis testing results. The final structural model supports four of the six hypotheses. Specifically, product/technology innovation affects brand knowledge (H2b: γ22 = 0.14, p < 0.05), Website quality affects perceived quality (H3: γ31 = 0.88, p < 0.01), and advertising intensity influences both perceived quality (H4a: γ41 = -0.13, p < 0.01) and brand knowledge (H4b: γ42 = 0.15, p < 0.01). We found no relationship between value-added features (H1), product/technology innovation (H2a) and perceived quality. The model is consistent with 17 our conceptual framework of brand equity – both brand knowledge (β23 = 0.26, p < 0.01) and perceived quality (β13 = 0.63, p < 0.01) are positive components of brand equity, and they are positive correlated with each other (β12 = 0.29, p < 0.01). <Insert Table 4 here> IV. STUDY TWO Study One validated the measures of the latent constructs and tested their relationships. To assess the robustness of the above results, we invited another group of working postgraduate students to fill in the survey, and at the same time investigated the sensitivity of the results with respect to the conceptualization and measurement of two constructs, brand knowledge and Website quality, that are both extensively studied in the extant literature. Specifically, our measure of brand knowledge (see Table 1) is ex ante consistent with the basic principal of the human capital model, which concerns the level of knowledge that a person possesses about the studied brand. Another popular view of brand knowledge that was advanced by Keller (1993) focuses on two sub-dimensions, brand awareness and association. We included additional items that measure these two sub-dimensions in the current survey.20 Similarly, our measure of Website quality focuses on the subjects’ impression on the overall excellence of the selected Websites. McKinney et al. (2002) separate Website quality into two sub-dimensions, information quality and system quality, and suggest that both of them could influence consumer behavior. We followed their characterization of information and system quality, and added new items to measure each of these two quality dimensions.21 The hypotheses about all the Webbased factors were then re-assessed using the new items. Table 5 lists the new items that we added in this round. <Insert Table 5 here> 20 The measures of brand awareness and associations were adapted from Aaker (1996), Droge (1989), Edell and Keller (1989), Keller (1991), Maheswaran (1994), Putrevu and Lord (1994), Sujan and Bettman (1989) and Yoo et al. (2000). 21 The items for information and system quality were adapted from Dabholkar et al. (1996) and McKinney et al. (2002). 18 We followed the same procedures as in Study One, and collected 92 usable responses.22 The subjects were part-time postgraduate students in a large university who have enrolled in two evening courses on management. The average age of the subjects was 27, and 32.6 percent of them are female. Most have extensive online experience; 76.1 percent of them spend more than 12 hours per week to surf the Internet. We selected two online recruitment services, JobsDB (http://www.jobsdb.com) and JobStreet (http://www.jobstreet.com), as stimuli to assess the generalizability of our findings. This is because recruitment services are relevant to working students – all subjects have previous experience in using at least one of the two services. Because the sample size of Study Two is relatively small, it is not feasible to estimate a CFA model that includes all the path coefficients and measurement errors. Instead, we averaged the measurement items for each individual construct, and then performed ordinary least squares regressions to examine the relationships between the constructs. Before testing the hypotheses, we first verified the measures of brand knowledge and Website quality that were used in Study One. A regression of brand knowledge on brand awareness and brand association produces: Brand knowledge = -0.10 + 0.51 × Brand awareness + 0.48 × Brand association. (0.12) (0.14) Both independent variables have positive and significant coefficients (p < 0.01), and R2 is 0.52, which means that they could explain 52 percent of the variations in brand knowledge. Similarly, both information and system quality contribute to Website quality: Website quality = 0.22 + 0.47 × Information quality + 0.44 × System quality. (0.12) (0.13) The two quality dimensions are positive and significant (p < 0.01) with R2 = 0.59. Hence our measures are consistent with previous wisdoms on brand knowledge and Website quality (Keller 1993; McKinney et al. 2002), which further strengthens the confidence in Study One’s findings. We estimated the model in Figure 1 with the new sample (with old and new measures of brand knowledge and Website quality) using least squares regressions. The results are reported in Table 6. Among all the significant relationships, except the effect of advertising intensity on perceived quality (which had a negative coefficient in Study One), the rest of the coefficients 22 Before the survey was administered, we conducted one round of pilot study with 31 working subjects to check the clarity and content validity of the items. Several items were reworded after the pilot study. 