8
JOURNAL OF CONSUMER PSYCHOLOGY, 14(1&2), 124-13 1 Copyright O 2004, Lawrence Erlbaum Associates, Inc. A Cognitive Model of Customer-Based Brand Equity for Frequently Purchased Products: Conceptual Framework and Empirical Results Girish N. Punj Department of Marketing University of Connecticut Clayton L. Hillyer Department of Marketing American International College The main purpose of this research is to identify the underlying cognitive structureof brandequity. Existing research on brand equity is used to identify 4cognitive"components"of customer-based brand equity.These are labeled asglobal brandattitude,strength ofpreference, brandknowledge, and brand heuristic. A conceptual frameworkof how these components(or subconstructs)are in- terrelated is proposed and empiricallytestedusing data from 2 frequentlypurchased productcate- gories. Covariance structure modeling is used as the analysis methodology. The results indicate that all the identified cognitivecomponentsare importantdeterminantsof customer-basedbrand equity. Specifically, the brand heuristic component serves as an important mediator in 2 "cogni- tive chains" that link global brand attitude to brand knowledge and global brand attitude to strength of preference,respectively. The study findings have important implications for design- ing equity maintenance strategies for frequently purchased products. Brand equity, a measure of the overall value of a brand (Keller, 1998), is a key concept in brand management. From a theoretical standpoint, researchers are interested in under- standing what brand equity denotes and how it is constituted. From a practical perspective, managers want to know how brand equity can be developed and maintained through the marketing activities of the firm. Despite extensive research, agreement on the definition and operationalization of customer-based brand equity re- mains somewhat elusive. A possible reason for this relates to the broadly inclusive manner in which the construct is frequently conceptualized. The dominant view is that cus- tomer-based brand equity is a set of brand-related associa- tions held by the consumer in memory (e.g., Keller, 1993). Under this perspective, brand equity is regarded as being largely attitudinal in nature, composed of beliefs, affect, and other subjective experiences related to the brand (i.e., brand attitude, brand image, etc.). Others have sought to include additional consumer con- structs, such as brand loyalty, purchase intentions, and brand commitment as part of its conceptualization (e.g., Aaker, 1991). They have argued that these constructs are also key to understanding customer-based brand equity, particularly as it relates to purchase behavior. In other words, brand equity is more than just a network of brand-related associations. Brand-equity research is not without its critics. Some have questioned the theoretical usefulness of such an all-inclusive approach to the conceptualization of customer-based brand equity (e.g., Barwise, 1993) and wondered whether it adds value to what is already known about consumer behavior through the use of traditional constructs such as brand loyalty and attitude toward the brand. Because of these concerns, there is a need for additional research on the cognitive under- pinnings of the brand-equity construct. In this article, we attempt to conceptually identify and em- pirically test the interrelations between the cognitive "com- ponents" of customer-based brand equity for frequently pur- Requests for reprints should be sent to Girish Punj, Department of chased products. Some important theoretical questions arise: Marketing, University of Connecticut, 2100 Hillside Road, Storrs, What are these cognitive components? How do they link with CT 06269-1041. E-mail: [email protected] one another? And, particularly, how do they link with the

A Cognitive Model of Customer-Based Brand Equity for Frequently Purchased Products: Conceptual Framework and Empirical Results

Embed Size (px)

Citation preview

JOURNAL OF CONSUMER PSYCHOLOGY, 14(1&2), 124-13 1 Copyright O 2004, Lawrence Erlbaum Associates, Inc.

A Cognitive Model of Customer-Based Brand Equity for Frequently Purchased Products: Conceptual

Framework and Empirical Results

Girish N. Punj Department of Marketing University of Connecticut

Clayton L. Hillyer Department of Marketing

American International College

The main purpose of this research is to identify the underlying cognitive structure of brandequity. Existing research on brand equity is used to identify 4cognitive "components"of customer-based brand equity. These are labeled asglobal brandattitude, strength ofpreference, brandknowledge, and brand heuristic. A conceptual frameworkof how these components (or subconstructs) are in- terrelated is proposed and empirically testedusing data from 2 frequently purchased productcate- gories. Covariance structure modeling is used as the analysis methodology. The results indicate that all the identified cognitive components are important determinants of customer-based brand equity. Specifically, the brand heuristic component serves as an important mediator in 2 "cogni- tive chains" that link global brand attitude to brand knowledge and global brand attitude to strength of preference, respectively. The study findings have important implications for design- ing equity maintenance strategies for frequently purchased products.

