Available online at www.sciencedirect.

com

Journal of Interactive Marketing 25 (2011) 169 177 www.elsevier.com/locate/intmar

Relationships among Brand Identity, Brand Image and Brand Preference: Differences between Cyber and Extension Retail Brands over Time
Maria Sksjrvi a & Saeed Samiee b,
a

Department of Product Innovation Management, Delft University of Technology, Delft, The Netherlands b Collins College of Business, The University of Tulsa, OK, USA Available online 24 May 2011

Abstract This study examines the relationships among brand identity, brand image, and brand preference in the context of cyber (pure online) and offline-based extension (traditional brick-and-mortar to online) retail brands over time. We test a conceptual model with survey data gathered over three time periods. Our results show that offline-based extension brands have an advantage over cyber brands when it comes to translating a brand identity into a successful brand image, especially in the early Internet stages (i.e., introduction and elaboration stages). Offline-based extension brands gain positive spillover effects from their offline-based counterparts, but such effects take time, and are not evident in the early Internet stage. Both types of brands have to work hard in the introductory stage to create a successful brand image that manifests into consumer preference for the brand. With regards to Internet use, we found that cyber brands have a slight disadvantage when moving from the elaboration stage to the fortification stage. 2011 Direct Marketing Educational Foundation, Inc. Published by Elsevier Inc. All rights reserved.
Keywords: Cyber brand; Extension brand; Brand identity; Brand image; Brand preference; Internet stage; Over time

Introduction Brands are among the most important intangible assets possessed by firms, contributing to greater value and market success (De Asis 2001; Shankar, Azar, and Fuller 2008). The brand premium is based on the equity that brands gain from being familiar, credible, and trustworthy, which lowers information costs and perceived risk involved with a purchase (Shankar, Azar, and Fuller 2008). A brand's worth is intimately tied to consumer reactions to product(s) or service(s) carrying a particular brand name. Recognizing this issue, much scholarly interest in the branding literature has centered on identifying ways of obtaining favorable consumer perceptions toward brands. For example, Shankar, Azar, and Fuller (2008) demonstrated that consumers' perceptions of a brand contribute
The first author gratefully acknowledges funding from the Academy of Finland. Corresponding author at: Collins College of Business, The University of Tulsa, 800 South Tucker Drive, Tulsa, OK 74104-9700, USA. E-mail address: samiee@utulsa.edu (S. Samiee).

to the brand's relative strength in a market, which drives a brand's value in the marketplace. The Internet has provided a particularly fruitful arena for branding (Bart et al. 2005). Online firms such as Amazon.com invest a great deal of resources in building a brand that elicits favorable reactions from consumers, as such reactions drive future purchase behavior. Of particular importance is eliciting favorable evaluations toward Internet brands that differ in their initial dispositions: cyber brands that only exist online (i.e., pure online brands) and have to build a brand from inception, and offline-based extension brands with marketplace counterparts (i.e., traditional brick-and-mortar brands) that serve as valuable bases for branding activities. Prior studies have shown that cyber and offline-based extension brands are likely to evoke different responses from consumers (Bart et al. 2005; Degeratu, Rangaswamy, and Wu 2000). Several studies also suggest that offline-based counterparts have an advantage over pure online brands. For example, Sksjrvi and Samiee (2007) demonstrate that offline-based brands are likely to have an advantage over pure online brands since they can gain positive spillover effects from their offline-

1094-9968/$ - see front matter 2011 Direct Marketing Educational Foundation, Inc. Published by Elsevier Inc. All rights reserved. doi:10.1016/j.intmar.2011.04.002

170

M. Sksjrvi, S. Samiee / Journal of Interactive Marketing 25 (2011) 169177

based counterparts. However, the extent to which such advantages are sustainable over time has not been examined. Although offline-based extension brands may have advantages in the beginning of their Internet stage, such effects may disappear after pure online brands have established themselves as viable competitors. A study of brand perceptions over time affords us the opportunity to uncover whether such patterns exist with respect to cyber and offline-based extension brands. Accordingly, the objective of this study is to explore the relationships among brand identity, brand image, Internet use, and brand preference for cyber and extension retail brands over time. The focus of the investigation is on the shift in consumer perceptions along the four constructs of interest to identify the advantages of each brand type over time. This investigation makes a contribution to the literature by examining how changes in core branding components affect consumer preferences for cyber and extension retail brands over time. Conceptual Model Degeratu, Rangaswamy, and Wu (2000) relied on information integration theory for examining how consumers integrate information available to them online as well as offline when evaluating Internet brands. According to their findings, the type of information that is important online and offline varies. Consumers often lack sensory information about online brands and, thereby, to a much greater extent rely on other kinds of information. Since consumers tend to lack any brand-related information regarding cyber retail brands at inception, other types of information such as price or product attributes become more important than the brand. This situation is exacerbated when the brand is just establishing itself online. In contrast, consumers have experience with marketplace brands and, therefore, tend to rely more on the brand and their prior evaluations of the brand instead of searching for new information (e.g., pertaining to price or product attributes). In this case, the brand name becomes a surrogate for other types of information that may be difficult to acquire. This suggests that consumers are likely to evaluate online and offline brands differently, and that such evaluations may vary over time. Aaker and Keller (1996), Aaker, Keller, and Joachimstaler (2000), and Keller (1993, 2003) identify two main sources of consumer preference for a particular brand: brand identity and brand image. Brand identity is defined as a unique set of brand associations that firms aim to create or maintain, whereas brand image is defined as consumer perceptions regarding a brand (Keller 2003, p. 66). Brand preferences over time mainly shift due to changes in these two components (McEnally and de Chernatony 1999). In line with Shankar, Azar, and Fuller (2008), we conceptualize brand identity and brand image as containing several subcomponents, with brand identity consisting of brand awareness, purpose, differentiation, and offerings (de Chernatony 1999), and brand image pertaining to brand credibility, brand character, consumers' overall attitude towards the brand, and consumers' feelings for the brand (de Chernatony 1999). In our conceptualization, brand image serves as a mediator between brand identity and preference. Brand identity