19 have the same signs as those in Study One. Hypotheses H2b, H3 and H4b are supported, but not H1 and H2a. The result of H4a is opposite to that in Study One. These indicate that most of the findings can be generalized across different subjects and online service industries. They are also insensitive to the measures that we used for brand knowledge and Website quality. V. OVERALL DISCUSSION AND CONCLUSIONS Consistent with prior brand equity theory (e.g., Aaker 1996; Erdem and Swait 1998; Keller 1993), both perceived quality and brand knowledge are positive components of online firms’ brand equity. Collectively, they explain 58 and 43 percent of the variations in brand equity in Study One and Two respectively. These signify their importance for online firms who are keen on building brand reputations. Many possible factors, including individual-, firm- and marketlevel characteristics, could influence the formation of quality and knowledge perceptions. In the following paragraphs, we discuss the findings pertaining to four firm-level Web-based marketing variables that an online firm could manipulate to influence consumer perceptions. Future works should extend this domain of knowledge by exploring the significance of individual and market characteristics in an online setting. The Effects of Web-based Marketing Variables Contrary to our prediction, the quality of value-added features does not have significant effect on perceived quality (H1). That is, providing good peripheral features to consumers does not seem to raise their perception of product quality. One candidate explanation is that the subjects could have failed to separate value-added features from Website quality (there is high shared variance between these two constructs; see Table 2), and the results could emerge because of collinearity. We have considered this possibility, and we purposely explained to the subjects in Study Two about the differences between value-added features, Website quality and perceived quality before the survey. Further, we added an alternative measure of Website quality (in terms of information and system quality) in Study Two, but the insignificant result of value-added features persists. Hence we believe the collinearity explanation may not be well grounded. Several other explanations might account for this finding. First, by nature, value-added features are useful to consumers who are less familiar with the online services. Given the ample Internet experience of the subjects, they might have neglected the add-on features provided by 20 the studied brands. This is particularly plausible if those value-added features are designed to provide on-line help or support for users to simply enjoy the basic functions, which is likely the case for search engine, browser-based email or online recruitment services. Empirical research in technology adoption/acceptance has shown that perceived ease of use may not affect adoption attitude and/or behavior for experienced users (see, e.g., Szajna 1996; Taylor and Todd 1995). Therefore, future works should examine the benefits of peripheral features to inexperienced Web users and/or more complicated online products. For instance, due to the variety and complexity of bank transactions, online banking could be a good testing ground for value-added features. Second, it is also possible for consumers to have pre-defined expectations on the products offered by the Websites, and whether such expectations are fulfilled might have dominated their perceptions of product quality. Empirical evidence has shown that consumer satisfaction toward self-service technologies might be driven by service functionality (Meuter et al. 2000). That is, people may care more about the presence and proper functionality of basic services rather than add-on peripherals. This is more likely for online services that serve well-defined goals or task objectives with little room for service extension. Note that in physical environments, a point-ofsales agent would typically approach and offer assistance to consumers, through which she may create unexpected services that the consumers appreciate (Bitner et al. 1990). It is difficult to deliver such surprises in a pre-programmed online environment. Hence it is possible for valueadded features to affect the quality perception