Brand equity, a measure of the overall value of a brand (Keller, 1998), is a key concept in brand management. From a theoretical standpoint, researchers are interested in under- standing what brand equity denotes and how it is constituted. From a practical perspective, managers want to know how brand equity can be developed and maintained through the marketing activities of the firm.

Despite extensive research, agreement on the definition and operationalization of customer-based brand equity re- mains somewhat elusive. A possible reason for this relates to the broadly inclusive manner in which the construct is frequently conceptualized. The dominant view is that cus- tomer-based brand equity is a set of brand-related associa- tions held by the consumer in memory (e.g., Keller, 1993). Under this perspective, brand equity is regarded as being largely attitudinal in nature, composed of beliefs, affect, and other subjective experiences related to the brand (i.e., brand attitude, brand image, etc.).

Others have sought to include additional consumer con- structs, such as brand loyalty, purchase intentions, and brand commitment as part of its conceptualization (e.g., Aaker, 1991). They have argued that these constructs are also key to understanding customer-based brand equity, particularly as it relates to purchase behavior. In other words, brand equity is more than just a network of brand-related associations.

Brand-equity research is not without its critics. Some have questioned the theoretical usefulness of such an all-inclusive approach to the conceptualization of customer-based brand equity (e.g., Barwise, 1993) and wondered whether it adds value to what is already known about consumer behavior through the use of traditional constructs such as brand loyalty and attitude toward the brand. Because of these concerns, there is a need for additional research on the cognitive under- pinnings of the brand-equity construct.

In this article, we attempt to conceptually identify and em- pirically test the interrelations between the cognitive "com- ponents" of customer-based brand equity for frequently pur-

Requests for reprints should be sent to Girish Punj, Department of chased products. Some important theoretical questions arise:

Marketing, University of Connecticut, 2100 Hillside Road, Storrs, What are these cognitive components? How do they link with CT 06269-1041. E-mail: [email protected] one another? And, particularly, how do they link with the

CUSTOMER-BASED BRAND EQUITY 1 25

brand name (or brand mark or logo) around which the asso- ciative memory network of brand-related associations is pre- sumed to exist?

The scope of this research is limited to established prod- ucts because it is for these products that the cognitive compo- nents of brand equity are likely to be the most well devel- oped. In addition, the brand-equity components are examined in the context of purchase behavior for these prod- ucts. Consequently, the implications from this research are mainly relevant for formulating strategies for brand-equity maintenance (as opposed to brand-equity development).

The primary approach to the conceptualization of cus- tomer-based brand equity (which hereafter is abbreviated as brand equity for ease of exposition) focuses on the con- sumer's brand associations. Here, the brand name (or brand mark or logo) plays an important role in the cognitive struc- ture underlying brand equity because it serves as the central node around which these brand associations form an associa- tive memory network. In essence, favorable brand associa- tions, including product-related (i.e., high quality), non-product-related (i.e. the spokesperson is cool), cogni- tive, and affective associations lead to positive brand evalua- tions or perceptions that can include affective dimensions. The effect of positive brand evaluations on brand equity is moderated by their accessibility in memory. Highly accessi- ble evaluations influence perception and subsequent equity development in the direction of existing evaluations (e.g., Fazio, 1986). These brand associations underlie the con- sumer's differential preference for a brand and subsequent behavior toward the brand.

Others have tried to further extend brand equity by includ- ing constructs, such as brand loyalty, brand awareness, per- ceived quality, in addition to brand associations (e.g., Aaker, 1991; Keller, 1993). However, these frameworks tend not to specify a formal structure among these additional constructs, choosing instead to list them as brand-equity elements. Such a specification can be conceptually problematic because the same construct then appears to play multiple roles. For in- stance, brand loyalty has been regarded as both a dimension and an outcome of brand equity (Morgan, 2000).