represents how firms aspire to be perceived, whereas brand image refers to how they are perceived. Brand identity does not directly influence consumer preferences. Rather, consumers interpret the firm's identity and translate it into an image (Keller 2003), and the image in turn influences consumer preferences (Keller 1993; Martnez and de Chernatony 2004). We also include Internet use in our model as a component of preference to convey the fact that consumers' experiences with a particular shopping channel may influence their preferences towards brands in that channel (Bart et al. 2005; Eastlick and Lotz 1999). Studies addressing Internet use show that consumers who extensively use a particular channel (such as the Internet) are more likely to differentiate the brands sold through that channel than shoppers who rarely use the channel (e.g., Eastlick and Lotz 1999). The marketing literature suggests four main components that drive the use of a given shopping channel: usage history, time spent (use duration), number of purchases (familiarity with channel), and spending (Bart et al. 2005; Eastlick 1996; Shim et al. 2001). The proposed conceptual framework modeling consumer perceptions of cyber and offline-based extension brands over time is shown in Fig. 1. Hypotheses On the Internet, a brand can be said to evolve in three stages: introduction, elaboration, and fortification, analogous to Park, Jaworski, and MacInnis' (1986) model of brand lifecycles. In the introductory stage, firms need to develop a set of activities to establish a brand's positioning in the market. As the external attributes are more important than brand-related aspects in this phase, it is important to develop a strong identity that can leverage positive associations with the brand (Degeratu, Rangaswamy, and Wu 2000; Keller 2003). In the elaboration stage, when the brand is becoming more important, the strategy should shift toward enhancing the value of the brand. In this stage, the key sources for brand equity should be preserved and amplified (Shankar, Azar, and Fuller 2008) and the positive and unique brand associations created in the introductory stage should be fortified (Keller 1993). In the fortification stage, the brand is linked with the firm's other

Brand type H2 H1 Brand identity H1 Brand preference H2 H3 Internet use Brand image H3

Internet stage

Fig. 1. Measurement model.

M. Sksjrvi, S. Samiee / Journal of Interactive Marketing 25 (2011) 169177

171

brands to reinforce all products in the brand's portfolio. In each Internet stage, consumer preference for a particular brand is likely to be sensitive to the amount of information available about the brand, potentially triggering changes in consumer preferences over time (Degeratu, Rangaswamy, and Wu 2000). The first hypothesis of our model concerns changes in brand identity over the stages of the Internet. Degeratu, Rangaswamy, and Wu (2000) suggest that establishment of an identity is likely to be more important in the introductory stage, when consumers are less familiar with a brand and its offering (Park, Jaworski, and Maclnnis, 1986). The purpose of brand identity is to establish a relationship between the customer and a brand which, in turn, expresses the brand's aspirations (Aaker and Keller 1996). As such, firms need to create a unique positioning of the brand by communicating the brand's benefits and its attributes to consumers (McEnally and de Chernatony 1999). In an online environment, the effort required for cyber and offline-based extension brands to create a meaningful brand identity is likely to be unequal. Cyber brands have to work much harder than offline-based extension brands to attain a positive preference because they have to create an identity from scratch. In contrast, an extension brand's connection to a brick-andmortar parent allows customers to transfer existing familiarity and attitudes regarding the offline-based brand to the firm's online extension brand at inception. Offline-based extension brands have parent brands with existing brand equity available for creating an identity for their new Internet-based brand, which results in positive spillover effects from the parent to the extension brand (Aaker and Keller 1990). In a seminal study, Shankar, Azar, and Fuller (2008) demonstrate the importance of spillover effects of brands from other categories than the focal brand. Focusing on Allstate Insurance, they showed that spillover effects occurred in many product categories (from auto to property and life insurance, and from property to auto insurance), and that such effects were positively and significantly related to brand strength. As such, offline-based brands benefit from leveraging their reputation in their online channel, thereby alleviating information asymmetries across markets. As cyber brands do not have spillover effects, they have to invest more in the creation of an identity than offline-based extension brands. This effect is likely to decrease over time, once cyber brands have created an online identity that can successfully convey their image. More formally stated: H1. The decrease in the effect of brand identity on brand image between the stages of the Internet is more pronounced for cyber than for offline-based extension brands. We further propose that brand image is likely to demonstrate a stronger relationship with consumer preferences in the early Internet stages. Following Degeratu, Rangaswamy, and Wu (2000), within the context of a new brand and a new shopping channel, consumers have to establish attitudes based on the available information. Creating a memorable brand image requires more investment from a cyber brand than from an extension brand. Offline-based extension brands leverage their offline-based counterparts to create associations in consumers' minds (Aaker and Keller 1990); in forming an image for an