of consumers only for more ambiguous products and, even with such products, more salient add-on features might be required to induce changes in consumers’ quality perceptions. In any case, online firms may wish to review the relevance of value-added features that they provide to their target customers. For browser-based email, search engine and recruitment services, it appears that the provision of peripheral services or tools is a giveaway to experienced Internet users, and these features might not have noticeable impact on strengthening brand equity. Product/technology innovation has a positive influence on brand knowledge (H2b), but it does not affect perceived quality (H2a). Apparently, being innovative helps an online firm impart more information and product knowledge to consumers, but it may not shift the firm’s products into better positions along the quality spectrum (cf. Schmalensee 1982). This could be again due 21 to the fact that our subjects are familiar with the studied Web-based services. Any quality signal brought forward by product/technology innovations (e.g., new indexing scheme, secure email facilities, etc.) might have been internalized within the industries, which make the innovation “transparent” to the subjects. Nevertheless, it is possible for more recent innovations, practices or technologies to exert a high influence on consumers’ quality perceptions. Future work should investigate whether the age of an innovation matters in an online environment. The finding that product/technology innovation increases brand knowledge is interesting and useful for consumer researchers and Website managers. It implies that being an innovator may help an online firm suppress consumer learning about the attribute and actual quality of its products. As demonstrated by Van Osselaer and Alba (2000), such a blocking effect is pervasive in experimental tasks involving consumer judgment, and it may grant the firm additional brand advantage because consumers may ascertain the excellence of its products based on their initial assessment and disregard later information that is valuable. Given that perception of innovation helps raise consumer knowledge of a brand, even if their products are inferior, Website managers may still want to emphasize their pioneering activities and innovations, and devise appropriate promotional campaigns to block subsequent learning by consumers. As hypothesized, Website quality has a significant positive influence on perceived quality (H3); this positive effect is stable across different Website quality measures. A well designed Website may help an online firm enhance its product quality to consumers. Therefore, other than focusing on product excellence, an online firm should also devote resources into constructing an appealing Website, which directly serves as a storefront for consumers. McKinney et al. (2002) provide a framework to guide the design of a good Website, which consists of dimensions such as relevancy, reliability and interactivity. A casual observation on the Internet suggests that most firms are indeed aware of the importance of having good Websites; many companies profess in designing and maintaining Websites for corporate clients. It is however instructive to emphasize that our stimuli consist of brands which offer purely online services. Hence the subjects may need more time to interact with the Websites in order to “consume” the services, which might have amplified the importance of Website quality. It would be interesting for future work to explore whether this finding continues to hold for online firms which sell physical products (e.g., books, electronics, etc.). 22 Advertising intensity reinforces the brand knowledge of consumers (H4b), and it lowers (raises) perceived quality in Study One (Two) (H4a). Consistent with theories in information economics (Nelson 1974; Stigler 1961), advertising confers brand information and knowledge to consumers, and its effect persists in online settings. Our results on H4a are mixed – advertising intensity lowers perceived quality of search engine and Web-based email services. This implies that the wear-out effect proposed by psychological research may indeed be pervasive in certain online industries (Alba et al. 1991). Further, this finding is consistent with the “desperation” and “no pain” explanation advanced by Kirmani and Wright (1989), and it suggests that the extent of advertising might have been excessive for the studied service providers. Currently, many online firms use banner advertisements, pop-up browser windows and mass emails to promote their products. Market forecasts have suggested that advertising expense on the Internet is growing at astronomical rates (Shamdasani et al. 2001), and it is widely held that Internet advertisements