PROPOSED COGNITIVE MODEL OF BRAND-EQUITY "COMPONENTS

Our goal is to develop a structural model of the cognitive components of brand equity. To this end, we use a pro- cess-model approach to combine attitudiizal and memory as- pects (e.g., Keller, 1993) and behavioral intention and con- sumption aspects (e.g., Maltz, 1991) of brand equity that can be related to purchase behavior. Specifically, if one classifies constructs along a cognitive-attitudinal-behavioral contin- uum, constructs more closely associated with attitude and memory are antecedents to constructs that lie closer to behav- ioral intention and consumption.

Over time, however, the proposed constructs and their in- terrelations have a recursive and reciprocal nature to them (especially for repeated, low-involvement product catego- ries). For example, consumption experience influences sub- sequent beliefs by reinforcing or altering them, which may, in turn, influence brand loyalty. However, the process-model described next is presented as a nonrecursive unidirectional model for the sake of conceptual clarity.

The first construct is the global brand attitude component, which is defined as the highest order (i.e., most inclusive and most accessible) abstract evaluation of a brand that subsumes specific brand associations and represents the combined ef- fect of all such associations. It is highly likely that global brand attitude is directly linked to the "central node" around which brand equity is formed (Keller, 1993). As mentioned earlier, the node serves as the activation mechanism for the subsequent retrieval of all other brand associations such as in an associative model of memory.

In every conceptualization of brand equity, brand attitude, however termed, plays a preeminent role in the representa- tion of brand equity (Dyson, Fan, & Hollis, 1996; Farquhar, 1989; Fazio, 1986). It is both highly accessible and diagnos- tic, and therefore consumers rely heavily on it in decision making, instead of attempting to recall and process specific brand associations (Lynch, Marmostein, & Weigold, 1988).

The next brand-equity component is brand heuristic, which can be defined as decision rules or heuristics used by the consumer that favor brands with high global brand atti- tude (i.e., brand equity). Global brand attitude's influence is mediated by the decision rules used by the consumer. In other words, the impact of brand associations is dependent on both their favorableness and their role in decision making. Sup- port for this comes from the heuristic-systematic informa- tion processing framework (e.g., Chaiken, 1987). In this framework, evaluative judgments (i.e., attitudes) available as input for decision making can be mediated by a decision rule. For example, heuristics such as "Let me buy the most famil- iar name" or "Which brand do I like the most?'tap global brand attitude, making it diagnostic. Decision rules such as "Let's buy what's on sale7' or "Buy the cheapest brand" may negate the salience of global brand attitude.

Furthermore, it is important to note that a favorable global brand attitude may promote the use of decision rules based on global brand attitude. Activation of global brand attitude is likely to activate heuristics used for the product category. Given favorable brand equity, we can expect the activated heuristic to tap associations that are available, to reduce cog- nitive effort, and to be diagnostic. Even in case of more delib- erative decision making, the salience of the brand name and attitude is likely to influence the choice of attributes to evalu- ate and the recall of other brand associations and equity com- ponents (Lynch et al., 1988).

Next, we turn to the behavioral intention and consump- tion aspects of brand equity, which can either relate to the consumption experiences with the brand or the strength of

126 PUNJ AND HILLYER

preference for it. We capture these two influences by using two components (i.e., subconstructs) that are labeled as brand knowledge and strength of preference. Both are influ- enced directly by brand heuristic and only indirectly by global brand attitude.

The brand knowledge may be defined as the accumu- lated experience with the brand, both direct and indirect. Research has shown that the level of consumption experi- ence is an important outcome of a favorable brand attitude and the memorability of the brand name or brand mark or logo (Alba & Hutchinson, 1987). Furthermore, repeated and satisfactory purchase experiences for these products can be expected to strengthen global brand attitude and the use of brand heuristics (e.g., Park & Srinivasan, 1994).

However, repeated purchases of a brand alone may not necessarily imply high levels of brand equity. Some routine purchases can be due to inertia that is no longer based on underlying preference. Also, situational factors (availabil- ity, price deals, etc.) that favor the use of non-brand-related heuristics may be more basic determinants of purchase be- havior than brand equity. In other words, whereas brand consumption can normally be expected to be strongly asso- ciated with brand equity, it may not be a sufficient condi- tion for the existence of brand equity (Maltz, 1991).

The strength of preference component is defined as the intensity of preference for the brand in comparison to other substitute brands that belong to the same product category. It captures the consistent predisposition to prefer a particular brand over others (i.e., relative brand equity). It serves as a proxy for the behavioral component of brand loyalty (e.g., Oliver, 1999). It also serves to help distin- guish inertia-driven repeat purchase behavior from brand-equity-driven purchase behavior.

