USC Marshall School of Business Marshall Research Paper Series Working Paper MKT 16-10 Brand Attachment and Brand Attitude Strength: Conceptual and Empirical Differentiation of Two Critical Brand Equity Drivers C.
W. Park Marshall School of Business, University of Southern California Deborah J. MacInnis Marshall School of Business, University of Southern California Joseph R. Priester Marshall School of Business, University of Southern California Andreas B. Eisingerich Imperial College London Dawn Iacobucci Owen Graduate School of Management, Vanderbilt University This paper can be downloaded without charge from theSocial Science Research Network electronic http://ssrn.
com/abstract=1605782 1 Electronic copy available at: http://ssrn. com/abstract=1605782 1 Brand Attachment and Brand Attitude Strength: Conceptual and Empirical Differentiation of Two Critical Brand Equity Drivers May 12, 2010 C. Whan Park Joseph A. DeBell Professor of Marketing ACCT 306C Marshall School of Business University of Southern California Los Angeles, CA 90089-0403 Phone: 213-740-7107; Fax: 213-740-7828 [email protected] usc. edu Deborah J. MacInnis Charles L.
and Ramona I. Hilliard Professor of Business Administration ACCT 306C Marshall School of BusinessUniversity of Southern California Los Angeles, CA 90089-0403 Phone: 213-740-5039 ; Fax: 213-740-7828 [email protected] edu Joseph Priester Associate Professor of Marketing ACCT 306C Marshall School of Business University of Southern California Los Angeles, CA 90089-0403 Phone: 213-821-5649; Fax: 213-740-7828 [email protected] usc. edu Andreas B. Eisingerich Assistant Professor of Marketing Imperial College Business School Imperial College London London, UK SW 7 2AZ Phone: +44(0)20-7594-9763; Fax: +44(0)20-7823-7685 a. [email protected]
ac. uk Dawn Iacobucci E. Bronson Ingram Professor in MarketingOwen Graduate School of Management 401 21st Avenue South Vanderbilt University Nashville, Tennessee 37203 Phone: 615-322-4075; Fax: 615- 343-7177 Dawn. [email protected]
vanderbilt. edu Forthcoming, Journal of Marketing Electronic Electronic copy available at: http://ssrn. com/abstract=1605782 2 Brand Attachment and Brand Attitude Strength: Conceptual and Empirical Differentiation of Two Critical Brand Equity Drivers Abstract Research has not verified the theoretical or practical value of the brand attachment construct in relation to alternative constructs, particularly brand attitude strength.The authors make conceptual, measurement, and managerial contributions to this research issue. Conceptually, they define brand attachment, articulate its defining properties, and differentiate it from brand attitude strength. From a measurement perspective, they develop and validate a parsimonious measure of brand attachment, test the assumptions that underlie it, and demonstrate that it indicates the concept of attachment. They also demonstrate the convergent and discriminant validity of this measure in relation to brand attitude strength.
Managerially, they demonstrate that brand attachment offers value over brand attitude strength in predicting (a) consumers’ intentions to perform difficult behaviors (those they regard as utilizing consumer resources), (b) actual purchase behaviors, (c) brand purchase share (the share of a brand among directly competing brands), and (d) need share (the extent to which consumers rely on a brand to address relevant needs including those brands in substitutable product categories). Keywords: Brand management, consumer behavior, marketing strategy, brand attachment, attitude strengthElectronic Electronic copy available at: http://ssrn. com/abstract=1605782 3 Academic researchers and practitioners in marketing have shown significant interest of late in studying consumers’ attachment to brands (Chaplin and Roedder John 2005; Park and MacInnis 2006; Schouten and McAlexander 1995; Thomson 2006).
As a construct that describes the strength of the bond connecting the consumer with the brand, attachment is critical as it should impact behaviors that foster brand profitability and customer lifetime value (Thomson, MacInnis, and Park 2005).At the same time, marketers have long invoked the constructs of attitude valence and strength as key antecedents to consumer behavior. Attitude valence is defined as the degree of positivity or negativity with which an attitude object (here a brand) is evaluated. Brand attitude strength is conceptualized as the positivity or negativity (valence) of an attitude weighted by the confidence or certainty with which it is held, i. e. , the extent to which it is seen as valid (Petty, Brinol, and DeMarree 2007).
Strong attitudes result from effortful thought about the attitude object (Petty and Cacioppo 1986), most often given its personal relevance. This effortful thought, and the confidence with which the attitude object is held, guide behavior. Brand attitude strength has been shown to predict behaviors of interest to firms, including; brand consideration, intention to purchase, purchase behavior, and brand choice (Fazio and Petty 2007; Petty, Haugtvedt, and Smith 1995; Priester et al. 2004). The rich history of research on brand attitude strength raises questions about the need for a construct such as brand attachment.
Does attachment provide value beyond measures of brand attitude strength? At present, the answer to this question is elusive, as research to date has not verified how brand attachment and brand attitude strength differ conceptually or empirically. Nor has research differentiated what unique consumer behaviors, if any, each predicts. The present research makes three key contributions pertinent to these issues. First, we differentiate the brand attachment construct from brand attitude strength conceptually, arguing Electronic copy available at: http://ssrn. com/abstract=1605782 that the two constructs have distinct conceptual properties and entail different formation processes. Second, we validate this distinction empirically, developing a novel scale that maps the conceptual properties of brand attachment and assessing its relationship to attitude strength.
Third, and most significantly, we empirically demonstrate that attachment and attitude strength have distinct behavioral implications. Brand attachment more accurately predicts intentions to perform behaviors that utilize significant consumer resources (time, monetary, reputational).It is also a stronger predictor of actual consumer behaviors than is brand attitude strength. These effects are observed in terms of consumer purchase behavior, brand purchase share (i. e. , choice among directly competing brands), and need share (i. e. , choice among brands targeting similar needs), and they are observed even after controlling for consumer inertia (i.
e. , past behaviors) and other potential factors. Beyond their theoretical significance, our results have significant managerial implications, suggesting that brand attachment serves as the ultimate destination for customer-brand relationships.As far as we are aware, this is the first article to examine this diverse set of behavioral outcomes from brand attachment and it is the first to demonstrate these effects in relation to attitude strength.
