Rajeev Batra, Aaron Ahuvia, & Richard P. Bagozzi
Using a grounded theory approach, the authors investigate the nature and consequences of brand love. Arguing that research on brand love needs to be built on an understanding of how consumers actually experience this phenomenon, they conduct two qualitative studies to uncover the different elements (“features”) of the consumer prototype of brand love. Then, they use structural equations modeling on survey data to explore how these elements can be modeled as both first-order and higher-order structural models. A higher-order model yields seven core elements: self–brand integration, passion-driven behaviors, positive emotional connection, long-term relationship, positive overall attitude valence, attitude certainty and confidence (strength), and anticipated separation distress. In addition to these seven core elements of brand love itself, the prototype includes quality beliefs as an antecedent of brand love and brand loyalty, word of mouth, and resistance to negative information as outcomes. Both the firstorder and higher-order brand love models predict loyalty, word of mouth, and resistance better, and provide a greater understanding, than an overall summary measure of brand love. The authors conclude by presenting theoretical and managerial implications. Keywords: brand management, brand attachment, brand loyalty, brand relationships, brand commitment lthough for decades researchers have studied how consumers form “like–dislike” attitudes toward brands, the past few years have seen a burgeoning interest among both practitioners and academics in consumers’ “love” for brands.1 Among practitioners, Roberts’s (2004) book Lovemarks expresses increased interest in this topic, and Bauer, Heinrich, and Albrecht (2009) recently documented a growing use of the concept of love in advertising. Academic research on brand love or related constructs has also been substantial (for reviews, see Albert, Merunka, and Valette-Florence 2008; Thomson, MacInnis, and Park 2005), finding it to be associated with positive word of mouth (WOM) and brand loyalty (Carroll and Ahuvia 2006; Fournier 1998; Thomson, MacInnis, and Park 2005), increased willingness to pay a price premium (Thomson, MacInnis, and Park 2005), and forgiveness of brand failures (Bauer, Heinrich, and Albrecht 2009), among other outcomes.

Brand Love

A

1We conceptualize brands as the totality of perceptions and feelings that consumers have about any item identified by a brand name, including its identity (e.g., its packaging and logos), quality and performance, familiarity, trust, perceptions about the emotions and values the brand symbolizes, and user imagery. The current work involves mainly discrete manufacturer brands and corporate brands. Further research is needed to establish its applicability to other branded objects and possessions.

Rajeev Batra is S.S. Kresge Professor of Marketing, Ross School of Business (e-mail: rajeevba@umich.edu), and Richard P. Bagozzi is Dwight F. Benton Professor of Behavioral Science in Management, Ross School of Business, and Professor of Clinical, Social and Administrative Sciences, College of Pharmacy (e-mail: bagozzi@umich.edu), University of Michigan. Aaron Ahuvia is Professor of Marketing, College of Business, University of Michigan–Dearborn (e-mail: ahuvia@umich.edu). All authors contributed equally. The research assistance of Ryan S. Elder and the helpful comments of Cele Otnes and Zeynep Gurhan-Canli are gratefully acknowledged. This article was accepted under the editorship of Ajay Kohli.
© 2012, American Marketing Association ISSN: 0022-2429 (print), 1547-7185 (electronic)

In consumer research, Shimp and Madden (1988) adapt Sternberg’s (1986) triangular theory of interpersonal love from psychology, and Ahuvia (1993) performs the first major empirical study. Fournier (1998) includes love as one of the core elements of consumers’ relationships with brands, and Ahuvia, Batra, and Bagozzi (2009), Carroll and Ahuvia (2006), and Albert, Merunka, and Valette-Florence (2008) explicitly study brand love. Related work spans self–brand connections (Escalas and Bettman 2003), consumers’ attachments to brands (Park et al. 2010; Thomson, MacInnis, and Park 2005), the construction of self-identity (Belk 1988), consumer–object bonds (Kleine, Kleine, and Allen 1995), and brand communities and reference groups (McAlexander, Schouten, and Koenig 2002). Although this interest suggests that brand love is an important marketing topic, little agreement exists as to what brand love is (see Albert, Merunka, and Valette-Florence 2008). Various definitions of brand love suggest that it has anywhere from 1 (Carroll and Ahuvia 2006) to 11 dimensions (Albert, Merunka, and Valette-Florence 2008), with most studies presenting differing conceptualizations. This disagreement persists in large measure because, as we discuss subsequently, most marketing studies have omitted the exploratory work needed in the early stages of research to establish the boundaries and contents of the key construct (Lincoln and Guba 1985). Instead, prior work has primarily substituted the vast psychological literature on interpersonal love (e.g., Aron and Westbay 1996; Sternberg 1986) and/or attachment (e.g., Bowlby 1979) for foundational exploratory research on brand love. In the psychological literature, definitions of different types of interpersonal love (e.g., romantic, compassionate/ altruistic) abound, many of which mention affection, attachment, intimacy, caring, intense longing, passion, and so on, depending on the specific type of love (Fehr 2006, pp. 226– 28). However, there are compelling reasons these conceptualizations of interpersonal love should not be applied 1
Journal of Marketing Volume 76 (March 2012), 1–16