Favorable levels of global brand attitude normally im- ply a high preference intensity. But, the perceived avail- ability of superior or equivalent brand substitutes can at- tenuate this relation. Although strength of preference indicates brand commitment, we know well that the latter does not necessarily translate into behavior. It is possible to form preferences without actually purchasing or con- suming the target brand or brands. For example, a con- sumer repeatedly exposed to advertisements and favorable word-of-mouth for a brand may develop a strong prefer- ence for the brand but may not be a prospective user for other reasons (e.g., not being in the target market). Thus, it appears that the strength of preference and brand knowl- edge constructs, whereas individually not sufficient, are jointly sufficient as brand-equity components.

In sum, the brand heuristic plays a key role by func- tioning as a mediating influence between global brand at- titude on the one hand, and the strength of preference and brand knowledge constructs on the other. It assumes that the activation of the brand-name node in memory follows the conscious awareness of global brand attitude. More favorable brand attitudes are posited to lead to stronger

brand-based heuristics. The formulation also allows for the possible independence of the strength of preference and brand knowledge constructs, for the reasons men- tioned earlier. It depicts two possible cognitive chains be- tween the four basic constructs that form our framework of brand equity.

Finally, to further evaluate the proposed model, three other models are also tested (see Figure 1) because they represent potential alternative explanations of the relation between brand heuristic and global brand attitude. First, one might expect global brand attitude to have its own direct effects on brand knowledge and strength of preference because of its centrality to the notion of brand equity and the fact that attitude is often used as a heuristic by itself. Consequently, Model A allows for direct effects on the outcome constructs and Model B treats global brand attitude and brand heuristic as independent vari- ables with their own effects. Second, it is possible to ques- tion the ordering of the constructs. It is possible that global brand attitude mediates brand heuristic, as depicted in Model C .

Pmpaa#d Model

Heuristic

SIrsngm Of Preference

Competing Model A Brand

Knowledge

HBurisliC Attitude

Preference

Competing Model B

Knowledge Attitude

Strenglh of Heurislic Preferenoe

Competing Model C A - z i J Heuristic

Preference

FIGURE 1 Proposed and competing models of brand-equity components.

CUSTOMER-BASED BRAND EQUITY 127

METHOD

The models were tested using data from two independent samples of consumers for two different frequently purchased

categories-soap and toothpaste. These categories were selected because their frequent and routine purchase imply that brand-equity-related components should be sta- ble, based on direct experience, and are more likely to in- volve the use of heuristics.

The data were collected in two separate surveys corre- sponding to the two product categories. The soap data were collected in a campus store of a large public university with a rural location. Respondents were "exit interviewed at the lo- cal campus store during a time when there were no special promotional or pricing tactics. A systematic sampling proce- dure was used and the interviews were counterbalanced across time-of-day and day-of-week over the study period. The toothpaste data were collected on the campus of a small private university located in a city. A sample of 100 consum- ers was "entrance" interviewed as they entered the student union. Due to the urban location of the second campus, stu- dents were asked about their preferences and consumption regardless of where they shopped for their toiletries.

The response rates for soap and toothpaste product cate- gories was 66% and 80%, respectively. The majority of re- spondents in each survey reported using their most-preferred brand for over 2 years, which indicated a stable consumption environment.

Measures

Global brand attitude was operationalized using separate in- dicators of beliefs, affect, and behavioral intention toward the brand in a manner consistent with Farquhar (1989) and simi- lar to Keller (1987). The belief items were culled from a larger set of items.

The five dimensions used to measure soap (toothpaste) beliefs were (a) smells (tastes) good-bad, (b) cleans (whit- ens) well-not well, (c) moisturizes (prevents cavities) well-not well, (d) lathers (freshens breath) well-not well, and (e) high quality-poor quality. Affect, defined as an over- all evaluative state, was measured on four dimensions: (a) like-dislike, (b) good-bad, (c) best-worst, and (d) good value-poor value. Behavioral intention was measured as the likelihood of purchasing the most-preferred brand on the next shopping trip.