incur lower costs (Dou et al. 2002). However, if an online firm over-promotes its brand or products, the excessive advertisements may bring inconvenience and frustration to consumers,23 which may ultimately lower the perceived quality of its products. In fact, taking into account the indirect paths, Study One suggests that advertising intensity has a net negative impact on brand equity. Therefore, online firms need to manipulate the nature and extent of advertising with care, and they should avoid over-selling their brands or products. By contrast, Study Two reveals that advertising intensity raises perceived quality of the recruitment services. This could be due to lower advertising expenses, and advertising may also give more assurance to job seekers about the reliability and quality of the service providers. These results collectively imply that advertising is an interesting variable for future studies – it is good in that it always raises consumers’ brand knowledge; but it is bad for some firms as it could reduce perceived quality. To conclude, three of the four Web-based marketing variables are found to be significant, and they jointly explain sizeable variations in perceived quality and brand knowledge. Through the later two constructs, they contribute to shaping the brand equities of the studied online firms. It is instructive to observe that the Internet has continually sparked new business practices and 23 See, for example, http://www.internetnews.com/IAR/article.php/1456271. 23 marketing models. Our studied marketing mix elements are by no means exhaustive, especially when much variation in brand knowledge is still not accounted for. It would be interesting to see if other Internet-specific variables or practices could help change consumer perceptions, which might ultimately be translated into brand equity. Further, another good extension to this study is to explore the significance and generalizability of our model across different product categories and online industries. We have tested the hypotheses on three online service industries; it would be interesting to see if these results hold for retailers that sell tangible goods on the Internet. 24 References Aaker, David A. Managing Brand Equity. New York: The Free Press, 1991. Aaker, David A. “Measuring Brand Equity across Products and Markets,” California Management Review, vol. 38, no. 3, Spring 1996, Berkeley, pp. 102-120. Aaker, David A. and Alexander L. Biel. Brand Equity and Advertising. Hillsdale, NJ: Lawrence Erlbaum Associates, 1993. Aaker, David A. and Robert Jacobson “The Value Relevance of Brand Attitude in HighTechnology Markets,” Journal of Marketing Research, vol. 38, no. 4, November 2001, pp. 485-493. Alba, Joseph W. and J. Wesley Hutchinson “Dimensions of Consumer Expertise,” Journal of Consumer Research, vol. 13, no. 4, March 1987, pp. 411-454. Alba, Joseph W., J. Wesley Hutchinson and John G. Lynch “Memory and Decision Making,” in Handbook of Consumer Behavior. Harold H. Kassarjian and Thomas S. Robertson, Eds. Englewood Cliffs, NJ: Prentice-Hall, Inc., 1991, pp. 1-49. Anderson, James C. and David W. Gerbing “Structural Equation Modeling in Practice: A Review and Recommended Two-Step Approach,” Psychological Bulletin, vol. 103, no. 3, 1988, pp. 411-423. Baker, Julie, A. Parasuraman, Dhruv Grewal and Glenn B. Voss “The Influence of Multiple Store Environment Cues on Perceived Merchandise Value and Patronage Intentions,” Journal of Marketing, vol. 66, no. 2, April 2002, pp. 120-141. Bitner, Mary Jo, Bernard H. Booms and Mary Stanfield Tetreault “The Service Encounter: Diagnosing Favorable and Unfavorable Incidents,” Journal of Marketing, vol. 54, no. 1, January 1990, pp. 71-84. Bowen, David E., Caren Siehl and Benjamin Schneider “A Framework for Analyzing Customer Service Orientations in Manufacturing,” Academy of Management Review, vol. 14, no. 1, 1989, pp. 75-95. Brucks, Merrie “The Effects of Product Class Knowledge on Information Search Behavior,” Journal of Consumer Research, vol. 12, no. 1, June 1985, pp. 1-16. Carpenter, Gregory S. and Kent Nakamoto “Consumer Preference Formation and Pioneering Advantage,” Journal of Marketing Research, vol. 26, no. 3, August 1989, pp. 285-298. Cobb-Walgren, Cathy J., Cynthia A. Ruble and Naveen Donthu “Brand Equity, Brand Preference, and Purchase Intent,” Journal of Advertising, vol. 24, no. 3, Fall 1995, pp. 25-40. Cronin, J. Joseph, Michael K. Brady and G. Tomas M. Hult “Assessing the Effects of Quality, Value, and Customer Satisfaction on Consumer Behavioral Intentions in Service Environments,” Journal of Retailing, vol. 76, no. 2, 2000, pp. 193-218. Dabholkar, Pratibha, Dayle I. Thorpe and Joseph O. Rentz “A Measure of Service Quality for Retail Stores: Scale Development and Validation,” Journal of the Academy of Marketing Science, vol. 24, no. 1, Winter 1996, pp. 3-16. 