Conceptual Distinction between Brand Attachment and Brand Attitude Strength Attachment Although research has examined attachment in interpersonal contexts, research in marketing suggests that consumers can also develop attachments to marketplace entities, including product brands (Fournier 1998; Keller 2003; Schouten and McAlexander 1995), celebrities (Thomson 2006), and special possessions (Ball and Tasaki 1992; Kleine and Baker 2004).Notably, despite 5 the growing popularity of the attachment construct, the conceptual properties of this construct remain elusive. Conceptual properties. Brand attachment is defined as the strength of the bond connecting the brand with the self. Consistent with attachment theory (Mikulincer and Shaver 2007), this bond is exemplified by a rich and accessible memory network (or mental representation) involving thoughts and feelings about the brand and the brand’s relationship to the self. Two critical factors reflect the conceptual properties of brand attachment: brand-self connection and brand prominence.Brand-self connection.
First, the idea that attachment involves a bond (with the brand included as part of the self) suggests that a critical aspect of attachment involves the cognitive and emotional connection between the individual and the self, defined here and elsewhere as brand-self connection (Chaplin and Roedder John 2005; Escalas and Bettman 2003; Escalas 2004). By categorizing the brand as part of the self, a consumer develops a sense of oneness with the brand, establishing cognitive links that connect the brand with the self.Though cognitive in its representation, this brand-self linkage is inherently emotional (Mikulincer and Shaver 2007; Thomson et al. 2005), involving myriad and potentially complex feelings about the brand, including sadness and anxiety from brand-self separation, happiness, and comfort from brandself proximity, and pride from brand-self display. Consumers can be connected to a brand because it represents who one is (e. g. , an identity basis) or because it is meaningful in light of goals, personal concerns, or life projects (an instrumentality basis, Mittal 2006).
Brand prominence.In addition to brand-self connection, previous research suggests that the extent to which positive feelings and memories about the attachment object are perceived to be top of mind also serves as an indicator of attachment. According to Mikulincer (1998) and 6 Collins (1996), positive memories about the attachment object (e. g.
, another person) are more prominent for individuals who are highly attached to an attachment object than they are for individuals who show weak attachment. The fact that brand-self connections develop over time and through experience suggests that brand-related thoughts and feelings become part of ne’s memory and vary in the perceived fluency or the ease with which they are brought to mind. We call this component brand prominence: Prominence reflects the salience of the cognitive and affective bond that connects the brand to the self. This salience is reflected by the perceived (1) ease and (2) frequency with which brand-related thoughts and feelings are brought to mind. Thus, consumers’ attachment in relation to two brands with the same degree of brand-self connection is greater for the brand that is perceived to be more prominent. Importance of both indicators.
Brand-self connection is a core component of attachment since it centrally reflects the definition of attachment as the bond connecting the individual with the brand. However, we suggest that the inclusion of brand prominence adds precision in measuring the “strength” of the bond connecting the brand with the individual. This is so for two reasons. First, when thoughts and feelings about the brand are highly accessible, prominence may exert a disproportionately strong influence on decision making (Alba and Marmorstein 1987) and ultimately consumer purchase behavior (Akcura, Gonul, and Petrova 2004).Specifically, consumers for whom brand-self connection is high and for whom associations are also prominent may be more likely to engage in relationship sustaining behaviors than those for whom the brand-self connection is high but prominence is low. This is true because the brand’s prominence makes relationship sustaining activities salient as well. Hence, one might observe greater behavioral commitment in the form of brand loyalty and other behaviors (e.
g. , positive 7 word of mouth, and more time, money, and energy spent on relationship sustaining behaviors) when both brand-self connection and prominence are high.This logic is also consistent with prior work (Akcura, Gonul, and Petrova 2004; Alba and Chattopadhyay 1986; Alba and Marmorstein 1987), which suggests that the prominence of feelings and thoughts can impact behavior by inhibiting recall of other thoughts and feelings (Alba and Chattopadhyay 1986). As relationship sustaining activities become more prominent, relationship inhibiting behaviors may be inhibited.
Second, we noted earlier that consumers can develop a strong brand-self connection because (a) the brand is part of one’s self-conception and/or (b) it has instrumental value.The former emphasizes brand-self connections in terms of who one is and one’s identity. One might expect that when brands are identity based (i.
e. , one’s iMac is seen as part of who one is and what one stands for) prominence is generally high since self-activation and brand activation cooccur. If so, adding prominence may add little to the assessment of attachment since prominence and brand-self connection covary. However, prominence may serve as an important indicator of attachment when consumers are connected to a brand given its instrumental value (i. e.
one’s iMac is important to fulfilling entertainment and work related goals). That is, when a brand has instrumental value attachment should be stronger when brand-related thoughts and feelings are more vs. less prominent. As prominence increases, brand-related thoughts and feelings are part of everyday life tasks, making brand attachment stronger. Brand attachment emotions. Emotions are often evoked when attachment is strong, as emotions are inherent to brand-self connection and prominence factors.
Indeed, the emotional nature of attachment has led to a measure of attachment based purely on emotions.Thomson et al. ’s (2005) 3-factor model characterizes brand attachment in terms of three emotional 8 components: (a) affection (characterized by the emotion items “affectionate,” “loved,” “friendly,” and “peaceful”), (b) passion (characterized by the items “passionate,” “delighted,” and “captivated”), and (c) connection (characterized by the items “connected,” “bonded,” and “attached”). Although we agree that attachments are emotional, in contrast to Thomson et al. (2005), the set of specific emotions underlying attachment is not central to our conceptualization (or measure) of attachment.Indeed, we are agnostic to the specific set of positive feelings linked to the brand. Feelings linked to brand-self connection and brand prominence could be numerous in type and different feelings may be idiosyncratically linked to specific person-brand autobiographical meanings and their prominence. Such feelings could include those noted by Thomson et al.
(2005). However, they could also include joy, excitement, pride, contentment, relief, nostalgia, or any other feelings retrieved from brand-self memories. Second, while passion may indeed characterize strong brand attachment as Thomson et al. 2005) suggest, the degree of passion linked to strong attachment may depend on the relationship’s evolutionary status. Research indicates that passion may wane as relationships progress (Ahuvia, Batra, and Bagozzi 2009). At the same time, relationship progression brings with it more brand-self experiences that should deepen the brand-self bond and enhance its salience.