Brand heuristic was measured using two 7-point Likert-type scale items. One item measured the tendency to use a brand-name heuristic directly ("I make my purchase se- lection of bath soap according to my favorite brand name, re- gardless of price"). The other item, which was reverse scored, measures a multiple brand heuristic in which a partic- ipant selects a brand from among an evoked set based on some special criterion such as price or availability ("I have more than one preferred brand of bath soap, and I make my

selection based on price, availability, etc."). See Jacoby and Chestnut (1978) for a similar measure.

For the toothpaste sample, the multiple-brand heuristic was replaced with a price heuristic measure ("My choice of toothpaste is based largely on price") because it was felt that subjects were more likely to use a price heuristic because of the higher retail cost of toothpaste compared to a bar of soap.

Strength of preference was measured using two variables, which have also been used in brand loyalty research. The first variable was Cunningham's (1967) stated-brand preference measure and was treated as an ordinal item. It asks the partic- ipants whether they will extend their search, postpone their purchase, or buy another brand if their most-preferred brand is unavailable. A willingness to extend search was treated as exhibiting the greatest strength of preference. The second variable was Pessemier's (1960) price-until-switching mea- sure, which captures the price elasticity of the most-preferred brand by finding the threshold at which incrementally pre- sented price increases lead to the purchase of another brand.

Brand knowledge was also measured using two variables. The first measure was the percentage of purchases repre- sented by the most-preferred brand during the previous month. A pretest had determined that the natural purchase cycle in the product category was approximately 30 days. The second measure was the number (net) of non-most-pre- ferred brands purchased over the last 8 months. This was cal- culated by subtracting the number of purchases of the most-preferred brand from the total number of soap pur- chases over the last 8 months. The aforementioned variables were measured at the individual level in correspondence to the respondent's most-preferred brand. This was the brand the respondent purchased and used most often.

Brand knowledge was treated as a behavioral construct because our model presumes relatively stable purchase be- havior. With repeated consumption of a brand, the diagnostic value of each subsequent consumption experience dimin- ishes, unless there are major changes to the brand or the us- age situation. Thus, its influence is already established by global brand attitude and, at least conceptually, does not feed back into global brand attitude. One can expect brand knowl- edge to more closely resemble actual consumption and be- have more like a behavioral construct under stable consump- tion. Accordingly, its operationalization was more in tune with actual purchase behavior as opposed to brand or cate- gory beliefs.

RESULTS

Preliminary Analyses

The correlations, means and standard deviations for all the measures in both the soap and the toothpaste data sets are provided in Table 1 . Preliminary analysis results of the soap and toothpaste data were similar. One-way analyses of vari-

128 PUNJ AND HILLYER

TABLE 1 Correlations, Means, and Standard Deviations

Soap

-

Single Mulri % Most Price Stated Brand Brand Purchase Brand Brand Preferred No. Until Brand Beliefs Affect Intent Heuristic Heuristic Brand Non-MPB. Switch Preference

Soap Brand beliefs Brand affect Purchase intention Single brand heuristic Multi brand heuristic % Most preferred brand Number non-MPB purchased Price until switching Stated brand preference Means Standard deviations

Toothpaste Brand beliefs Brand affect Purchase intention Single brand heuristic Price heuristic % Most preferred brand Number Non-MPB purchased Price until switching Stated brand preference Means Standard deviations

ance (ANOVAs) revealed that only non-most-preferred brand purchases for toothpaste varied significantly by brand. Colgate and Crest users (56% of respondents) were more likely to use fewer non-most-preferred brands. All in all, the data is not biased by brand.

Congeneric measures of reliability for all measures were generated by estimating a measurement model for both toothpaste and soap using LISREL VII. Some measures may be below ideal levels; these measures represented the lower bound for indicator reliability (Jijreskog & Sorbom, 1989) and were deemed acceptable given the quasi-experimental nature of the research. The high coefficient of determination for independent variables and the low chi-square value for each model indicate strong measurement models (Bollen, 1989).

Convergent validity for both sets of data was satisfied as all indicator loadings were highly significant at p = .05 or better. All other constructs demonstrated acceptable discriminant validity (see Gerbing & Anderson, 1988) except for brand knowledge and brand heuristic for the toothpaste data. Nevertheless, the constructs were kept separate for sev- eral reasons. First, they are conceptually distinct from our theoretical perspective because heuristics are inputs and the consumption measures are outputs. Second, the two con- structs achieved statistical discriminant validity in the soap sample, and an important goal of the toothpaste sample was

to see whether these results could be replicated. Third, the modification indexes reported by LISREL, a diagnostic for improving the overall fit of the model, indicated that allow- ing the respective indicators to load on the other construct or constructs would not significantly improve the model.