directly to brand love. We argue that brand love needs to be conceptualized from the ground up, built on a deep understanding of how consumers experience it, and only then should valid connections be made to the interpersonal love literature. Thus, the current research begins with two qualitative studies that provide a grounded and evidence-based foundation for our subsequent studies of brand love. Consistent with research on interpersonal love (see Fehr 2006, 2009), we find that brand love, as consumers experience it, is best represented as a higher-order construct including multiple cognitions, emotions, and behaviors, which consumers organize into a mental prototype. These include, but go beyond, brand attachment (Thomson, MacInnis, and Park 1995) and self–brand connections (Escalas and Bettman 2003). Using survey data, we then develop a valid and parsimonious structural equations model of the brand love prototype that, because of its grounding in the two qualitative studies, uses significantly broader emotional and self-related constructs than prior work (e.g., a sense of natural comfort and fit, a feeling of emotional connectedness and bonding, a deep integration with a consumer’s core values, a heightened level of desire and interaction, a commitment to its long-term use, attitude valence and strength). We show that our multicomponent model of the brand love prototype greatly expands understanding of the consumer experience of brand love. It also explains more of the variation in repeat purchase intention, positive WOM, and resistance to negative information about the brand than a summary measure of brand love. Through this richer understanding of brand love, we gain insight into how brand liking can potentially be changed into brand love, and we draw theoretical and managerial implications. Progress in brand love research has been hindered by a lack of exploratory studies that guide subsequent measurement and theory development. This has led to two major problems: assuming the equivalence of brand love and interpersonal love and the perception of brand love as an emotion rather than a relationship.

specific content. For example, sexual passion is a feature of romantic love but not of parental love; thus, theories about parental love cannot be applied directly to romantic love. Similarly, theories of interpersonal love cannot be applied directly to brand love. Albert, Merunka, and Valette-Florence (2008) provide a similar type of analysis to our Studies 1 and 2. They find that 11 dimensions underlie brand love: passion, a longduration relationship, self-congruity, dreams, memories, pleasure, attraction, uniqueness, beauty, trust (satisfaction), and a willingness to state this love. However, as they note (p. 1073), they fail to find the aspects of attachment and commitment found in most prior studies, and their results could be idiosyncratic given their choice of the three specific images used to depict brand love. Thus, further studies to understand how consumers experience brand love are still needed. The existing literature also does not adequately distinguish between the love emotion and the love relationship. The love emotion is a single, specific feeling, akin to affection (Richins 1997), which, like all emotions, is short term and episodic. In contrast, the love relationship, like the friendship relationship, can last for decades and involves numerous affective, cognitive, and behavioral experiences (Fournier 1998). Extant brand love research sometimes studies the love emotion and sometimes studies the love relationship, but it rarely acknowledges the distinction. The first step in understanding brand love is to uncover the implicit definition of love that consumers are using when they say they love a particular brand or product. Prior research has found that fuzzy and complex concepts such as emotions or love (Fehr 2006; Shaver et al. 1987)—concepts not amenable to definition in terms of necessary and sufficient criteria (Fehr 2006, p. 227)—are best described as prototypes (Rosch 1975). A prototype is a list of attributes (i.e., prototype features) that people associate with a particular kind of thing, in this case love (see Fehr 2006). These attributes are organized into a central or most typical exemplar of that category (Shaver et al. 1987, p. 1062), such as love, or a subcategory, such as romantic love, parental love, brand love, and so on. The more of these prototype features a relationship or an emotion has, and the more central those attributes are to the prototype, the more likely a consumer is to consider it some type of love. Because prototypes are cultural models, researchers have found high levels of similarity in interpersonal love prototypes across gender, sexual orientation, and age. Unlike classical definitions (Fehr and Russell 1991), which are consciously formulated to be precise, prototypebased definitions are fuzzy (Shaver et al. 1987) in two ways. First, prototype definitions are always characterized by fuzzy boundaries, which, in the current context, means that a typical consumer will view some brands as definitely loved, some brands as definitely not loved, and other brands falling into a “sort-of-loved” middle category. Second, proBrand Love as an Emotion Versus a Relationship