Thus, while time may be associated with waning passion, it may also be associated with enhanced attachment. Representing attachment based on passion may not fully capture all relationships characterized by strong attachment.Third, attachment is more than emotions; it is reflected by mental representations (richcognitive schemata) that include brand-self cognitions, thoughts, and autobiographical brand memories (Berman and Sperling 1994; Mikulincer and Shaver 2007) that may not be captured by 9 measures of emotions. As such, we do not include emotions as factors that indicate brand attachment. Instead, we reason that our two factor model of attachment (brand-self connection and brand prominence) captures the emotions that accompany attachment.
1 Brand attachment and brand relationship quality.The brand attachment concept also shares some conceptual resemblance to Fournier’s (1998) seminal concept of brand relationship quality (BRQ). Both concepts propose similar outcomes (e. g.
, accommodations, devaluation of alternatives). Furthermore, Fournier’s (1998) concept includes brand-self connection as one of the six indicators of BRQ, assessing the “quality, depth, and strength” of a consumer’s relationship with a brand (Fournier 1998, p. 363). However, our measure is designed to reflect only the strength dimension as it pertains to brand-self connection.Moreover, BRQ is designed to accommodate a host of relationships types (e. g.
, best friends, kinships, dependencies, and enslavements) and thus accommodates relationships types that can be positive, neutral, or negative. However, brand attachment specifies neither relationship type, nor does it accommodate negative relationships. Differentiating Brand Attachment from Brand Attitude Strength Brand attachment and brand attitude strength share several similarities. Both are psychological constructs that reference a brand. Both involve assessments of “strength.
” (i. e. , of the bond or the attitude).Both assume that high levels of their respective constructs are based on substantial processing regarding the brand. Both have implications for marketing-relevant consumption behaviors, such as brand purchase, repeat purchase, and willingness to recommend a brand. Moreover, we surmise that when consumers are strongly attached to a brand, they can also have 10 a positive and strong attitude toward it.
However, we regard brand attachment and brand attitude strength as distinct constructs because they differ in several fundamental respects. First, the constructs differ in the nature of affect they implicate.Whereas attachment implicates hot affect from the brand’s linkage to the self (Mikulincer and Shaver 2007), strong brand attitudes reflect evaluations and cold affect (Cohen and Areni 1991) involving a judgment about the brand. This difference in affect has important implications for brand behaviors as discussed later. In this sense, the constructs differ in their motivational power, with the emotional and self-implications underlying attachment serving as a more powerful driver of behavior. Second, although both constructs involve assessments of strength, the entity to which “strength” applies differs.
With attachment, what is strong is the bond that connects the brand with the self. Bonds are stronger (a) as connections between the brand and self become closer and (b) as brand-related thoughts and memories are more prominent. With strong attitudes, what is strong is one’s judgment of the goodness or badness of the brand. Thus, with attachment, strength references the brand self-relationship.
Such strength is indicated by the connection between the self and the brand and a subjective sense of brand prominence. With strong attitudes, strength references the attitude object and the confidence with which it is held.Such strength is often indicated by objective indicators of attitude accessibility. Moreover, the factors that lead to variation in strength vary. With strong brand attitudes, strength varies not as a function of brandself connections or the prominence of brand thoughts, but rather as a function of the confidence with which the judgment is rendered (Petty, Brinol, and DeMarree 2007). Third, the constructs differ in their range of valence. Strong attitudes can range from positive to negative, such that attitude strength is conceptualized on a bi-polar valence dimension.Attitudes thus range from strong-positive to weak-positive to weak-negative to strong-negative.
11 Thus, positive and negative ends anchor the attitude strength continuum and behavior is linked with either end of that continuum. In other words, just as strong positive attitudes predict behavior (e. g. , purchase) strong negative attitudes also predict behavior (e. g. , purchase avoidance). In contrast, attachments are always positive.
The opposite of a strong attachment is a weak attachment. What varies is not the valence of the attachment but rather the strength of the bond connecting the brand with the self and its prominence.Finally, whereas attachment is largely time-dependent, brand attitude strength need not be. Specifically, attachment includes relationship based working models (mental representations) that reflect prominent autobiographical and episodic memories concerning oneself and the attachment object.
Such models also include procedural knowledge about how the brand can regulate one’s emotions (Collins and Read 1994; Mikulincer and Shaver 2007). Such self-brand links develop over time (Mikulincer and Shaver 2003). In contrast, strong brand attitudes need not be time dependent.They are based on thoughtful processing (elaboration) and can be formed in a limited time as long as the information on which they are based is both persuasive and the result of elaboration. Because attachments develop over time while strong brand attitudes need not, attachment may reflect a more advanced stage of relationship development. Predicting the Differential Impact of Brand Attachment and Brand Attitude Strength Having conceptually distinguished brand attachment from brand attitude strength, we turn now to understanding whether they predict different outcomes.This is a novel issue as prior research has not yet distinguished the differential effects each predicts.
Thomson et al. (2005) have demonstrated that their measure of emotional attachment and attitude valence have distinct 12 effects, with attachment better predicting brand loyalty and willingness to pay a price premium. However, their research did not study attitude strength. Since attitude strength is more closely tied to actual purchase behavior than is attitude valence (Fazio 1995; Petty, Haugtvedt, and Smith 1995; Priester et al. 004), a more convincing case for the power of attachment would be made if its impact were different than that of attitude strength. Increasing research shows that attitude strength predicts purchase behavior, with the direction of the behavior (being inclined or disinclined toward purchase) varying as a function of whether attitude valence is strongly positive or strongly negative (Fazio 1995; Petty, Haugtvedt, and Smith 1995).
Most often studied are relatively simple behaviors, such as purchase intentions or product choice (Fazio, Powell, and Williams 1989; Petty, Haugtvedt, and Smith 1995).We add to the literature by suggesting that within a given consumption context, behaviors can be conceptualized along a hierarchy that reflects their enactment difficulty. Difficulty is conceptualized as the extent to which the behaviors expend economic, social, psychological, time, or physical resources. We use an extended version of self-expansion theory to develop these ideas. As described below, our theorizing predicts novel behavioral outcomes that have not been linked with brand attachment or strong brand attitudes, specifically, intentions and actual behaviors, including actual purchase, brand purchase share, and need share.