Proposed Model

The proposed model exhibited a good overall fit to the soap data (see Table 2; x2 = 27.02 with 21 df; p = .17; adjusted goodness-of-fit index = .94; root mean square residual [RMSR] = .066). The largest standardized residual was 2.22 (purchase intention and percentage most-preferred brand). The percentage of residuals in magnitude greater than 2.0 was 2.8% (1136). There were no significant correlated errors among the disturbances.

All hypothesized paths were significant, and the signs of the coefficients were in the correct direction. Furthermore, the model supports the notion that global brand attitude and brand knowledge were mediated by brand-name heuristic. The coefficient for the indirect effect of global brand attitude on brand knowledge was positive and significant (.048; t = 1.65). This suggests that there is correspondence between global brand attitude and brand knowledge but that it is medi- ated by brand heuristic.

CUSTOMER-BASED BRAND EQUITY 129

TABLE 2 Covariance Structure Analysis Results

Model Soap Parameter Toothpaste Parameter

Proposed model Global brand attitude-+Brand heuristic 1.75" 1.36 Brand heuristic-+Brand knowledge .03b .08 Brand heuristic-+Strength of preference .28 .29 Chi-square 27.02 (df = 21) 25.18 (df = 24) P-value .17 .40 AGFI~ .94 .90 RMSRe ,066 ,057

Competing model A Global brand attitude+Brand heuristic 1.76 1.36 Global brand attitude-+Brand knowledge -.08 -.04C Global brand attitudejstrength of preference .1 I C .02C Brand heuristic4Brand knowledge .08 .08 Brand heuristic4Strength of preference .2 1 .28 Chi-square 30.62 (df = 22) 25.00 (df = 22) P-value .10 .30 AGFI .93 .89 RMSR ,084 ,057

Competing model B Global brand attitude4Brand knowledge -.08 -.04C Global brand attitude+Strength of preference .1lc .02C Brand heuristic-+Brand knowledge .08 .08 Brand heuristic+Strength of preference .21 .28 Chi-square 30.62 (df = 22) 25.00 (df = 22) p value .10 .30 AGFI .93 .89 RMSR ,084 ,057

Competing model C Brand heuristic4 Global brand attitude .2 1 .19b Global brand attitude -+Brand knowledge .10 .09b Global brand attitude +Strength of preference .5 1 .30 Chi-square 103.34 (df = 24) 230.24 p value .OO .OO AGFI .79 .48 RMSR .28 ,125

Note. aall path coefficients significant a t p < .01 or better unless noted. bsignificant at p < .05. %ot significant at p > .05. d~djus ted Goodness-of-Fit Index. eRoot Mean Square Residual

The model also fit the toothpaste data very well (see Table 2; x2 = 25.18 with 24df;p = .40; adjustedgoodness-of-fit index = .90; RMSR = .057). The largest standardized residual was 2.44 (purchase intention and percentage most-preferred brand). The percentage of residuals in magnitude greater than 2.0 was 5.6% (2136). There were no significant correlated er- rors among the disturbances. All hypothesized paths were sig- nificant, and the signs of the coefficients were in the correct di- rection. The toothpaste data replicates the soap results.

Competing Models

Competing Models A and B allowed for independent effects of global brand attitude on brand knowledge and strength of preference. The difference between them is that in Model A, brand heuristic is causally linked to global brand attitude (es- timating path parameter), whereas in Model B both are inde- pendent (estimating correlation). From a statistical stand-

point, they are equivalent models in that the overall fits of the models are the same.

For the soap data, Model A and B both fit the data (see Ta- ble 2; ~2 = 30.62 with 22 df; p = . l o ; adjusted goodness-of-fit index = .93). In both models, the global brand atti- tude-strength of preference link was not significant. Also, the link between global brand attitude-brand knowledge was negative.

For the toothpaste data, Model A and B both fit the data (see Table 2; x2 = 25.18 with 22 df; p = .30; adjusted good- ness-of-fit index = 3 9 ) . In both models, the links between global brand attitude-strength of preference and global brand attitude-brand knowledge were not significant.