The Brand Love Prototype

Limitations of Extant Brand Love Research

Rather than exploring brand love in an open-ended manner with consumers, most extant brand love research begins with a chosen theory of interpersonal love and then creates scale items to apply this theory to a marketing context. This approach presents a potential problem if brand love is not directly analogous to the particular theory of interpersonal love being used. Research suggests that this problem is significant (Aggarwal 2004; Richins 1997). We are not suggesting that, because brand love may be different from interpersonal love, it is not a “real” type of love. There are multiple kinds of interpersonal love (e.g., romantic, parental, compassionate/altruistic), all of which are real (Fehr 2009), yet they vary from one another in their 2 / Journal of Marketing, March 2012

Assuming the Equivalence of Brand Love and Interpersonal Love

totype definitions are fuzzy because their features frequently include not only elements of the phenomenon itself but also antecedents and outcomes. For example, in Shaver et al. (1987, p. 1076), the prototype of the emotion fear includes antecedents (e.g., the threat of harm), attributes of fear itself (e.g., feelings of nervousness, cognitive inability to focus, behavioral acts of hiding or crying), and outcomes (e.g., self-comforting). This creates fuzziness because not all elements of a prototype are necessarily attributes of the core phenomenon itself. When researchers elicit prototype features, it is often necessary to investigate whether some sort of dimensional reduction might be possible. For example, Fehr (1988) identifies 68 features for the love prototype, and Aron and Westbay (1996) use factor analysis to extract three underlying latent factors from them. Similarly, after obtaining similarity scores for 135 prototypicality-rated emotion terms, Shaver et al. (1987) cluster-analyze them hierarchically to determine how they might be split up at different levels of abstraction. Thus, at the maximal degree of abstractness, these emotion terms could be classified simply as being positive or negative; at the other, more subordinate, end of the tree, the analysis yielded 25 clusters: “The cluster-analytic results therefore provide three sets of candidates for basicness: a 2-term list at a high level of abstraction (essentially, positive vs. negative emotions), a 5- or 6-term list, and a 25-term list” (Shaver et al. 1987, p. 1068). In other words, the set of features that constitute a prototype can often be hierarchically organized at different degrees of abstractness. Uncovering mental prototypes presents a challenge because they are tacit knowledge structures and thus are not easily verbalized. To produce a description of a largely tacit mental prototype, it is necessary to get respondents to use the prototype for its natural purpose, observe (or imagine) themselves doing this, and then report their observations to the researcher. Both Studies 1 and 2 follow this data collection strategy by prompting respondents to use their own love prototype to determine whether various brands or other items are clearly loved, on the borderline between loved and not loved, or clearly not loved and then reporting the criteria they used to make these classifications. This is similar to Study 2 of Shaver et. al (1987) in that the authors elicited the features of each type of emotion from respondents by asking them to list what they believed and felt and how they acted when they experienced the emotional state being studied. Our three studies move from being open and exploratory to being more focused and confirmatory. Study 1 is designed to provide the widest possible lens on brand love. Rather than directing consumers to talk specifically about brands, we asked respondents about “things that they love” (excluding only other people). Casting such a wide net enables us to place brand love in context and reduce the risk of overlooking important related phenomena. In Study 2, the interviews are narrower in scope, focusing specifically on loved brands. Finally, in Study 3, we conduct a quantitative survey examining loved brands in a consumer electronics context. Because the conventional reporting of qualitative studies (Studies 1 and 2) requires extensive quotations and discursive analysis, which is not possible within the length constraints

of this journal, we present them here only in summary. We detail the methodology and illustrative quotations in the Web Appendix (http://www.marketingpower.com/jm_ webappendix), and longer versions are available on request. Methodology