extension brand, consumers only need to update their existing image of the brand's offline-based counterpart. For example, in creating an image for Barnesandnoble.com, consumers can access their image of Barnes and Noble's physical stores and update it with the online association (cf. Boush and Loken 1991). In contrast, when a cyber brand is introduced, consumers have to create an entirely new image for it. No previous category association exists to which consumers can relate and, therefore, they have to create a new category for evaluating the brand (Boush and Loken 1991). In an online environment, information is relatively easy to obtain and, as more information about relevant attributes becomes available, the relative importance weights of the attributes already available should decrease (Degeratu, Rangaswamy, and Wu 2000). Hence, we propose: H2. The decrease in the effect of brand image on brand preference between the stages of the Internet is more pronounced for cyber than for offline-based extension brands. As consumers become accustomed to a particular retail channel and feel comfortable in using it, the impact of the shopping channel is likely to grow stronger over time (Eastlick 1996). For example, Bart et al. (2005) found a positive relationship between online shopping experience and consumer intentions to buy from a site. Retail channel use is likely to have a larger impact on offline-based extension brands than on cyber brands. Once consumers start using a cyber brand, they will automatically associate the brand with the Internet and the online environment. Brands such as Amazon.com do not exist in other contexts and marketplaces and their only shopping channel association is with the Internet. In contrast, consumers are already accustomed to using the offline-based counterpart of extension brands. In this case, Internet use is likely to have a significant impact on whether they switch from using the physical outlet to using the Internet channel. Consumers who are already comfortable using the Internet for shopping are also more likely to start using the Internet extension of offline-based brands. Thus, we posit: H3. The decrease in the effect of Internet use on brand preference between the stages of the Internet is more pronounced for cyber than for offline-based extension brands. Methodology The centrality of Internet brand preference in this study required data collection for multiple cyber and offline-based extension brands at three different points in time. The data for cyber and offline-based extension brands were gathered separately. The first set of data for this study was collected in 2000, when Internet brands were in an early stage of development. The second and third waves of data were collected in 2005 and 2009, when the use of the Internet had proliferated and online purchases had become routine for a significant portion of the population. All three sets of data were gathered from respondents with access to a computer and the Internet in a Midwestern city which is frequently targeted as a test market.

172

M. Sksjrvi, S. Samiee / Journal of Interactive Marketing 25 (2011) 169177

Data pertaining to cyber and offline-based extension brands ideally should be gathered from individuals who have had the opportunity to be exposed to these brands. Thus, as recommended by Churchill (1999), this study relied on a nonprobability judgment or purposive sampling method in which respondents were screened for computer literacy, Internet access, and Internet shopping experience. All respondents were working adult businesspeople from local firms who were asked to complete two sequential questionnaires administered about three weeks apart. Thus, each respondent completed one questionnaire concerning cyber brands and approximately three weeks later a second one regarding offline-based extension brands. With the exception of brand names pertaining to cyber and offline-based extension brands, the two research forms were identical. We measured respondents' familiarity with the parents of offline-based extension brands on a seven-point scale. The mean for brand familiarity was 5.9, indicating that the use of the extended brands in this study was appropriate. For the first data collection effort in 2000, 114 respondents completed both forms. In 2005, two completed forms were received from 99 consumers. The third data collection in 2009 consisted of 150 respondents. Considerable turnover in employment occurred between the first and second data collection. As a result, and consistent with other longitudinal studies that rely on primary data, our consumer samples differed somewhat over the three data collection periods. Questionnaire Development All fourteen Internet retail brands used in the study both cyber brands and offline-based extension brands were among the top 100 retail eCommerce sites according to the rankings by the National Retail Federation for 1999 (Reda 1999). Even though all Internet brands were in their relative infancy in 1999, the top 100 brands were among the most recognized and widely used cyber and offline-based extension brands. All of the brands used in the study were launched between the years 1995 to 1999. Thus, in 1999, all of the brands targeted in this investigation were in their introductory stage. To minimize product category bias or retail brand salience and interest among respondents, for each cyber brand included in the study, a closely matched counterpart extension brand was also targeted for investigation. The seven retail cyber brands selected were matched as closely as possible with a retail extension brand: CDNow.com with TowerRecords.com, eToys.com with Toys-RUs.com, Amazon.com with BarnesandNoble.com, Egghead.com with CompUSA.com, Shopping.com with WalMart.com, eTrade. com with CharlesSchwab.com, and Expedia.com with Citibank. com. Some cyber brands included in the first wave of data collection were acquired or forced out of business and liquidated. These brands were dropped from our model. Cyber and offline-based extension brands were matched along several characteristics. First, all brands selected were in the same broad industry, i.e., retail business. Second, the retail focus for each pair of brands (that for which they are known) was also closely matched. Third, we broadly matched each brand's focused offering (e.g., the breadth of book titles offered for both cyber and