25 Dodds, William B., Kent B. Monroe and Dhruv Grewal “Effects of Price, Brand, and Store Information on Buyers’ Product Evaluations,” Journal of Marketing Research, vol. 28, no. 3, August 1991, pp. 307-319. Dou, Wenyu, Ulrik Nielsen and Chee Ming Tan “Using Corporate Websites for Export Marketing,” Journal of Advertising Research, vol. 42, no. 5, 2002, pp. 105-115. Dowling, Michael J. and Jeffrey E. McGee “Business and Technology Strategies and New Venture Performance: A Study of the Telecommunications Equipment Industry,” Management Science, vol. 40, no. 12, December 1994, 1663-1676. Droge, Cornelia “Shaping the Route to Attitudinal Change: Central Versus Peripheral Processing Through Comparative Versus Noncomparative Advertising,” Journal of Marketing Research, vol. 26, no. 2, May 1989, pp. 193-204. Edell, Julie and Kelvin Lane Keller “The Information Processing of Coordinated Media Campaigns,” Journal of Marketing Research, vol. 26, no. 2, May 1989, pp. 149-163. Erdem, Tulin and Joffre Swait “Brand Equity as a Signaling Phenomenon,” Journal of Consumer Psychology, vol. 7, no. 2, 1998, pp. 131-157. Falk, R. Frank and Nancy B. Miller. A Primer for Soft Modeling. Akron, OH: The University of Akron Press, 1992. Farquhar, Peter H. “Managing Brand Equity,” Marketing Research, vol. 1, no. 3, 1989, pp. 2433. Ford, Gary T., Darlene B. Smith and John L. Swasy “Consumer Skepticism of Advertising Claims: Testing Hypotheses from Economics of Information,” Journal of Consumer Research, vol. 16, no. 4, March 1990, pp. 433-441. Fornell, Claes and David F. Larcker “Evaluating Structural Equation Models with Unobservable Variables and Measurement Error,” Journal of Marketing Research, vol. 18, no. 1, February 1981, pp. 39-50. Goodman, Paul S., Mark Fichman, F. Javier Lerch and Pamela R. Snyder “Customer-Firm Relationships, Involvement, and Customer Satisfaction,” Academy of Management Journal, vol. 38, no. 5, October 1995, pp. 1310-1324. Hair, Joseph F., Rolph E. Anderson, Ronald L. Tatham and William C. Black. Multivariate Data Analysis with Readings, 5th ed. Upper Saddle River, NJ: Prentice-Hall, Inc, 1998. Haubl, Gerald and Valerie Trifts “Consumer Decision Making in Online Shopping Environments: The Effects of Interactive Decision Aids,” Marketing Science, vol. 19, no. 1, Winter 2000, pp. 4-21. Henard, David H. and David M. Szymanski “Why Some New Products are more Successful than Others,” Journal of Marketing Research, vol. 38, no. 3, August 2001, pp. 362-375. Holbrook, Morris B. and Kim P. Corfman “Quality and Value in the Consumption Experience: Phaedrus Rides Again,” in Perceived Quality: How Consumer View Stores and Merchandise, eds., Jacob Jacoby and Jerry C. Olson, Lexington, MA, 1985, pp. 31-57. Howard, John A. Consumer Behavior: Application of Theory. New York: McGraw-Hill, 1977. 26 Isobe, Takehiko, Shige Makino and David B. Montgomery “Resource Commitment, Entry Timing, and Market Performance of Foreign Direct Investments in Emerging Economies: The Case of Japanese International Joint Ventures in China,” Academy of Management Journal, vol. 43, no. 3, June 2000, pp. 468-484. Jacoby, Jacob and Jerry C. Olson. Perceived Quality: How Consumer View Stores and Merchandise. Lexington, MA, 1985, pp. xi-xiv. Jacoby, Jacob, George J. Syzabillo and Jacqueline Busato-Schach “Information Acquisition Behavior in Brand Choice Situations,” Journal of Consumer Research, vol. 3, no. 4, March 1977, pp. 209-216. Janiszewski, Chris and Stijn M.J. van Osselaer “A Connectionist Model of Brand-Quality Associations,” Journal of Marketing Research, vol. 37, no. 3, August 2000, pp. 331-350. Kamakura, Wagner A. and Gary J. Russell “Measuring Brand Value with Scanner Data,” International Journal of Research in Marketing, vol. 10, no. 1, 1993, pp. 9-22. Keller, Kevin Lane “Memory Factors in Advertising: The Effects of Advertising Retrieval Cues on Brand Evaluations,” Journal of Consumer Research, vol. 14, no. 3, December 1987, pp. 316-333. Keller, Kevin Lane “Cue Compatibility and Framing in Advertising,” Journal of Marketing Research, vol. 28, no. 1, February 1991, pp. 42-57. Keller, Kevin Lane “Conceptualizing, Measuring, and Managing Customer-Based Brand Equity,” Journal of Marketing, vol. 57, no. 1, January 1993, pp. 1-22. Keller, Kevin Lane. Strategic Brand Management: Building, Measuring and Managing Brand Equity. Upper Saddle River, NJ: Prentice-Hall, Inc, 1998. Kim, Jinwoo, Jungwon Lee, Kwanghee Han and Moonkyu Lee “Businesses as Buildings: Metrics for the Architectural Quality of Internet Businesses,” Information Systems Research, vol. 13, no. 3, September 2002, pp. 239-254. Kirmani, Amna “Advertising Repetition as a Signal of Quality: If it’s Advertised so Much, Something Must be Wrong,” Journal of Advertising, vol. 26, no. 3, Fall 1997, pp. 77-86. Kirmani, Amna and Peter Wright “Money Talks: Perceived Advertising Expense and Expected Product Quality,” Journal of Consumer Research, vol. 16, no. 3, December 1989, pp. 344353. Kotler, Philip H. Marketing Management: Analysis, Planning, and Control, 8th ed. Englewood Cliffs, NJ: Prentice-Hall, Inc, 1994. Lieberman, Marvin B. and David B. Montgomery “First-Mover Advantages,” Strategic Management Journal, vol. 9, Special Issue: Strategy Content Research, Summer 1988, pp. 41-58. Lohse, Gerald L. and Peter Spiller “Electronic Shopping,” Communications of the ACM, vol. 41, no. 7, July 1998, pp. 81-87. 