Self-expansion theory provides a basis for these predictions. Self-Expansion Theory and Behaviors Self-expansion theory (Aron et al. 2005) posits that individuals possess an inherent motivation for self-expansion, a desire to incorporate others (here brands) into one’s conception of “self.
” The more an entity (brand) is included in the self, the closer is the bond that connects them. 13 Attachment develops over time as relationships between the self and the entity evolve. Through time, a cognitive reorganization takes place such that the self expands to include the entity.Individuals develop a positive feeling of one-ness with the entity (Aron et al. 1992) and tend to view the entity’s resources as their own (Mittal 2006). We add to self expansion theory by proposing that consumers who are attached to brands are not just recipients of the brand’s resources (i. e. , consumers come to regard the brand’s resources as their own); they also actively invest their own resources in the brand so as to maintain their brand relationship.
Thus, consumers who are highly attached to a brand are more motivated to expend resources of their own in the process of self-expansion.Such resources include the allocation of (a) social resources, like defending the brand to others and derogating alternatives (e. g. , Johnson and Rusbult 1989), (b) financial resources, as evidenced by a willingness to pay a higher price for the brand (Thomson et al. 2005) or the willingness to devote a greater share of one’s expenditures to the brand (as opposed to brands in the same or related product categories), and (c) time resources, as illustrated by involvement in brand communities and brand promotion through social media (Muniz and O’Guinn 2001; Schouten and McAlexander 1995).Hence, the more attached one is to the brand, the more likely the individual is to move from an egocentric to a more reciprocal brand relationship involving sharing one’s resources with the brand. As such, consumers who are highly attached to a brand should treat the brand preferentially and engage in restorative behaviors that ensure brand relationship continuation (Aron et al 1992; Aron et al. 2005; Mikulincer 1998).
Impact on Intentions to Perform Difficult Behaviors 14 We expect that consumers’ intent to enact difficult behaviors (those that use more of their own resources) is greater when attachment is strong (vs. eak). This is so since a feeling of oneness is accompanied with hot affect, which is highly motivational (Mikulincer and Shaver 2007). Because attached consumers see brands as part of themselves and have salient thoughts and feelings about the brand, they should be more willing to utilize greater resources of their own; resources that require the enactment of difficult behaviors so as to maintain that relationship.
The greater the attachment, the more difficult the behavior the consumer is willing to enact in order to maintain the brand relationship.Attitude strength should be less able to predict these relationship maintaining, sustaining and restoring behaviors because the brand is not connected to the self and hence is less strongly linked to resource allocation for the purposes of sustaining a brand relationship. This novel extension of self-expansion theory leads us to predict: H1: Brand attachment is a better indicator of a consumer’s intentions to perform difficult behaviors than is brand attitude strength. Impact on Actual Purchase Behavior Whereas H1 examines intentions, a long history of research shows that intentions and actions do not always correspond.
This is because situational (e. g. , a brand unavailability), normative (e. g. , social constraints), behavioral (e. g. , habits), and financial (price increase) constraints may preclude intentions from being actualized into behaviors (Fishbein and Ajzen 1975; Shepphard, Hartwick, and Warshaw 1988).
When the behavior itself involves significant resources, enactment difficulty will be even higher. We posit that brand attachment will act as a stronger predictor of actual difficult-to-enact behaviors compared to strong brand attitudes. Strongly 15 ttached consumers incorporate a brand as part of their self and hold salient thoughts and feelings about it. On the basis of perceived oneness with a brand, consumers should be more motivated to enact relationship sustaining behaviors that are difficult to perform than consumers with strong brand attitudes. Therefore, we predict: H2: Brand attachment is a better indicator of a consumer’s actual purchase behavior than is brand attitude strength. Brand Purchase Share We also expect that brand attachment better predicts a brand’s purchase share, defined as the share of a brand among directly competing brands (e.
. , if the number of competing brands a consumer purchases in a given product category is 0, the focal brand’s purchase share is 100%). When consumers are strongly attached to a brand, competing brands will be less prominent and linked less strongly to the self. As a result, competing brands will be less likely to be regarded as substitutes. Work in attachment theory and psychology indicates that individuals perceive attached objects as irreplaceable; other objects will not serve as substitutes (Bowlby 1980). Thus, a consumer who is strongly attached to a brand of running shoes (e.
g. Nike) is less likely to use competing brands. In contrast, brand attitude strength does not necessarily have the same implications for brand purchase share.
One can have a strong positive attitude toward one brand while also having a similar strong positive attitude toward another brand. Thus, we anticipate: H3: Brand attachment is a better indicator of brand purchase share (the share of a brand among directly competing brands) than is brand attitude strength. 16 Need Share We also expect that brand attachment better predicts the brand’s share of use among substitutable alternatives, specifically, need share.For example, a consumer who is strongly attached to a brand of soft drinks is not only less likely to buy competing soft drinks, but also less likely to buy other beverages (e. g. , tea, coffee, water, juice). Likewise, a consumer who is attached to her iPhone may not only be more likely to allocate more of her monetary resources to the iPhone (vs.
competing cell phone alternatives), but also more likely to use her iPhone as a source of information and entertainment compared to competing need categories (e. g. , newspapers, TV, magazines).
The brands to which consumers are highly attached capture consumers’ mind and heart. Therefore, attached consumers would be less likely to rely on alternatives, even in other categories that fill the same need. Thus, we predict: H4: Brand attachment is a better indicator of brand need share (the relative use of a brand compared to substitutable alternatives) than is brand attitude strength.
Study 1: Measuring Brand Attachment We first developed a scale designed to map the conceptual properties of the brand attachment construct noted above.We generated a set of items designed to tap brand-self connections and the prominence of brand thoughts and feelings. Ten indicators of attachment were generated, five each for the brand-self connection and prominence components. All items were evaluated on 11point scales anchored by 0 (= “not at all”) and 10 (= “completely”). 17 We analyzed consumers’ responses to the 10-item scale using three very different brands (Quaker Oats Oatmeal, the Apple iPod, and a local university).
Through exploratory factor analyses using oblique factor rotation, we reduced the 10-item scale to 8 items.The full list of items comprising this scale and analyses pertinent to this pretest is presented in Table 1. Although the resulting 8-item scale (5 items reflecting brand-self connection and 3 items representing brand prominence) is not unusually long for academic use, we sought to develop a more parsimonious scale that would lend itself to marketing practice. We therefore selected items that best map the conceptual definition of the two attachment components based on statistical grounds (strong factor loadings and reliability tests).