Study 1 consisted of 70 structured telephone interviews lasting between 10 and 60 minutes each and 10 follow-up depth interviews lasting two to four hours each. Respondents were between 23 and 45 years of age, highly educated, urban, and approximately equally male and female. Study 1 examined all types of noninterpersonal love including, but not limited to, brand love (e.g., love for consumptionrelated objects, activities such as eating and dancing). In 10 follow-up depth interviews, respondents compared loved and not-loved items, as well as interpersonal and noninterpersonal love. Study 2 focuses specifically on loved brands (e.g., Apple, Victoria’s Secret). It includes 18 detailed interviews, lasting approximately two hours each, with college students. Respondents discussed brands of their own choosing in various categories (e.g., consumer electronics, clothing) or meeting specific criteria (e.g., a brand they have used for a long time). They also discussed interpersonal love, to allow for a comparison with brand love. Because we analyzed the 70 telephone interviews from Study 1 in part by comparing frequencies of responses, we assessed intercoder reliability, and a proportional reduction in loss statistic of .87 confirmed coding reliability (Rust and Cooil 1994). We analyzed the depth interviews in Study 1 and all interviews from Study 2 using a grounded theory approach (Strauss and Corbin 1994, p. 283), combined with methods developed by McCracken (1988). In Study 1, we found that 96% of the respondents claimed to love something other than another person, and 72% percent viewed at least one object or activity as being loved in the strictest, most literal sense of the word. In Study 2, 100% of the respondents claimed to either “love” or “sort-of-love” at least one brand, and 89% put at least one brand into the “love” (as opposed to “sort-of-love” or “not love”) category. Therefore, noninterpersonal love in general, and brand love in particular, were commonly reported experiences among respondents. Our analysis yielded ten major components: high quality, linkages to strongly held values, beliefs that the brand provided intrinsic rather than extrinsic rewards, use of the loved brand to express both current and desired self-identity, positive affect, a sense of rightness and a feeling of passion, an emotional bond, investments of time and money, frequent thought and use, and length of use (for more detail about each component, see the Web Appendix at http://www. marketingpower.com/jm_webappendix). Great quality/qualities. When talking about loved brands, respondents’ comments almost invariably began with a list of the perceptions about the brand’s many attractive qualiBrand Love / 3 Elements of the Brand Love Prototype

Studies 1 and 2

Self-identity. Respondents strongly identified with the things they loved, reflecting the important function of loved brands in expressing existing identities and enacting desired identities (Belk 1988; Escalas and Bettman 2003, 2005). This identity link occurred through both the consumer’s direct relationship with the object and the loved brand’s facilitation of interpersonal relationships (McAlexander, Schouten, and Koenig 2002). These findings match prior research on brand love (Ahuvia 2005) and interpersonal love (Aron and Aron 1996). Because talking about a brand with other people is an important part of identity construc4 / Journal of Marketing, March 2012

Intrinsic rewards. There is a common distinction between performing an act to get something (extrinsic rewards) and doing it because you love it (intrinsic rewards) (Babin, Darden, and Griffin 1994, p. 645). A loved brand provides intrinsic rewards when it creates psychological states such as happiness, which are perceived as being part and parcel of using the product (e.g., Pinkberry frozen yogurt is delicious). Loved brands commonly provide both intrinsic and extrinsic rewards, so providing extrinsic rewards was not a problem per se. However, when brands provided only extrinsic rewards, respondents often felt they did not really love the brand but rather were just using it to get something else that they did love.

Strongly held values and existential meaning. While loved brands were praised for providing a wide variety of benefits, such as comfort, transportation, entertainment, exercise, relaxation, and so on, brands were more likely to be loved when they also connected to something the respondent believed was deeper, such as self-actualization, close interpersonal relationships (Richins 1994), existential meaning, or religious or cultural identities. For example, one respondent loved Canon because her hobby was producing creative photo albums of friends that she gave as gifts; thus, she saw Canon cameras as intimately bound up in these meaningful social relationships. Other loved brands succeeded through their marketing in establishing brand meanings that connected to deeply held values (e.g., Apple represents creativity and self-actualization).