27 Maheswaran, Durairaj “Country of Origin as a Stereotype: Effects of Consumer Expertise and Attribute Strength on Product Evaluations,” Journal of Consumer Research, vol. 21, no. 2, September 1994, pp. 354-365. Mandel, Naomi and Eric J. Johnson “When Web Pages Influence Choice: Effects of Visual Primes on Experts and Novices,” Journal of Consumer Research, vol. 29, no. 2, September 2002, pp. 235-245. McKinney, Vicki, Kanghyun Yoon, Fatemeh “Mariam” Zahedi “The Measurement of WebCustomer Satisfaction: An Expectation and Disconfirmation Approach,” Information Systems Research, vol. 13, no. 3, September 2002, pp. 296-315. Meuter, Matthew L., Amy L. Ostrom, Robert I. Roundtree and Mary Jo Bitner “Self-Service Technologies: Understanding Customer Satisfaction with Technology-Based Service Encounters,” Journal of Marketing, vol. 64, no. 3, July 2000, pp. 50-64. Nault, Barrie R. and Albert S. Dexter “Added Value and Pricing with Information Technology,” MIS Quarterly, vol. 19, no. 4, December 1995, pp. 449-464. Nelson, Phillip “Advertising as Information,” Journal of Political Economy, vol. 82, no. 4, 1974, pp. 729-754. Nelson, Phillip “Information and Consumer Behavior,” Journal of Political Economy, vol. 78, no. 2, 1970, pp. 311-329. Nunnally, Jum C. Psychometric Theory. McGraw-Hill, NY, 2nd ed., 1978. Olshavsky, Richard W. “Perceived Quality in Consumer Decision Making: An Integrated Theoretical Perspective,” in Perceived Quality: How Consumer View Stores and Merchandise, eds., Jacob Jacoby and Jerry C. Olson, Lexington, MA, 1985, pp. 3-29. Ostrom, Amy and Dawn Iacobucci “Consumer Trade-Offs and the Evaluation of Services,” Journal of Marketing, vol. 59, no. 1, January 1995, pp. 17-28. Palmer, Jonathan W. “Web Site Usability, Design, and Performance Metrics,” Information Systems Research, vol. 13, no. 2, June 2002, pp. 151-167. Park, Chan Su and V. Srinivasan “A Survey-Based Method for Measuring and Understanding Brand Equity and its Extendibility,” Journal of Marketing Research, vol. 31, no. 2, May 1994, pp. 271-288. Park, C. Whan, David L. Mothersbaugh and Lawrence Feick “Consumer Knowledge Assessment,” Journal of Consumer Research, vol. 21, no. 1, June 1994, pp. 71-82. Putrevu, Sanjay and Kenneth R. Lord “Comparative and Noncomparative Advertising: Attitudinal Effects under Cognitive and Affective Involvement Conditions,” Journal of Advertising, vol. 23, no. 2, June 1994, pp. 77-90. Raaijmakers, Jeroen G. W. and Richard M. Shiffrin “Search of Associative Memory,” Psychological Review, vol. 88, no. 2, March 1981, pp. 93-134. Rao, Akshay R. and Kent B. Monroe “The Effect of Price, Brand Name, and Store Name on Buyers’ Perceptions of Product Quality: An Integrative Review,” Journal of Marketing Research, vol. 26, no. 3, August 1989, pp. 351-357. 28 Ratchford, Brian T. “The Economics of Consumer Knowledge,” Journal of Consumer Research, vol. 27, no. 4, March 2001, pp. 397-411. Ratcliff, Roger and Gail McKoon “A Retrieval Theory of Priming in Memory,” Psychological Review, vol. 95, no. 3, 1988, pp. 385-408. Robinson, William T. “Sources of Market Pioneer Advantages: The Case of Industrial Goods Industries,” Journal of Marketing Research, vol. 25, no. 1, February 1988, pp. 87-94. Robinson, William T. “Product Innovation and Start-up Business Market Share Performance,” Management Science, vol. 36, no. 10, October 1990, pp. 1279-1289. Robinson, William T. and Claes Fornell “Sources of Market Pioneer Advantages in Consumer Goods Industries,” Journal of Marketing Research, vol. 22, no. 3, August 1985, pp. 305-317. Schmalensee, Richard “Product Differentiation Advantages of Pioneering Brands,” American Economic Review, vol. 72, no. 3, June 1982, pp. 349-365. Sharma, Subhash. Applied Multivariate Techniques. USA: John Riley & Sons, Inc., 1996. Shamdasani, Prem N., Andrea J S Stanaland and Juliana Tan “Location, Location, Location: Insights for Advertising Placement on the Web,” Journal of Advertising Research, vol. 41, no. 4, 2001, pp. 7-21. Simon, Carol J. and Mary W. Sullivan “The Measurement and Determinants of Brand Equity: A Financial Approach,” Marketing Science, vol. 