For brand-self connection these items are: (1) “To what extent do you feel that you are personally connected to (Brand Name)? ” (abbreviated as “Connected”), and (2) “To what extent is (Brand Name) part of you and who you are? ” (abbreviated as “Part of who you are”). These items represent the identity and instrumentality bases of brand-self connection, respectively that we described earlier. For brand prominence, the items are: (1) “To what extent are your thoughts and feelings towards (brand name) often automatic, coming to mind seemingly on their own? (abbreviated as “Automatic”), and (2) “To what extent do your thoughts and feelings towards (brand name) come to you naturally and instantly? ” (abbreviated as “Naturally”).
These two items, which reflect (a) myriad brand-relevant thoughts and feelings and (b) the lack of control over their activation, capture the retrieval ease and frequency characteristic of brand prominence. Note that the “Automatic” item explicitly captures the frequency (“often”) of brand-related thoughts and feelings. These two items were also phrased to describe multiple (vs. single) thoughts and feelings, thereby capturing the frequency aspect of brand prominence. The brand prominence measure does not distinguish the number and valence of thoughts and feelings about a brand 18 since the number and distribution of positive vs. negative feelings linked to the brand are reflected in the degree of brand-self connection.
It should be noted that reducing the number of indicators (8 eight to 4) provides a more parsimonious scale without significant loss of reliability. Reduction also provides a more conservative test of our hypotheses. 2 Insert Table 1 hereStudy 2: Attachment and Attitude Strength as Distinct Constructs Study 2 was designed to (a) test the assumption that both brand-self connection and brand prominence are indicators of brand attachment and (b) demonstrate that attachment and attitude strengths are empirically discriminable.
Method Participants and design. One hundred eight undergraduate marketing students completed a booklet in exchange for partial course credit. The booklet asked respondents to report their thoughts and feelings towards the Apple iPod. The survey was administered in a group setting (30-40 per group). Measures.All responses were provided on 11-point scales anchored by 0 (not at all) and 10 (completely). The 4-item brand attachment scale was used (r = . 91 for indicators of brand-self connection; r = .
71 for indicators of brand prominence). Two items measured separation 19 distress: “To what extent would you be distressed if the iPod were discontinued? ” and “To what extent is it difficult to imagine life without the iPod? ” (r = 82). Brand attitude strength was assessed by attitude valence and a set of items regarded as corresponding with strength. Respondents indicated the extent to which they viewed the Apple iPod as good (+5) vs.
ad (-5); positive (+5) vs. negative (-5); and the extent to which they liked it (+ 5) vs. disliked it (-5), with ? = .
73. These items indicate attitude valence. 3 Respondents used five additional items to rate the extent to which the Apple iPod is (a) important to them and (b) self-relevant, as well as (c) the extent to which they have thought about the brand, (d) are confident with their brand evaluation, and (e) are certain regarding their brand evaluation (? = . 73). 4 These items are commonly used with attitude valence to indicate of attitude strength (Krosnick et al. 993).
Brand attitude strength was assessed as a single-order factor reflecting the multiplicative product of attitude valence weighted by the confidence/certainty with which this attitude is held. Our results are unchanged by including self-relevance and importance of the attitude. All analyses below use only attitude certainty and confidence as indicators because the joint product of attitude valence weighted by confidence and certainty is most consistent with the conceptual definition of brand attitude strength (Brinol and Petty 2009).Notably, the results below are similar when we assessed brand attitude strength (a) as the average of the items indicating attitude valence and confidence/certainty or (b) as a two-factor model with valence and confidence/certainty as separate second-order factors. Test of Assumptions 20 Importance of the two-factor brand attachment model. We conducted two confirmatory factor analyses of the items representing attachment; one in which both the brand-self connection and prominence factors were allowed to correlate (r = . 37; ? 2 (3) = 18. 37), and a second in which they were forced to be perfectly correlated (? (4) = 223.
11). The change in ? 2 (?? 2 (1) = 204. 74, p < . 001) reveals that the first analysis better fits the data, confirming the two factor attachment model.
Evidence for discriminant validity. To verify that brand attachment and brand attitude strength are discriminable constructs, we conducted two confirmatory factor analyses; one in which both constructs were allowed to correlate (r = . 66, ? 2(3) = 3. 80), and a second in which the two factors were forced to be perfectly correlated (? 2(4) = 27. 70).
The difference between the two models was significant (?? (1) = 23. 90, p < . 001), suggesting that brand attachment and brand attitude strength are related but empirically distinct.
These results also demonstrate that the brand attachment scale meets the criteria of convergent and discriminant validity as it is related to but distinct from brand attitude strength. Had attachment been simply a stronger form of attitude strength, we are unlikely to have observed distinct factors. Evidence for convergent validity. Separation distress is regarded as an emotional indicator of attachment (Bowlby 1980; Thompson et al. 005). That is, the more attached one is to an entity, the more distress one feels at the prospect of losing one’s relationship with that entity. Separation from (loss of proximity to) an attachment object creates emotional distress, inducing negative feelings like depression, anxiety, and a loss of self. Adults reveal distress from attachment figure loss (e.
g. , Berman and Sperling 1994; Simpson 1990), whereas in a consumer context, loss of possessions is also mourned. Indeed, Bowlby (1980) regards separation distress as a natural concomitant of attachment.As Hazan and Zeifman (1999) note, “separation 21 distress… is the data from which the existence and regulator role of the attachment behavioral system is inferred” (p. 351). The existence of separation distress is thus regarded as evidence for the existence of attachment. The 1985 “New Coke” fiasco and the death of Michael Jackson illustrate two such examples of the relationship between brand attachment and separation distress.
The validity of the brand attachment scale would therefore be evidenced by showing a relationship between attachment (and its components) and separation distress.A regression analysis was thus performed to examine whether brand-self connection and prominence as indicators of attachment both independently predict this known indicator of attachment (separation distress). The results reveal two main effects (namely, brand-self connection; ? = . 50, F(1, 104) = 15. 6, p . 05). The difference between the two coefficients was significant (z = 3. 87, p < .