ties, such as its exceptional performance, trustworthiness, good-looking design, and so on. However, while some forms of interpersonal love are said to be unconditional, not a single respondent made this claim for brand love. Instead, loved brands were praised for being the best available (e.g., best in every way, best value for money, best on some important attribute), and simply knowing that a better brand existed was commonly offered as a reason for not loving a particular brand. The only complaint that came up regularly in discussions of loved brands was the high price of some higher-end brands. But even for these brands, consumers felt satisfied, believing that this high price was justified. For lower-priced items, being considered an exceptional value for the money was a commonly mentioned virtue of loved brands. For a few luxury goods that respondents fantasized about owning, an exorbitant price was mentioned as making the product even more special; however, this was not reported when the respondent had actually paid for the product.

tion (Holt 1997), high levels of WOM should also be associated with brand love.

Passionate desire and a sense of natural fit. Respondents talked about a sense of natural fit and harmony between themselves and their loved brands. For some, this sense of “rightness” about the relationship included a strong desire for that brand, reflecting the higher-arousal, hotter aspects of brand love frequently called passion (Belk, Ger, and Askegaard 2003). Albert, Merunka, and Valette-Florence (2008) find that passion is the first dimension in brand love, and Bauer, Heinrich, and Albrecht (2009) argue that passion is the most managerially relevant aspect of brand love. When this sense of natural fit with a brand was combined with passionate desire, it was sometimes expressed as “love at first sight.”

Positive affect. Respondents described their experience with loved brands in positive emotional terms, and this tendency was even more prevalent for loved consumption activities. This affect covered the lower-arousal emotions termed “affection” (Thomson, MacInnis, and Park 2005) and “warm-hearted” feelings (Richins 1997) typical of “companionate love” (Hatfield 1988).

Emotional bonding and anticipated heartbreak. Feeling bonded with, and emotionally connected to, a brand emerged as an important aspect of brand love (e.g., Fournier 1998; Thomson, MacInnis, and Park 2005). In addition to these positive emotions, other researchers have also noted that consumers are likely to feel a strong desire to maintain proximity with their loved objects, even feeling “separation distress” when they anticipate or experience being distanced from them (Hazan and Zeifman 1999; Park et al. 2010; Thomson, MacInnis, and Park 2005). In our interviews, such bonding and attachment was frequently evident in pervasive comments that respondents knew they loved a brand because it was unique and irreplaceable and thus would be missed if lost.

Willingness to invest. Respondents reported investing high levels of time, energy, and money into loved brands. These investments highlight the importance of the brand and integrate it more deeply into the consumer’s identity, thus increasing attachment to the brand. Because respondents anticipated separation distress if they were to lose a loved brand (Hazan and Shaver 1994), they are likely to be price insensitive with respect to the brand (Thomson, MacInnis, and Park 2005). Frequent thought and use. Fournier (1998) argues that for a brand to become a legitimate “relationship partner,” the consumer must engage in frequent, interactive behaviors with it, and Park et al. (2010) use the construct of brand prominence to capture frequent thinking about, and use of, brands to which a consumer is strongly attached. Similarly, we find that having frequent interactions with, or thoughts about, a brand is an important aspect of brand love. Indeed, every respondent in Study 2 considered how much time they spent using or thinking about a brand a key criterion for how much they loved it.

Length of use. Having a long history with a brand was a frequently mentioned feature of brand love. This shared history can give the loved brand an important place in the respondent’s personal identity narrative. Because past behavior is often a good predictor of future behavior (Guadagni and Little 1983), it implies greater loyalty to loved brands (Thomson, MacInnis, and Park 2005). Note that our results are significantly different from those of Albert, Merunka, and Valette-Florence (2008). Whereas Albert, Merunka, and Valette-Florence find that (1) functional quality, (2) brand loyalty, (3) the attachment emotion, and (4) the ability of the brand to make the consumer feel good were associated with lower levels of love, our two studies reveal them—as might be expected—to be positive components. Hierarchical Organization of Brand Love Prototype Features

These findings have implications for the strength of attitudes (Krosnick et al. 1993) toward such loved brands. Because attitudes that rest on more frequent, direct experiences tend to be more strongly held, consumers’ attitudes toward loved brands should therefore come to be strongly held. Many of the typical indicators of attitude strength (e.g., greater attitude extremity, more certainty and importance, greater affective–cognitive consistency, faster response latency, more frequent thinking and talking about the attitude object; Krosnick et al. 1993) are therefore logical elements of brand love.