extension booksellers was extensive). Fourth, cyber and their matching offline-based extension brands were introduced at about the same time. Fifth, cyber brands and their extension counterparts selected for the study were at the same stage of development on the Internet (that is, had existed for approximately the same length of time during which they were actively engaged in brand building programs). Two points regarding the matching of cyber and offline-based extension brands are noteworthy. First, Amazon.com had a slight advantage in terms of its inception over BarnesandNoble.com, and second, Expedia. com and Citibank.com were matched as purveyors of services. The cyber and extension brand questionnaires were pretested for content validity and relevance on a pretest sample of male and female subjects 18 to 50 years old. The pretest subjects did not report any problems in completing the research instruments and, hence, no major change to the forms was deemed necessary. Measures For each retail brand type (cyber and extension), summated scales across all brands were used to develop the indicators for the four main constructs of this study: brand identity, brand image, brand preference, and Internet use. These measures are shown in Appendix 1. Each item of each construct was measured using a seven-point rating scale, ranging from strongly disagree to strongly agree. The only exception was brand attitude, which was measured along five dimensions (good-bad, like-dislike, pleasantunpleasant, high quality-low quality, and good value-poor value) adapted from Mitchell and Olson (1981) and Holbrook and Batra (1987), using seven-point scales to represent brand image. Overall, the database for this study includes 140 separate measures across fourteen Web brands across three time periods. Data Analysis Partial least squares (PLS) modeling was used to test the hypotheses. PLS is a structural equation modeling technique with several key advantages over LISREL for the present study. PLS modeling accommodates the use of categorical variables and, in contrast to LISREL, does not demand a large sample size (Chin 1998). Further, PLS avoids factor indeterminacy and inadmissible solutions (Fornell and Bookstein 1982; Gopal, Bostrom, and Chin 1992). It is well suited for marketing data which often do not satisfy the requirements of multinormality, interval scaling, or the sample size for maximum likelihood estimation (Fornell and Bookstein 1982). PLS results are also easier to interpret, as path coefficients are like regression weights (equivalent to standardized beta weights in regression) and indicator loadings as principal component loadings within the context of the model (Falk and Miller 1992; Gopal, Bostrom, and Chin 1992). Despite its advantages, PLS modeling has certain limitations. PLS is not as robust as LISREL and does not focus on estimating causal paths. Instead, a key purpose of PLS is to test portions of a theoretical model, making it better suited for exploratory rather than confirmatory research (Chin 1998). In conducting PLS analysis, constructs can be defined as either reflective or formative. Reflective constructs are typical

M. Sksjrvi, S. Samiee / Journal of Interactive Marketing 25 (2011) 169177

173

of classical test theory and factor analysis models and are invoked in an attempt to account for observed variances or covariances (Fornell and Bookstein 1982). Preference was modeled as a reflective construct, as its indicators are assumed to covary. In contrast, formative constructs are not designed to account for observed variables; that is, they are used to minimize residuals in the structural relationship (Chin 1998). Here, the constructs are conceived as consisting of separate components that need not covary, as they measure different parts of the construct (Chin 1998). In the present study, brand identity, brand image, and Internet use were measured as formative constructs. All of these variables consist of independent components that jointly form the constructs. The model fulfills the criteria for the validation of formative constructs, as the main dependent variable is represented by a reflective construct (Diamantopoulos and Winklhofer 2001). Results The results for the measurement model are shown in Table 1. To validate the reflective indicators, factor loadings, average variance extracted (AVE), and composite reliability are examined. The results pertaining to preference, a reflective measure in our model, are also shown in Table 1. All values exceed their cut-off points; that is, .60 for loadings [], .70 for composite reliability [c], and .55 for AVE. For formative indicators, only the weights can be interpreted in the context of the model. The weights obtained in time 1 (t1), time 2 (t2), and time 3 (t3) are shown in Table 2. To test our hypotheses, we coded brand type (cyber vs. extension) and Internet stage (introduction vs. elaboration stage, elaboration vs. fortification stage, and introduction vs. fortification stage) as moderating variables. The moderating variables were estimated for each of the paths in the model, as shown in Fig. 1. The results are shown in Table 3. Hypothesis 1 posits that the decrease in the effect of brand identity on brand image between Internet stages is more pronounced for cyber than for offline-based extension brands. The brand type brand identity interaction on brand image was not significant between t1 and t2 (p N .10), but marginally significant between t2 and t3 ( = .353, t = 1.20, p b .10), as well as between t1 and t3 ( = .353, t = 1.20, p b .10). These results
Table 1 Loadings, composite reliability, and AVE for preference. Brand type Construct Indicators t1 Cyber brands Preference Visits Overall preference Likelihood of buying Certainty of buying Previous purchases Visits Overall preference Likelihood of buying Certainty of buying Previous purchases .76 .78 .85 .86 .69 .82 .65 .89 .84 .75 c