The appropriate test for nested models is to test whether any reduction in the overall chi-square by estimating addi- tional parameters is significant. For the toothpaste data, adding the extra links did not significantly improve the model (Ax2 = .18, 2 dB. Unfortunately, it was necessary to

130 PUNJ AND HILLYER

fix some additional parameters such that the competing models have an additional degree of freedom, rather than two less than the proposed model. However, in both com- peting models, more parameters of interest are estimated. A comparison of the chi-square values also indicates that the models are not significantly different from the proposed model (Ax2 = .3.02, 1 df).

Competing Model C switched the roles of global brand at- titude and brand heuristic. For both the soap and toothpaste data, neither model fit the data. The overall chi-square values were 103.34 0, = .00,24 df) and 230.2 0, = .00,29 df), respec- tively. All of the links were positive and significant 0, < .05).

DISCUSSION

Overall, there is strong empirical evidence in support of the proposed model. The proposed model is superior to the com- peting models most clearly with the toothpaste data set as the competing Models A and B models are nested and do not provide a better fit. It can be argued that the soap data also provides strong supporting evidence because despite having more parameters of interest estimated, it does not signifi- cantly fit the data better than the proposed model. Competing Model C provides evidence that the order of the global brand attitude-brand heuristic link is as proposed.

There are several important implications that emerge from the model tests. First, the proposed model suggests that brand associations themselves are not enough to capture the notion of brand equity. If brand associations were enough, then one would have expected to see (a) significant links from global brand attitude to both brand knowledge and strength of pref- erence and/or (2) a lack of significance between global brand attitude and brand heuristic. Only the global brand atti- tudejbrand knowledge link for the soap data was signifi- cant. The coefficient was surprisingly negative, suggesting that attitude is not enough to predict purchase behavior in our data.

Second, the contention that brand heuristic mediates global brand attitude is strongly supported. This suggests that the real value of brand equity may reside in its capability to influence "second-order" cognitive constructs. It appears that the power of brand equity lies in its ability to influence components that are proximate to consumption.

Third, that both the global brand attitude + brand heuris- tic 4 brand knowledge and global brand attitude -+ brand-name heuristic + strength of preference cognitive chains provide a parsimonious empirical representation of the hierarchical sequence in which the equity components are related. It should be noted that in all the models tested, the LISREL VII modification indexes never indicated that there would be a significant improvement in the overall model if either of the brand knowledge-strength of preference links were estimated. In other words, they have a complimentary role in the representation of brand equity.

Obviously, the generalizability of our findings is limited but at least appears to be valid across two independent sam- ples and for two different product categories. Also, it should be remembered that over time, the outcome constructs should influence the input constructs, a feedback process that was not modeled in this study. As such, the role of strength of preference and brand knowledge, in particular, will be en- hanced.

SUMMARY AND IMPLICATIONS

The empirical findings allow one to speculate on the similari- ties between the components of brand equity and brand loy- alty. Brand loyalty is a relatively well-understood construct with an established research tradition. In contrast, research on brand equity is still in a state of evolution. Based on our findings, it appears that both share the same attitudinal com- ponents, but differ in terms of their behavioral associations. For instance, the notion of consumption is germane to any characterization of brand loyalty. But it may not be so for brand equity.

Also, it seems that the behavioral aspects of brand loyalty are difficult to isolate from a multitude of situational pur- chase influences that act as "equity eroders." Such a finding is consistent with evidence that consumers are loyal to a set of brands (e.g., Oliver, 1999), with actual choice being situationally determined just prior to purchase.

In addition, the strength and valence components of pref- erence appear to play different roles in the representation of brand equity and brand loyalty. Although preference is an in- gredient in both constructs, it takes on a different (and more potent) meaning in the case of brand equity. It seems that preference in the former instance embodies both the strength and valence dimensions. Given such a formulation, brand eq- uity can be regarded as being less subject to erosion than brand loyalty.

Another important distinction appears to be the transfer- ability of brand equity. No similar notion exists in the brand loyalty literature despite its long research tradition. This again points to the role of the strength dimension mentioned previously because transferability cannot occur without strength. Clearly, this is yet another important area for addi- tional research on brand equity.