We noted previously that the set of features that constitute a prototype can often be hierarchically organized at different levels of abstractness (Aron and Westbay 1996; Shaver et al. 1987). For example, Shaver et al. (1987) find that their 135 prototypicality-rated emotion terms can be classified at the most abstract level as simply being positive or negative, slightly less abstractly into 5 or 6 categories (hierarchically subsumable under the positive/negative higher-level categorization), or at an even less abstract (lower) level into 25 categories. The grounded theory process of category creation requires the researcher to actively interpret the raw data and code constructs that emerge from the respondent data as signifiers, parts, properties, or instances of other coded constructs. Some of these relationships—“is an instance of,” “is part of,” and “is a property of”—suggest a hierarchical structure. As coding takes place, the researcher may find that some instances of two coded constructs (e.g., higherorder needs and deeper meanings) can best be subsumed under a single higher-level category (e.g., strongly held values and existential meaning). This process of combining codes into more general constructs was a central part of the analysis for Studies 1 and 2: We hierarchically reduced the original list of more than 75 codes to the 10 major themes presented previously. We began by noting the proliferation of conceptualizations of brand love and argued that this was due to (1) a lack of the exploratory qualitative research typically conducted Discussion and Conclusion

The applicability of interpersonal love theories to brand love. Our respondents often stated that although they genuinely loved some brands, this was a different form of love than interpersonal love. Respondents would sometimes compare brand love to interpersonal love in a way that suggested that the brand love prototype was partially based on their understanding of interpersonal love but also modified to fit a consumer context. Not surprisingly, the most widely noted difference in our data was that brand love was often described as a less important relationship than interpersonal love. However, two other important differences emerged as well. First, while interpersonal love contained a strong element of altruistic concern for the beloved, this was not found in brand love. Consumers were concerned with what the brand could do for them, not what they could do for the brand. Second, in healthy interpersonal relationships, when we love someone, they return our love through their helpful behaviors toward us and by occasionally experiencing the love emotion toward us. In contrast, respondents noted that brands do not experience emotions and therefore could not return a person’s love in that way (though brands were viewed as returning the consumer’s love when the brand benefited the consumer; see also Harding and Humphreys 2011). None of the prior brand love studies based on interpersonal love theories included all the aspects of the brand love prototype uncovered here. Thus, we conclude that the interpersonal love theories that have been used as the basis for past work do not provide a suitable theoretical foundation for brand love research. It is beyond the scope of this study to discuss all the theories of interpersonal love that have been applied to brand love. However, because Sternberg’s (1986, 1997) triangular theory is by far the most frequently adopted framework to explain consumers’ love for brands, some discussion of it is appropriate. Sternberg’s theory holds that various types of love are created through different admixtures of passion, intimacy, and decision/commitment. Passion is the least problematic aspect of Sternberg’s theory because it is plausibly translated into the passionate attraction consumers feel for brands. Intimacy refers to feelings, thoughts, and actions that are connected to the experience of warmth, closeness, and bondedness in loving relationships. Although Sternberg (1997) views intimacy as including ten subcomponents, the integration of the beloved into the self is not among them (nor is this phenomenon included in passion or decision/ commitment). In contrast, the current research finds integration of the loved brand into the consumer’s identity to be a central aspect of brand love (e.g., Ahuvia 2005; Albert, Merunka, and Valette-Florence 2008; Escalas and Bettman 2005; Fournier 1998). Thus, theories of brand love derived from Sternberg’s theory are likely to omit the ways consumers use brands to construct or project their identity. Finally, Sternberg’s decision/commitment component is largely irrelevant to brand love. The decision part of this component refers to a person’s conscious choice to view their relationship as love, with all the normative implicaBrand Love / 5

when developing a new topic area and (2) a failure to differentiate between the love emotion and the love relationship. Next, we discuss these issues in the light of our findings.