show that cyber brands have to work harder than offline-based extension brands to translate their identity into a successful brand image, especially when the market has developed. This effect is likely due to the positive synergistic effects offlinebased extension brands gain over time as spillover from their offline-based counterpart. The Internet stage brand identity interaction on brand image was significant and negative between t1 and t2 ( = .052, t = 2.58, p b .01), and marginally significant between t2 and t3 ( = .479, t = 1.37, p b .10) as well as between t1 and t3 ( = .353, t = 1.39, p b .10). These findings demonstrate that brands have to work less on translating an identity into a brand image as they develop on the Internet. Taken together, these results show that although cyber brands have to work harder than offline-based extension brands in translating their brand identity into a brand image, the advantage held by offline-based extension brands diminishes over time. The main effect of brand type was not significant between t1 and t2 (p N .10), but gained significance over time and became marginally significant between t2 and t3 ( = .519, t = 1.32, p b .10), as well as between t1 and t3 ( = .410, t = 1.32, p b .10). The main effect of Internet stage was significant between t1 and t2 ( = .121, t = 2.78, p b .01), and marginally significant between t2 and t3 ( = .513, t = 1.44, p b .10) as well as between t1 and t3 ( = .491, t = 1.57, p b .10). The main effect of brand identity was significant between t1 and t2 ( = .624, t = 9.90, p b .01), t2 and t3 ( = .263, t = 2.68, p b .01), and t1 and t3 ( = .543, t = 8.59, p b .01). These results provide support for H1. Hypothesis 2 posits that the decrease in the effect of brand image between the Internet stages is more pronounced for cyber than for offline-based extension brands. The brand type brand image interaction on brand preference was not significant at any time ( p N .10). The Internet stage brand image interaction on brand preference was significant between t1 and t2 ( = .017, t = 1.64, p b .05) and between t1 and t3 ( = .611, t = 3.35, p b .01), but not between t2 and t3 ( p N .10), showing that brands, in general, have to work harder on building a brand image in their early stages on the Internet. The main effects of brand type were not significant at any time ( p N .10), whereas the main effect of Internet stage was marginally significant between t1 and t2 ( = .117, t = 1.50, p b .10), significant between t1 and t3 ( = .468, t = 3.07, p b .01), and non-significant between t2 and t3 ( = .190, t = .27, p N .10). These results provide partial support for H2.

t2 AVE .62 .72 .67 .72 .87 .78 .84 .84 .68 .69 .78 c .87 AVE .57 .89

t3 .75 .67 .80 .72 .74 .67 .74 .79 .87 .70 c .86 AVE .55

Extension brands

Preference

.90

.63

.88

.59

.87

.57

174

M. Sksjrvi, S. Samiee / Journal of Interactive Marketing 25 (2011) 169177

Table 2 Changes in measurement model over time. Cyber brands Construct Brand identity Indicators Brand awareness Brand purpose Brand differentiation Brand offerings Brand credibility Brand character Overall attitude toward the brand Brand feelings Usage history Time spent (use duration) Number of purchases (familiarity with channel) Spending Visits Overall preference Likelihood of buying Certainty of buying Previous purchases Weight t1 .314 .461 .344 .273 .160 .094 .005 .790 .142 .210 1.169 1.717 .192 .346 .289 .286 .136 Weight t2 .496 .724 .192 .124 .311 .200 .130 .637 .562 .454 .409 .770 .169 .328 .194 .373 .254 Weight t3 .422 .884 .176 .272 .736 .062 .367 .093 .092 .108 .185 .824 .292 .335 .282 .174 .270 Offline-based extension brands Weight t1 .356 .685 .029 .111 .712 .149 .064 .264 .172 .531 .916 .225 .233 .267 .294 .286 .177 Weight t2 .093 .724 .118 .302 .229 .376 .362 .991 .231 .852 .166 .581 .303 .256 .316 .216 .216 Weight t3 .146 .798 .183 .229 .488 .157 .386 .365 .548 .884 .314 .487 .128 .321 .328 .319 .203