01). 26 Brand attachment also better predicts moderately difficult behaviors (? = . 52, p < . 01) than does brand attitude strength (? = .
47, p < . 01). Again, the difference between the ? coefficients is significant (z = 2. 1, p < . 01). Combined, these results strongly support H1. Although not hypothesized, we found that brand attitude strength (? = .
46, p < . 01) was as strong a predictor as brand attachment (? = . 45, p < . 01) of brand behaviors regarded as least difficult to perform (z = . 07, p > . 05). We discuss this finding further in the General Discussion section.
In addition, we compared the predictive power of two different attachment models, (1) one with both brand-self connection and brand prominence, as indicators of attachment, and (2) another with only the brand-self connection component.This analysis explores whether brand prominence as an indicator of brand attachment is necessary for the prediction of behavioral intentions. The results show no significant difference in the one-component (self-connection) vs. two-component (self-connection and prominence) brand attachment measure’s predictive ability of most difficult (? ’s = . 84 and . 82, p < . 001, respectively), moderately difficult (? ’s = .
87 and . 82, p < . 001, respectively), and least difficult to perform behaviors (? ’s = . 82 and . 77, p < . 001, respectively).
The difference in coefficients and model fit (?? (7) = 10. 62; ?? 2(5) = 5. 31; ?? 2(11) = 6. 90; p > . 05, respectively) were not significant. This result is interesting and is explored further in Study 4 and the Discussion section. Discussion Study 3 replicates the measurement effects observed in Study 1. In addition, and consistent with H1, the study supports the idea that brand attachment better predicts consumers’ intentions to enact difficult behaviors than does brand attitude strength.
These results further confirm the idea that brand attachment and brand attitude strength are different constructs that have different 7 behavioral outcomes. Finally, we find that the two- and one-component brand attachment measures predict intent to perform behaviors equally well. Because Study 3 relies on intentions to perform behaviors, versus performance of actual behavior, Study 4 extends Study 3 by examining the importance of brand-self connection and prominence as indicators of attachment. Study 4 also tests the remaining hypotheses using actual purchase data. Study 4 Method Study 4 involved the collaboration with a large European retail bank listed on the stock exchange.
The bank serves over 15 million customers across 17 international markets. Because the investments, and in some cases life savings, of individuals are involved, this context is ideal for testing the ability of both constructs to predict actual behavior. We obtained contact details for 2,000 customers who were randomly selected from one of the firm’s branch networks. Prior to our quantitative study, we conducted 41 telephone interviews with randomly selected customers.
Our aim was to discuss the meaning of the items comprising our measures with customers and reduce item ambiguity.A subsequent pretest of the questionnaire with 52 randomly selected customers explicitly asked participants to point out any ambiguity in responding to individual questions. Such pretesting was deemed necessary since the original scale was developed in the United States and tested among US consumers. Use of the scale in Europe necessitated minor wording changes to ensure question clarity. A finalized version of the questionnaire was then mailed with an informative cover letter, prepaid return envelopes, and a thank you note with small book of commemorative stamps worth 8 $3 as an incentive.
The cover letter explained the purpose of the study, assured that individual responses would not be shared with anyone outside the research team, and thanked the participant. Three weeks after the first mailing, we followed up with the same questionnaire and a small thank you card for participation. Our sampling effort generated 701 responses.
Four questionnaires were dropped due to insufficient questionnaire completion. The final set of 697 usable responses reflected a 34. 85% effective response rate.We compared early and late responses, following Armstrong and Overton’s (1977) recommended procedure. No indication of response bias was found.
Moreover, a check with a manager from the collaborating firm revealed that the demographic profile of respondents was representative of the firms’ customer base. Measures. Measures of brand attachment, brand attitude strength, and separation distress were identical to those used in the prior studies. Dependent measures. Actual purchase was operationalized by collecting actual purchase data about the individual customer from the collaborating firm.
Actual purchase behavior was measured by summing the sales of all investments for each individual customer over the most recent 6-month period after the survey. Brand purchase share was operationalized by asking respondents to indicate how many banks (in addition to the collaborating firm) they are currently using for their various financial services (e. g. , checking account, savings, investments, loans, etc.
). When respondents indicated they exclusively use the collaborating firm brand, purchase share was indicated as 100%.Need share was measured by asking respondents to indicate the extent (out of 100%) to which they use the collaborating firm for all their financial services. Questions and directions included, “To what extent out of 100% do you use (company name) for all your financial services (savings, investments, loans, etc. )? If you use only (company name) for your financial services, you indicate 100%. If you use more than one financial service 29 institution, including non-banking institutions such as investment firms, insurance companies, etc. , indicate the percentage of your use of (company name) among all the financial service institutions.
Control variables. We also collected data on several control variables. To predict changes in actual behavior that are not driven by consumer inertia, we included past purchase, which was collected similarly to our customer purchase behavior measure and accounted for the six months prior to our survey, as a predictor. We also accounted for gender and relationship length. Prior research indicates the role of relationship length as a likely proxy for customer inertia (Colgate and Lang 2001) and the effect of gender on brand choice (Meyers-Levy and Sternthal 1991).Insert Figure 3 here Results The results replicated the previous studies which tested the assumptions underlying the (a) conceptual properties of brand attachment, (b) the two-factor model and the second-order model of brand attachment, (c) the measure’s discriminant and convergent validity in relation to brand attitude strength,7 and (d) the stronger relationship between attachment and separation distress in relation to attitude strength.
These results are not reported given space constraints. Test of H2: Actual customer purchase behavior.To test whether brand attachment better predicts actual customer purchase behavior (sales as reported by the company) than does brand attitude strength, we used a model in which attachment was represented by prominence and brand-self connection as second-order factors. The results revealed that brand attachment 30 significantly predicts actual behavior (? = .
14, p < . 01), even after accounting for past behavior (? = . 60, p < . 001), relationship length (? = . 03, p > . 05), and gender (? = . 02, p > .
05). Brand attitude strength did not predict actual customer behavior (? . 05, p > . 05) (Figure 3). The difference between the coefficient for attachment and brand attitude strength was significant (z = 2. 44, p < .