Etic themes: Structural hierarchy, positive attitudes, and resistance to negative brand information. Grounded theory involves the generation of etic, theory-driven interpretations. As interviews were coded for the presence or absence of certain constructs, they were also coded for links between the constructs. Some of these links reveal that constructs formed a system of hierarchical categories. We chose the ten themes reported previously because, in our judgment, they provided a useful balance of detail and parsimony. However, it would also have been possible to produce a more detailed analysis by breaking these ten themes down into smaller units or a more parsimonious analysis by combining some of these themes into fewer, even more gen6 / Journal of Marketing, March 2012

The brand love emotion or the brand love relationship? When consumers described their love for a brand to us, they invariably described a broad and long-term consumer– brand relationship, with multiple interrelated cognitive, affective, and behavioral elements, rather than a specific, single, transient love emotion. Indeed, the love emotion itself was rarely mentioned as part of that brand love relationship, whereas other emotions (e.g., happiness when thinking about the brand, anxiety about possibly losing the loved brand) were frequently discussed. Thus, we use the term “brand love” to refer to a consumer–brand relationship that corresponds with the brand love prototype described previously and use the terms “brand love emotion” or “love emotion” to refer to the specific affective state called love.

tions this entails. In contrast, in brand love, consumers rarely consciously choose to define their relationship with a brand as “love”—at least before a researcher asks them about it. Furthermore, the commitment part of Sternberg’s triangular theory does not refer to a sense of bondedness to the brand. (Recall that Sternberg considers this sense an aspect of intimacy.) Rather, Sternberg’s commitment refers to a perceived normative, moral obligation to maintain the relationship even in the face of a much better alternative. Although respondents may have been resistant to negative information about their loved brands, if the poor performance of a loved brand became undeniable, respondents reported that they would not maintain their love for the brand. Thus, as Aggarwal (2004) observes, in brand love, the norms of a commercial marketplace often replaced the norms of interpersonal relationships, implying that Sternberg’s interpersonal decision/commitment component cannot be applied to brand love without changing it into something quite different from what Sternberg intended. In conclusion, we stress that we do not mean to imply that brand love researchers should abstain from (1) citing interpersonal love research as sources of hypotheses or even (2) citing parallels between findings on brand love and interpersonal love as relevant supporting evidence. However, our findings do suggest that when theories of interpersonal love are used to construct measures of brand love, important variables may be omitted from the study and unnecessary ones included. Moreover, this lacuna is unlikely to be detected through an analysis of data collected only on dimensions suggested by the interpersonal love literature.

Recall that in the literature, a prototype is typically conceived of in terms of close antecedents, the phenomenon itself, and proximal consequences (Shaver et al. 1987). Shaver et al. (1987, p. 1078) find that an antecedent to the love emotion is “the judgment that the loved one provides something the person wants, needs or likes.” Equivalently, in our brand love prototype, we consider aspects of the loved brand characterized as great quality/qualities antecedent to brand love, because people are attracted to things that provide them with needed benefits (Murstein 1988) and it is difficult to conceive of brand love in the absence of necessary quality. In our respondent interviews, their positive evaluation of the loved brand’s quality was widely recounted as an antecedent to their coming to love it. The core phenomena of brand love would then include different cognitions (e.g., about self-identity), feelings and sense of connectedness and fit, and behaviors (e.g., frequent interactions, resource investments) that our qualitative studies identified as being part of experienced brand love. Because brands cannot be loved without also being liked and evaluated highly, these core brand love features should also include attitude valence, even though brand love con-

Elements of the Brand Love Prototype: Antecedents, Core Elements, and Consequences

eral, categories. Thus, Studies 1 and 2 reveal that features of the love prototype are grouped hierarchically (see Study 3). A second etic finding is the important role of positive attitudes in brand love. Positive attitudes about the loved brands were too ubiquitous to be viewed as their own theme, in that they played a role in framing consumer discourse about a great many other themes. The connection noted here between love and positive attitudes is a more modest version of claims made in prior research, which have gone so far as to define love as an attitude: “Love is an attitude … involving predispositions to think, feel and behave in certain ways” (Rubin 1970, p. 265, italics in original). Hendrick and Hendrick (2006, p. 150) also describe love as “attitude/belief systems that include an emotional core … and are related to patterns of … behavior.” In addition to positive attitude valence, brand love also displayed many characteristics of attitudinal strength. Attitude strength is composed of multiple dimensions (Krosnick et al. 1993), many of which (e.g., greater attitude extremity and intensity, more certainty and importance, affective–cognitive consistency, more frequent thinking and talking about the attitude object) are clearly evident in the interviews. This is not surprising, because in Studies 1 and 2, loved brands were considered very important, were related to at an intense level (because they connected to deeper meanings and identities and were invested in more), were thought and talked about more, and evoked strong affective responses. The extensive and extreme positivity toward loved brands suggests a third etic finding: resistance to negative brand information. Such resistance is also suggested by the finding that loved brands become integrated into the consumer’s identity and people naturally tend to resist negative information about themselves (Ahearne, Bhattacharya, and Green 2005).