Brand image

Internet use

Preference

Hypothesis 3 proposes that the decrease in the effect of Internet use between the Internet stages is more pronounced for cyber than for offline-based extension brands. We did not find support for this hypothesis. The only significant interaction effect was between t2 and t3 for the brand type Internet use, which was marginally significant ( = .311, t = 1.32, p b .10). The main effect of Internet use was non-significant between t1 and t2 (p N .10), significant between t2 and t3 ( = .561, t = 2.56, p b .01), and marginally significant between t1 and t3 ( = .086, t = 1.27, p b .10). H3 is thus not supported. Validation As our database includes two purveyors of services from different industries (i.e., Citibank and Expedia), we reexamined our findings by excluding these two brands. The industries to which the remaining pure online and offline-based brands belong were matched and, therefore, consistent. From among the leading brands reported by Stores (Reda 1999),
Table 3 Changes in model across all timelines. t1 Brand type brand identity brand image Internet stage brand identity brand image Brand type Internet stage Brand identity Brand type brand image brand preference Internet stage brand image brand preference Internet stage Internet use brand preference Brand type Internet use brand preference Brand type Internet stage Brand image Internet use .336 .052 .292 .121 .624 .369 .017 .054 .175 .293 .117 .518 .078
vs t2

Citibank and Expedia were broadly matched as service firms, although the industries in which they operate (travel and banking) may give rise to differences in brand assessment. We conducted a robustness check to determine whether our results for the proposed model would remain consistent if these two brands were excluded from analysis. The results in Table 4 show that there are slight differences in the beta coefficients without these brands. These differences improved the proposed model. The proposed effects for H1 are strengthened (see Table 4) by increasing the significance level of the brand type brand identity interaction effect as well as the Internet stage brand identity interaction effect on brand image from p b .10 to p b .05. The main effect of brand type was strengthened in the same manner (i.e., the significance level increased from p b .10 to p b .05). These results do not change the findings reported earlier with Expedia and Citibank included in the data; however, their exclusion improved the proposed model. Therefore, our proposed model can be considered robust across brands.

t2 t .73 2.58 .66 2.78 9.90 .58 1.64 .29 1.17 .46 1.50 5.74 1.06 p Ns b.01 Ns b.01 b.01 Ns b.05 Ns Ns Ns b.10 b.01 b.10

vs t3

t1 T 1.22 1.37 1.32 1.44 2.68 .12 .45 .15 1.32 .89 .27 2.44 2.56 p b.10 b.10 b.10 b.10 b.01 Ns Ns Ns b.10 Ns Ns b.01 b.01

vs t3

Hypothesis t 1.26 1.39 1.32 1.57 8.59 .59 3.35 .05 .84 .48 3.07 6.30 1.27 p b.10 b.10 b.10 b.10 b.01 Ns b.01 Ns Ns Ns b.01 b.01 b.10 H1 H1 Main effect Main effect Main effect H2 H2 H3 H3 Main effect Main effect Main effect Main effect

.504 .479 .519 .513 .263 .069 .302 .035 .311 .515 .190 .297 .561

.405 .434 .410 .491 .543 .302 .611 .010 .134 .245 .468 .594 .086

M. Sksjrvi, S. Samiee / Journal of Interactive Marketing 25 (2011) 169177 Table 4 Model robustness check: Expedia/Citibank excluded. t1 Brand type brand identity brand image Internet stage brand identity brand image Brand type Internet stage Brand identity Brand type brand image brand preference Internet stage brand image brand preference Internet stage Internet use brand preference Brand type Internet use brand preference Brand type Internet stage Brand image Internet use .328 .055 .288 .124 .637 .352 .022 .084 .168 .287 .125 .527 .090
vs t2

175

t2 t .71 2.60 .70 2.76 9.92 .77 1.65 .36 1.13 .53 1.60 5.78 1.10 p Ns b.01 Ns b.01 b.01 Ns b.05 Ns Ns Ns b.10 b.01 b.10

vs t3

t1 T 1.64 1.66 1.72 1.44 2.68 .84 .63 .30 1.40 .85 .26 2.32 2.49 p b.05 b.05 b.05 b.10 b.01 Ns Ns Ns b.10 Ns Ns b.01 b.01

vs t3

Hypothesis t 1.65 1.70 1.82 1.55 8.62 .84 3.01 .05 .90 .53 3.00 6.35 1.28 p b.05 b.05 b.05 b.10 b.01 Ns b.01 Ns Ns Ns b.01 b.01 b.10 H1 H1 Main effect Main effect Main effect H2 H2 H3 H3 Main effect Main effect Main effect Main effect