There are several future research directions suggested. One direction is testing the model with a high-involvement product category. It is possible, as consumers engage in more deliberate decision making, that the role of the brand heuris- tic diminishes as consumers adopt more elaborate and active decision strategies.

Another direction is expanding and testing the role of brand knowledge and strength of preference as antecedents to global brand attitude and brand heuristics. This also brings up the differences in brand-equity development and mainte- nance. One would expect that brand knowledge, in particular,

CUSTOMER-BASED BRAND EQUITY 131

plays a more prominent role in the development of brand eq- uity because the consumer receives diagnostic information from early brand consumption. Thus, the model presented here may need to be adapted for cases of brand-equity devel- opment.

In sum, this article proposed a model for the cognitive components of customer-based brand equity that derive for related preference, choice, and consumption variables. Based on past brand-equity research, the framework attempts to disaggregate the construct into four structural components: global brand attitude, brand heuristic, brand knowledge, and strength of preference. The empirical findings indicated that the components of brand equity form two cognitive chains that link together two important stages in the preference re- sponse hierarchy.

REFERENCES

Aaker, D. A. (1991). Managing brand equity. New York: Free Press. Alba, J. W., & Hutchinson, J. W. (1987). Dimensions of consumer expertise.

Journal of Consumer Research, 13, 41 1 4 5 4 . Barwise, P. (1993). Brand equity: Snark or boojum? International Journal of

Research in Marketing, 10, 93-1 04. Bollen, K. (1989). Structural equations with latent variables. New York:

Wiley. Chaiken, S. (1987). The heuristic model of persuasion. In M. Zanna, J.

Olson, & C. Herman (Eds.), Social injluence: The Ontario Symposium (pp. 3 4 9 ) . Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.

Cunningham, S. M. (1967). Perceived risk and brand loyalty. In D. Cox, (Ed.), Risk taking and information handling in consumer behavior (pp. 507-523). Boston: Harvard University Press.

Dyson, P., Farr, A,, and Hollis, N. S. (1996). Understanding, measuring, and using brand equity. Journal of Advertising Research, 36(6), 9-21.

Farquhar, P. H. (1989, September 1). Managing brand equity. Marketing Re- search, 24-33.

Fazio, R. H. (1986). How do attitudes guide behavior? In R. M. Sonentino & E. T. Higgins (Eds.), Handbook of morivation and cognition: Foundations of social behavior (pp. 20-43). New York: Guilford.

Gerbing, D. W., & Anderson, J. C. (1988). An updated paradigm for scale development incorporating unidimensionality and its assessment. Journal of Marketing Research, 25, 186-192.

Jacoby, J., & Chestnut, R. W. (1978). Brand loyalty measurement and man- agement. New York: Wiley.

Joreskog, K. G., & Sorbom, D. (1989). LISREL VII: Analysis of linearstruc- tural relationships by maximum likelihood, instrumental variables, and least squcrres methods. Mooresville, IN: Scientific Software.

Keller, K. L. (1987). Memory factors in advertising: The effect of advertis- ing retrieval cues on brand evaluations. Journal of Consumer Research, 14, 316-333.

Keller, K. L. (1993). Conceptualizing, measuring, and managing cus- tomer-based brand equity. Journal of Marketing, 57(l), 1-22.

Keller, K. L. (1998). Strategic brand management: Building, measuring and managing brand equity. Upper Saddle River, NJ: Prentice Hall.

Lynch, J. G., Jr., Marmorstein, H., & Weigold, M. F. (1988). Choices from sets including remembered brands: Use of recalled attributes and prior overall evaluations. Journal of Consumer Research, 15, 225-233.

Maltz, E. (1991). Managing brand equity: A conference summary (Report No. 91-1 10). Cambridge, MA: Marketing Science Institute.

Morgan, R. P. (2000). A consumer-oriented framework of brand equity. In- ternational Journal of Market Research, 42, 65-78.

Oliver, R. L. (Ed.). (1999). Whence consumer loyalty [Special issue]. Jour- nal of Marketing, 63, 3 3 4 4 .

Park, C.-S., & Srinivasan, V. (1994). A survey-based method for measuring and understanding brand equity and its extendability. Journal of Mar- keting Research, 31, 271-288.

Pessemier, E. A. (1960). An experimental method for estimating demand. Journal of Business, 33, 373-383.