12, no. 1, Winter 1993, pp. 28-52. Srinivasan, Srini S., Rolph Anderson and Kishore Ponnavolu “Customer Loyalty in eCommerce: An Exploration of its Antecedents and Consequences,” Journal of Retailing, vol. 78, no. 1, 2002, pp. 41-50. Stigler, George J. “The Economics of Information,” Journal of Political Economy, vol. 69, no. 3, 1961, pp. 213-225. Sujan, Mita and James R. Bettman “The Effects of Brand Positioning Strategies on Consumers’ Brand and Category Perceptions: Some Insights from Schema Research,” Journal of Marketing Research, vol. 26, no. 4, November 1989, pp. 454-467. Swait, Joffre, Tulin Erdem, Jordan Louviere and Chris Dubelaar “The Equalization Price: A Measure of Consumer-Perceived Brand Equity,” International Journal of Research in Marketing, vol. 10, no. 1, 1993, pp. 23-45. Szajna, Bernadette “Empirical Evaluation of the Revised Technology Acceptance Model,” Management Science, vol. 42, no. 1, 1996, pp. 85-92. Szymanski, David M. and Richard T. Hise “e-Satisfaction: An Initial Examination,” Journal of Retailing, vol. 76, no. 3, 2000, pp. 309-322. Taylor, Shirley and Peter Todd “Assessing IT usage: The Role of Prior Experience,” MIS Quarterly, vol. 19, no. 4, 1995, pp. 561-570. Vakratsas, Demetrios and Tim Ambler “How Advertising Works: What do We Really Know?” Journal of Marketing, vol. 63, no. 1, January 1999, pp. 26-43. 29 Van Osselaer, Stijn M.J. and Joseph W. Alba “Consumer Learning and Brand Equity,” Journal of Consumer Research, vol. 27, no. 1, June 2000, pp. 1-16. Van Waterschoot, Walter and Christophe Van den Bulte “The 4P Classification of the Marketing Mix Revisited,” Journal of Marketing, vol. 56, no. 4, October 1992, pp. 83-93. Willcocks, Leslie P. and Robert Plant “Pathways to E-Business Leadership: Getting from Bricks to Clicks,” Sloan Management Review, vol. 42, no. 3, Spring 2001, pp. 50-59. Yoo, Boonghee, Naveen Donthu and Sungho Lee “An Examination of Selected Marketing Mix Elements and Brand Equity,” Journal of the Academy of Marketing Science, vol. 28, no. 2, Spring 2000, pp. 195-211. Zeithaml, Valarie A “Consumer Perceptions of Price, Quality and Value: A Means-End Model and Synthesis of Evidence,” Journal of Marketing, vol. 52, no. 3, July 1988, pp. 2-22. 30 Value-Added Features (ξ1) γ11 (H1) Product/ Technology Innovation (ξ2) γ21 (H2a) β13 γ31 (H3) γ22 (H2b) Brand Equity (η3) β12 Website Quality (ξ3) γ41 (H4a) Advertising Intensity (ξ4) Perceived Quality (η1) Brand Knowledge (η2) β23 γ42 (H4b) Figure 1. Research Model 31 n.s. Mt – Ms Mc – Mt s. Mt – Mu1 or Mt – Mu2 n.s. or Mt – Mu3 or Mt – Mu4 Accept Mt s. – significant n.s. – non-significant Ms – Measurement (saturated) model; Mt – Hypothesized structural model; Mc – Next most likely constrained model with respect to Mt (Omit PQ → BK); Mu1 – Next most likely unconstrained model with respect to Mt (Add VA → BE); Mu2 – Next most likely unconstrained model with respect to Mt (Add AI → BE); Mu3 – Next most likely unconstrained model with respect to Mt (Add WQ → BE); Mu4 – Next most likely unconstrained model with respect to Mt (Add PI → BE). VA – Value-added features; AI – Advertising intensity; WQ – Website quality; PI – Product/technology innovation; PQ – Perceived quality; BK – Brand knowledge; BE – Brand equity Figure 2. Sequential Chi-square Difference Tests 32 Table 1. Items and Measurement Properties Item1 Standardized loading2 Value-added features (VA) (composite reliability = 0.74; Cronbach alpha = 0.73; AVE = 0.49)3 The quality of the value-added features offered by X is impressive. The value-added features provided by X are desirable. X provides low standard value-added features.4 0.78 (15.18) 0.70 (13.30) 0.60 (10.96) Product/technology innovation (PI) (composite reliability = 0.89; Cronbach alpha = 0.89; AVE = 0.67) When improving products or technologies, X has been fast. X is fast in advancing its products or technologies. It appears that X is in the forefront in terms of new product or technology development. X has been fast in enhancing its products or technologies. 0.79 (16.61) 0.84 (18.28) 0.74 (15.09) 0.89 (20.04) Website quality (WQ) (composite reliability = 0.88; Cronbach alpha = 0.88; AVE = 0.64) The Website quality of X is well maintained. I am impressed with the Website quality of X. X maintains a high standard of Website quality. The Website quality of X is superior. 