.553 .552 .523 .511 .290 .122 .289 .046 .310 .510 .189 .286 .546

.522 .423 .511 .488 .555 .332 .590 .009 .145 .257 .452 .599 .088

Discussion and Conclusions Our goal in this study was to investigate consumer perceptions of online retail brands by focusing on the shift in perceptions over three main variables (brand identity, brand image, and Internet use) as brands transition to different stages on the Internet. We examined these changes across two different types of brands: cyber brands and offline-based extension brands. The present research contributes to the extant literature on branding by showing that offline-based brands have an advantage over cyber brands when it comes to translating a brand identity into a successful brand image. Traditional offline-based brands gain positive spillover effects from their offline-based counterparts; however, such effects take time and are not evident in the early Internet stage. Both types of brands have to work hard in the early Internet stage to create a successful brand image that manifests into consumer preference for the brand. With regards to Internet use, we found that cyber brands had a slight disadvantage between t2 and t3. (i.e., when moving from the elaboration stage to the fortification stage), probably due to the umbrella branding strategies used by offline-based brands. The results show that for cyber brands, the largest effort involves building brand identity, especially in the introductory Internet stage (i.e., when the market is developing). Cyber brands need to create market awareness and brand equity from inception and, for these brands, a focus on brand identity is more important than a focus on other brand components. In contrast, marketing strategies for offline-based extension brands should encourage consumers to shop online in addition to the physical retail locations, with the ultimate goal of using the extension brand as a means of gaining additional market penetration and share. Theoretical and Managerial Implications Our study makes a theoretical contribution by demonstrating that offline-based extension brands gain positive spillover effects from their brick-and-mortar parents, but that such effects take

time to materialize. Further, our results indicate that the advantage held by the offline-based brick-and-mortar brands does not extend itself into brand image; both types of brands have to make an effort to create a successful brand image, especially in the early Internet stages. Our results also show that when moving from the elaboration stage to the fortification stage, cyber brands have a slight disadvantage with respect to Internet use, as this may be the time when offline-based brands engage in expansion via umbrella strategies. These results bring novel insight into the literature on branding, and suggest insights for researchers who wish to take a longitudinal perspective on branding efforts. For managers, this study shows that consumers may react differently to brand-building efforts across Internet stages. In the early stages, much effort needs to be expended on establishing a viable identity (especially in the case of cyber brands) and a favorable image for the brand. In later stages, the focus should shift toward cross-channel strategies to gain favorable spillover effects. Cyber brands, by virtue of lacking offline-based counterparts, can imitate this strategy by cross-advertising their brands and cooperating with offline-based entities, e.g., by engaging in cobranding or ingredient branding strategies. Such strategies provide consumers with opportunities to associate cyber brands with the more tangible offline-based entities. In general, brand management over time poses several managerial problems. Awareness regarding the manner in which consumer preferences for online brands change over time affords firms the opportunity to more effectively manage brand-building efforts, for example, by allocating marketing resources and brand-building budgets to the branding components that will have the greatest influence. Limitations Our results reflect several shortcomings in the design and implementation of this research. First, as with any cyclical concept, the necessary time period to transition from the introductory period to later stages is undefined. Although our estimates of the necessary time lapse between stages are

176

M. Sksjrvi, S. Samiee / Journal of Interactive Marketing 25 (2011) 169177

adequately long and reasonable, the newness of the Internet makes such estimates particularly vulnerable to error. Thus, we cannot be certain that longer (shorter) periods between stages of data collection would not result in somewhat different outcomes. Second, despite our efforts to closely match cyber and offlinebased extension brands, the retail brands comprising cyber and offline-based extension brands are not identical. In the case of Expedia and Citibank, which operate in different industries, we set these firms aside and successfully retested and validated our model. Still, a comparison of non-identical sets of entities over time may in part explain the differences in effects across brand

types and Internet stages. That is, we cannot rule out alternative reasons for differences in our results. Third, our three samples of respondents were matched along several key dimensions important to this investigation; however, the subjects and response rates differed somewhat over time. Ideally, data should be gathered from identical samples across time periods. Fourth, alternative paths and explanations for differences across Internet stages cannot be ruled out. As such, non-causality is also a limitation in our study. Finally, single-item measures, even when used in formative constructs, are prone to measurement error and future research should make greater use of multi-item measures.

Appendix 1. Variables Used in the PLS Models Respondents were asked to indicate the extent to which they agree/disagree with the following statements on 7-point scales for each of the seven cyber brands and seven offline-based extension brands (i.e., 140 individual measures). Separate questionnaires for cyber and offline-based extension brands were administered three weeks apart. The same questions and brands were used in 2000, 2005 and 2009. The constructs and their corresponding question(s) are shown below: Operationalization of constructs.
Construct Brand identity Items (summated scales across brand types) Brand awareness Brand purpose Brand differentiation Brand offerings Brand credibility Brand character Overall attitude toward the brand Question (identical questions for each of the 14 brands for t1 and t2) I am familiar with this web brand This web brand has a well-defined use/purpose This web brand is uniquely different from its competitors (stores, catalogs, or web sites) This web brand offers a broad range of products This web brand is credible This web brand lacks character (reverse coded) 5-item measure (7-point scale), summated for each brand Good-Bad Likedislike Pleasantunpleasant High Qualitylow quality Good Valuepoor value I have a good feeling about this web brand How long have you been using the Internet? Approximately how much time do you spend on the Internet each week? During 1999/2004, how many times have you purchased something for personal or household use using the Internet? How much have you spent on personal or household purchases on the Internet in 1999/2004? I visit this web site frequently I prefer this web brand over others like it It is likely that I will purchase this brand/do business with this web brand I definitely will buy from this web brand I have previously purchased from this web brand