01). The result that brand attachment is a stronger predictor of actual customer behavior than is brand attitude strength is in line with our prediction and conceptualization of brand attachment and its predicted effects. A set of analyses that compared only brand-self connection vs. brand-self-connection and prominence as indicators of attachment showed that the latter model demonstrated better prediction (? .
11 vs. ? = . 18, p’s < . 001, respectively; ?? 2(8) = 49.
83, p < . 001). These results suggest that attachment (as represented by both brand-self connection and brand prominence subscales) successfully predict actual purchase, confirming that both are useful predictors of actual behavior. Test of H3: Brand purchase share. We tested whether brand attachment significantly predicts brand purchase share with a model in which attachment was again represented by brand prominence and brand-self connection as second-order factors.
The results showed that brand attachment is a stronger predictor of brand purchase share (? = . 65, p < . 001) than is brand attitude strength (? = . 21, p < . 001), even after accounting for past behavior (? = . 10, p < . 05), relationship length (? = . 05, p > .
05), and gender (? = . 03, p > . 05; Figure 3). The difference between the effects of brand attachment and brand attitude strength was significant (z = 8.
86, p < . 001). These results support H3.Furthermore, we compared the results of two models, one in which attachment was represented by our two factor measure and the other in which attachment was represented by brand-self connection only. As expected, the two-component brand attachment measure predicted brand purchase share significantly better than did brand-self 31 connection alone (? = .
64 vs. ? = . 20, p’s < .
001, respectively; ?? 2(8) = 56. 40, p < . 001). These results further underscore the importance of including prominence (with brand-self connection) as a predictor of brand purchase share.Test of H4: Brand need share. We tested H4 by asking respondents to indicate the extent (out of 100%) to which they use the collaborating firm for all their finance services, even services that are performed by other institutions (e. g.
, investment firms, insurance companies). Structural equation models show that attachment represented by both prominence and brand-self connection as second-order factors was a significantly stronger predictor of brand need share (? = . 83, p < . 001) than was strong brand attitudes (? = .
26, p < . 01; z = 11. 86, p < . 001) (see Figure 3).
Past behavior (? = . 02, p > . 05), relationship length (? = . 03, p > .
05), and gender (? = . 02, p > . 05) had no significant effects. These results support H4. We also compared a model that represented attachment by only brand-self connection with a model in which both brand-self connection and prominence were indicators of attachment. Once again, we observed that the two factor brand attachment model more strongly predicted brand need share than did brand-self connection alone (? . 82 vs. ? = .
29, p’s < . 001, respectively; ?? 2(8) = 49. 20, p < . 001). Discussion The use of real purchase data in Study 4 supports H2-H4 and helps us generalize findings to a context involving actual customer behavior. Study 4 demonstrated that brand-self connection and brand prominence both indicate attachment and that the inclusion of both strongly predicted company reported sales, brand purchase share and need share compared to brand-self connection alone.They also outperformed brand attitude strength as predictors of these outcomes.
These effects were observed even when accounting for alternative factors. Taken together, the results of 32 Study 4 corroborate the important role of brand attachment and strongly support the notion that brand attachment and brand attitude strength are different constructs that have different outcomes related to behavior, brand purchase-share, and need share. General Discussion Summary and ImplicationsThe objective of this paper was to address the critical yet unexplored question of whether brand attachment adds value as a construct of interest to marketing and consumer researchers compared to that of brand attitude strength. Building on prior work on attachment and the self-expansion theory our paper makes 3 significant contributions.
The first is a conceptual contribution that articulates the properties of brand attachment and distinguishes this construct from brand attitude strength. The second is a measurement contribution.We developed a managerially viable scale that taps the indicators of brand attachment. We also demonstrated that (a) both the brand-self connection and prominence dimensions are critical and non-redundant indicators of attachment, (b) the scale is strongly related to a known emotional indicator of attachment, separation distress, supporting the scale’s convergent validity, and (c) the brand attachment scale (as represented by brand self-connection and prominence) is empirically related but distinct from brand attitude strength, supporting its convergent and discriminant validity.
The third is a significant theoretical and managerial contribution. We hypothesize and find that the more strongly consumers are attached to a brand, the more willing they are to forsake personal resources to maintain an ongoing relationship with that brand. Thus, they are willing to express an intent to engage in difficult behaviors—those that require investments of 33 time, money, energy, and reputation, so as to maintain (or deepen) a brand relationship. In addition to behavioral intent, we also show that attachment represented by both brand-selfconnection and prominence is a significantly etter predictor than brand attitude strength of actual behaviors. Considering brand purchase and need share in particular, managers have much to gain by an effort to build stronger brand attachment.