0.77 (16.20) 0.82 (17.64) 0.81 (17.38) 0.80 (16.93) Advertising intensity (AI) (composite reliability = 0.89; Cronbach alpha = 0.89; AVE = 0.63) X is intensively advertised on any media. I find it easy to come across an advertisement for X on any media. Advertisements for X can be seen frequently on any media. Advertisements of X are common on any media. It is rare to see an advertisement for X on any media.4 0.72 (14.59) 0.76 (15.75) 0.90 (20.46) 0.87 (19.14) 0.70 (14.02) Perceived quality (PQ) (composite reliability = 0.80; Cronbach alpha = 0.80; AVE = 0.57) X is of high quality. X appears to be very reliable. The quality of X is poor.4 0.85 (18.29) 0.71 (14.19) 0.70 (14.03) Brand knowledge (BK) (composite reliability = 0.90; Cronbach alpha = 0.90; AVE = 0.75) I know much about X. I have strong knowledge of X. I have very good understanding on X. 0.81 (17.16) 0.92 (20.67) 0.86 (18.92) Brand equity (BE) (composite reliability = 0.89; Cronbach alpha = 0.88; AVE = 0.61) X is better than any other Y service with same quality. 0.73 (14.88) Compared to another Y service with identical quality, X is of a higher value to me. 0.90 (20.28) In the presence of a Y service that is just as good, it is more worthwhile to use X. 0.85 (18.76) If there is a Y service that is not different from X in any way, I will give X a higher rating. 0.74 (15.33) It makes sense to use X instead of any other Y service even if they are the same. 0.66 (13.05) 1 X = one of the studied brands in the corresponding category; Y = search engine or Web-based email. 2 t-statistics in parentheses. All t-statistics are significant with p < 0.01. 3 AVE = average variance extracted. 4 Data are reverse-coded. 33 Table 2. Shared Variance between the Latent Constructs Construct VA AI WQ PL PQ BK BE VA 0.49 AI 0.00 0.63 WQ 0.45 0.01 0.64 PI 0.24 0.05 0.31 0.67 PQ 0.38 0.00 0.61 0.28 0.57 BK 0.07 0.03 0.09 0.10 0.11 0.75 BE 0.20 0.00 0.34 0.17 0.38 0.18 0.61 Comparison Mt / M s Mc / M t Mt / Mu1 Mt / Mu2 Mt / Mu3 Mt / Mu4 + s. = significant (p ≤ 0.01); n.s. = non-significant (p > 0.01). Hypothesis H1 (γ11) H2a (γ21) H2b (γ22) H3 (γ31) H4a (γ41) H4b (γ42) + Table 3. Summary of the SCDTs χ2 χ2 df Outcome+ Conclusion ST ND difference 1 2 439.29 435.27 4.03 4 n.s. – 450.51 439.29 11.22 1 s. – 439.29 437.71 1.58 1 n.s. Accept Mt 439.29 437.15 2.14 1 n.s. Accept Mt 439.29 436.06 3.23 1 n.s. Accept Mt 439.29 438.98 0.32 1 n.s. Accept Mt Table 4. Hypotheses Testing Path Standard Path t-statistic coefficient error 0.03 0.10 0.30 VA → PQ 0.06 0.05 1.02 PI → PQ 0.14 0.08 1.75 PI → BK 0.88 0.11 8.37 WQ → PQ -0.13 0.04 -3.22 AI → PQ 0.15 0.06 2.39 AI → BK Conclusion+ Not supported Not supported Supported Supported Supported Supported Except H4a: AI → PQ, all hypotheses are evaluated using one-tailed tests. The t-statistic of γ22 is significant at p < 0.05. All other significant t-statistics have p < 0.01. 34 Table 5. New Measurement Items Item+ Information Quality (composite reliability = 0.81) The website of X contains relevant information. I can understand the information from the website of X. Information provided by the website of X is reliable. The website of X covers a broad scope of information. The information obtained from the website of X is adequate. Most of the information from the website of X is useful. System Quality (composite reliability = 0.75) It is easy to find what I need from the website of X. The hyperlinks found in the website of X are good. The accessibility of the website of X is good. The website of X provides me with an interactive experience. The website of X is user-friendly. I find it easy to navigate the website of X. It is entertaining to surf the website of X. Brand Awareness (composite reliability = 0.61) It is easy to recognize X. Some characteristics of X come to my mind easily. I am aware of X. I can recall how the logo of X looks like. Brand Association (composite reliability = 0.91) I have a favorable opinion of X. I like X very much. I have a pleasant experience with X. The decision to use X is wise. X has been a good choice. X has a personality. My choice of X is based on its unique features. I have a clear image of the type of person who would use X. There are features that can describe X. Some features of X are beneficial. X provides superior features. X has several beneficial features. The features offered by X are useful. + X = one of the studied brands. 35 Hypothesis / parameters H1 (γ11) H2a (γ21) H2b (γ22) H3 (γ31) H4a (γ41) H4b (γ42) β12 β13 β23 Table 6. Results of Study Two+ With Study One With new BK and Path measures WQ measures++ -0.01 (0.11) 0.02 (0.10) VA → PQ 0.13 (0.13) 0.06 (0.12) PI → PQ *** 0.30 *** (0.09) 0.35 (0.15) PI → BK 0.80 *** (0.15) 0.57 *** (0.13) WQ → PQ 0.11 * (0.06) 0.14 ** (0.06) AI → PQ 0.08 * (0.05) 0.22 *** (0.08) AI → BK 0.05 (0.13) 0.29 *** (0.08) PQ → BK 0.15 ** (0.09) 0.45 *** (0.09) PQ → BE 0.39 *** (0.08) 0.92 *** (0.11) BK → BE + Standard errors in parentheses. Except H4a: AI → PQ, all hypotheses are evaluated using one-tailed tests. *** p < 0.01; ** p < 0.05; * p < 0.10. ++ A composite measure of brand awareness and association was used in place of brand knowledge; website quality was measured by information and system quality. 36
© Copyright 2026 Paperzz