Brand image

Internet use

Brand feeling Usage history Time spent (use duration) Number of purchases (familiarity with channel) Spending

Brand preference

Visits Overall preference Likelihood of buying Certainty of buying Previous purchases

References
Aaker, David A. and Kevin L. Keller (1990), Consumer Evaluations of Brand Extensions, Journal of Marketing, 54, January, 27-41. and (1996), Building Strong Brands. New York: The Free Press. , , and Erich Joachimstaler (2000), Brand Leadership. New York: The Free Press. De Asis, V. Karen (2001), Managing Brands Over Time. Business World. March 22). Bart, Yakov, Venkatesh Shankar, Fareena Sultan, and Glen L. Urban (2005), Are the Drivers and Role of Online Trust the Same for All Web Sites and Consumers? A Large-Scale Exploratory Empirical Study, Journal of Marketing, 69, 4, 13352.

Boush, David M. and Barbara Loken (1991), A Process-Tracing Study of Brand Extension Evaluation, Journal of Marketing Research, 28, February, 16-28. de Chernatony, Leslie (1999), Brand Management through Narrowing the Gap between Brand Identity and Brand Reputation, Journal of Marketing Management, 15, 13, 15779. Chin, Wynne W. (1998), The Partial Least Squares Approach to Structural Equation Modeling, in Modern Methods for Business Research, George A. Marcoulides, ed. Mahwah, NJ: Lawrence Erlbaum Associates, 295336. Churchill, Gilbert A. (1999), Marketing Research: Methodological Foundations. New York: The Dryden Press. Degeratu, Alexandru M., Arvind Rangaswamy, and Jianan Wu (2000), Consumer Choice Behavior in Online and Traditional Supermarkets: The

M. Sksjrvi, S. Samiee / Journal of Interactive Marketing 25 (2011) 169177 Effects of Brand Name, Price, and Other Search Attributes, International Journal of Research in Marketing, 17, 1, 5578. Diamantopoulos, Adamantios and Heidi M. Winklhofer (2001), Index Construction with Formative Indicators: An Alternative to Scale Development, Journal of Marketing Research, 38, 2, 26977. Eastlick, Mary Ann (1996), Consumer Intention to Adopt Interactive Teleshopping, Working Paper No. 96113. Marketing Science Institute. and Sherry Lotz (1999), Profiling Potential Adopters and NonAdopters of an Interactive Electronic Shopping Medium, International Journal of Retail and Distribution Management, 27, 20923. Falk, R. Frank and Nancy B. Miller (1992), A Primer for Soft Modeling. Akron, Ohio: The University of Akron Press. Fornell, Claes and Fred L. Bookstein (1982), Two Structural Equation Models: LISREL and PLS Applied to Consumer Exit-Voice Theory, Journal of Marketing Research, 19, 4, 44052. Gopal, Abhijit, Robert P. Bostrom, and Wynne W. Chin (1992), Applying Adaptive Structuring Theory to Investigate the Process of Group Support Systems Use, Journal of Management Information Systems, 9, 3, 4569. Holbrook, Morris B. and Rajeev Batra (1987), Assessing the Role of Emotions as Mediators of Consumer Responses to Advertising, Journal of Consumer Research, 14, 3, 40420. Keller, Kevin L. (1993), Conceptualizing, Measuring, and Managing Customer-Based Brand Equity, Journal of Marketing, 57, 122.

177

(2003), Strategic Brand Management. 2nd ed. New Jersey: PrenticeHall. Martnez, Eva and Leslie de Chernatony (2004), The Effect of Brand Extension Strategies on Brand Image, Journal of Consumer Marketing, 21, 4, 3950. McEnally, Martha R. and Leslie de Chernatony (1999), The Evolving Nature of Branding: Consumer and Managerial Considerations, Academy of Marketing Science Review, 99, 2, 138. Mitchell, Andrew A. and Jerry C. Olson (1981), Are Product Attribute Beliefs the Only Mediator of Advertising Effects on Brand Attitude?" Journal of Marketing Research, 18, 3, 31832 (August). Park, C. Whan, Bernard J. Jaworski, and Deborah J. Maclnnis (1986), Strategic Brand Concept-Image Management, Journal of Marketing, 50, 4, 13545. Reda, S. (1999), Groundbreaking Consumer Survey Tracks Nation's Biggest On-line Merchants, Stores, 81, 9, V2 (September). Sksjrvi, Maria and Saeed Samiee (2007), Non-Price Antecedents of Consumer Preference for Cyber and Extension Brands, Journal of Interactive Marketing, 21, 1, 2235. Shankar, Venkatesh, Pablo Azar, and Matthew Fuller (2008), Practice Prize PaperBRAN*EQT: A Multicategory Brand Equity Model and its Application at Allstate, Marketing Science, 27, 4, 56784. Shim, Soyeon, Mary Ann Eastlick, Sherry L. Lotz, and Patricia Warrington (2001), An Online Prepurchase Intentions Model: The Role of Intention to Search, Journal of Retailing, 77, 397416.