In addition to the managerial implications noted above, the results of the present research offer other important managerial and future research implications. First, while the brand attitude strength construct may capture a brand’s mind share of a consumer, attachment is uniquely positioned to capture both heart and mind share.Our finding that attachment better predicts actual behavior than brand attitude strength is of significant importance to managers. Brand attitude strength does not fully reflect the extent to which a brand has successfully captured consumers’ heart. When assessing customer-brand relationships, we therefore encourage managers to incorporate brand attachment in their brand evaluation matrices. Linking attachment with responses to brand attitude measures and actual purchase data will afford managers a more detailed picture of how current brand management efforts relate to future sales.Future Research Though our findings are provocative, they raise additional research issues. First, given the uniquely strong effects of brand attachment shown here, additional research is needed on how marketers can enhance brand attachment (by fostering brand-self connection and prominence). In addition to studying brand attachment antecedents, future research might also examine the relative impact of brand attachment and brand attitude strength on several metrics of brand equity. According to various metrics, a brand’s financial value to the firm is typically affected by 34 he brand’s (a) unit price (Pt), (b) unit marketing costs (MCt) and (c) the number of units sold (Q). Close examination of these three components suggests they are directly tied to and reflect the nature and intensity of customers’ attachment to a brand. Thus, the stronger the brand attachment, the higher the unit price that the brand can bear (willingness to pay a price premium; Thomson et al. 2005). Strong attachments also induce a devaluation of competing alternatives (Johnson and Rusbult 1989) and result in greater willingness to stay in the relationship (Drigotas and Rusbult 1992).These intentions and behaviors all influence the stability of the Q component and reduce the costs of customer retention. Finally, strong attachments toward brands impact brand loyalty, willingness to promote a brand, and engender a relative insensitivity to reciprocity by one’s partner (e. g. , active marketing effort by a brand to reinforce or appreciate its customers’ loyalty; Thomson et al. 2005). Such outcomes should both impact the Q component and make the MC component more cost efficient. Attitude strength may be less strongly related to these brand equity metrics.Hence, although prior research suggests that strong brand attitudes can be an important driver of brand equity (a fact which we do not dispute), we suggest that there is added value to examining attachment, as it may predict brand equity drivers (e. g. , price, unit marketing costs and the number of units sold) more strongly than do strong brand attitudes. Additional research is also needed to shed light on several interesting results observed in Studies 3 and 4. One issue is that whereas brand attachment in Study 4 best predicted the results when attachment was indicated by both brand-self connection and prominence.In Study 3, however only brand-self connection was necessary. Two reasons may underlie these results. First, Study 3 predicted intentions whereas Study 4 predicted actual behaviors. Perhaps prominence is unnecessary as an attachment indicator when only intentions (vs. behavior) are assessed. Second, perhaps for most respondents in Study 3 the brand (Nike) is more identity-based than 35 instrumentality-based. We reason that brand prominence contributes more when attachments are based on the brand’s instrumentality in goal achievement.However, future research should examine the boundary conditions under which prominence plays a role as a critical indicator of attachment. A second interesting result comes from Study 3, which showed that brand attitude strength and attachment are equally good as predictors of easy- to- perform behaviors like brand switching. It is possible that when behaviors are easy to perform, either attachment or strong attitudes predicts behavior. It is also possible that brand attitude strength better predicts movement away from a brand (e. g. , switching) than does attachment since strong negative attitudes predict brand rejection.Attachment may better reflect approach (than avoidance) responses. Future research should also compare the brand attachment measure developed here with the pictorial measure of the inclusion of another person in the self developed by Aron, Aron and Smollan (1992). That measure represents closeness in terms of the degree of pictorial overlap between the self (represented as a circle) and another individual (represented as a different circle). The greater the overlap in the circles the closer the individual is to the other. Applied to brands, we speculate that Aron et al. s (2005) measure corresponds closely with the brand-self connection component of attachment, and in particular items that reflect the identity overlap (vs. instrumentality) of the brand. Yet, this issue deserves future work. Research might also examine whether brand attitude strength and brand attachment reflect different stages of a brand relationship, each of which must be managed to strengthen brand equity. Indeed, Study 4 provides evidence of such time dependence with brand attachment (see Endnote 7). Perhaps the first stage of brand attachment entails relationship establishment 6 which develops through brand purchase. This stage is best represented by positive brand attitudes. Such attitudes may become strong when they are based on thoughtful processing. When the brand offers resources in the service of self-expansion, consumers may subsequently develop strong connections between the brand and the self as well as mental models of the brand and the self, from which brand-related thoughts and feelings are easily and frequently accessed. At this second stage, strong brand attitudes develop into brand attachments.It is perhaps at this point that the positive relationship between self-associations with a brand and brand attitudes converge (Gawronski, Bodenhausen, and Becker 2007; Greenwald et al. 2002; Prestwich et al. , 2010; Tietje and Brunel 2005; Zhang and Chan 2009). The act of choosing an object can result in the creation of associations between the self and the chosen object (Gawronski, Bodenhausen, and Becker 2007; Tietje and Brunel 2005). The brand’s prominence and its linkage to the self may incline consumers to invest resources of their own in the service of maintaining a brand relationship.Such resources are revealed by brand loyalty, brand defense, and other behaviors that are difficult to perform and require the use of valued resources. These behaviors are typical outcomes of brand attachment. According to this perspective, attitude strength and brand attachment are not competing constructs. Rather, they may represent different stages of a brandcustomer relationship. Both are critical. Whereas brand attachment may represent a more desirable destination for managers than strong brand attitudes, the development of strong brand attitudes may be a relevant and necessary step.Finally, we should note that our paper emphasizes attachment as a positive bond between the consumer and the self. Other research suggests that consumers can have strong negative dissociations between the brand and the self. Such negative dissociations can results in brand relationships described as enmities (Fournier 1998) and motivations to inflict harm on the brand 37 (as is true with “bad behavior” from sports fans). Research on these strong aversive brand relationships as the counterpart of strong attachment relationships is richly deserving of future research. 38REFERENCES Ahuvia, Aaron, Rajeev Batra, and Richard Bagozzi (2009), Love, Desire and Identity: A Theory of the Love of Things,” in The Handbook of Brand Relationships, Deborah J. MacInnis, C. Whan Park, and Joseph R. Priester, eds. New York: M. E. Sharpe, 342-57. Akcura, Tolga M. , Fusun F. Gonul, and Elina Petrova (2004), “Consumer Learning and Brand Valuation: An Application on Over-the-Counter Drugs,” Marketing Science, 23 (1), 156-69. Alba, Joseph W. and Amitava Chattopadhyay (1986), “Salience Effects in Brand Recall,” Journal of Marketing Research, 23 (November), 363-69. ________ and Howard Marmorstein (1987), “The Effects of Frequency Knowledge on Consumer Decision Making,” Journal of Consumer Research, 14 (June), 14-25. Armstrong, Scott J. and Terry S. Overton (1977), “Estimating Nonresponse Bias in Mail Surveys,” Journal of Marketing Research, 14 (August), 396-402. Aron, Arthur, Elaine N. Aron and Danny Smollan (1992), “Inclusion of Other in the Self-Scale and the Structure of Interpersonal Closeness”, Journal of Personality and Social Psychology, 63 (4) 596-612. _________, Debra Mashek, Tracy McLaughlin-Volpe, Stephen Wright, Gary Lewandowski, and Elaine N.Aron (2005), “Including Close Others in the Cognitive Structure of the Self,” in Interpersonal Cognition, Mark W. Baldwin, ed. New York: Guilford Press, 206-32. Ball, Dwayne A. and Lori H. Tasaki (1992), “The Role and Measurement of Attachment in Consumer Behavior,” Journal of Consumer Psychology, 1 (2), 155-72. Berman, William H. and Michael B. Sperling (1994), “The Structure and Function of Adult Attachment,” in Attachment in Adults: Clinical and Developmental Perspectives, Michael B. Sperling and William H. Berman, eds. New York: Guilford Press, 3-28. 39