tains many elements that go beyond such valence (e.g., the self-related cognitions and distinct feelings documented previously). Furthermore, as discussed in the “Etic Themes” subsection, the core of brand love also should encompass attitudinal strength, including greater attitude extremity and intensity, more certainty and importance, and more frequent thinking and talking about the attitude object (e.g., Krosnick et al. 1993).2 To understand which aspects of the prototype are best viewed as consequences of brand love, we must consider the purpose of mental categorization. As Barsalou (1991, p. 58, emphasis added) points out,
The purpose of categorization is to identify information in memory that provides useful inferences. Upon accessing a category for an entity, a tremendous amount of knowledge becomes available that is useful in a variety of ways … the origins of the entity,… its probable behavior, its implications for the perceiver’s goals, or actions for interacting with it successfully.

also need to study how these components are organized: whether some of them conceptually and empirically combine into higher-order structures or split up further at lower levels of abstraction. Thus, we conducted Study 3, which builds on our grounded theory study and uses structural equation modeling to address these and other questions. Pretest data collection. We generated items to measure the brand love prototype features and pretested them with undergraduate students (n = 133). The items expressed the extent (“very much” to “not at all”) to which respondents believe the brand possessed the listed characteristic for them personally (e.g., “Lets you present yourself to others as the kind of person you want to be”). We supplemented these items with items that visually depicted possible degrees of overlap between two concentric circles, one representing the brand and the other their personal identity (adapted from Bagozzi and Lee 2002). We also located items from existing scales for the constructs of attitude strength (Krosnick et al. 1993, p. 1150) and brand loyalty (Thomson, MacInnis, and Park 1995). For positive feelings, we combined the items Thomson, MacInnis, and Park (2005, p. 80) use with nine more: contented, relaxed, hopeful, calming, wanting, longing for, sense of desire, fun, and exciting. Methodology

Consistent with Barsalou’s argument that prototypes serve useful purposes, we conceive the brand love prototype as helping consumers arrive at useful outcomes in their relationship with brands. In the interpersonal love domain, previous research has found relationship stability (similar to loyalty) to be a relational outcome typical of relationships consistent with prototypical rather than nonprototypical love (Fehr 2006, p. 238). Analogously, we conceptualize brand love consequences as greater brand repurchase intentions, willingness to pay a higher price, engagement in positive WOM, and resistance to negative information. As Albert, Merunka, and Valette-Florence (2008) note, even if we know (from Studies 1 and 2) what features jointly constitute brand love, we still need to know which components are the most important for brand love to be strong. We

Study 3: Structural Equation Modeling of Brand Love

2Recently, Park et al. (2010) examined the relationship between “brand attachment” (similar to brand love) and their interpretation of attitude strength and claim empirical evidence for discriminant validity between the two. However, their conceptualization and measurement of attitude strength are not the same as ours (which relies heavily on Krosnick et al. 1993). They use most items that constitute our two attitude strength factors (e.g., the frequency with which the brand comes to mind, how quickly these brandrelated thoughts come to mind) to measure a different construct they label “brand prominence.” Just as we keep our two attitude strength factors within our brand love prototype, they too keep their brand prominence factor within their brand attachment construct. Given the lack of common agreement on what constitutes attitude strength (see Krosnick et al. 1993), it is not surprising that different authors and studies use this term in varying ways. More research is needed to arrive at a more definitive understanding of the construct and operationalization of attitude strength (and related constructs). Of note, although Park et al. focus on the different outcomes of brand attachment/prominence and attitude strength, they do not study the antecedents and core elements of these in depth, as we do here.

Preliminary category identification and scale refinement. We analyzed the pretest data using exploratory factor analysis to group the items in the survey data into 16 factors (shown in Table 1). Respondents from Studies 1 and 2 viewed 14 of these factors as core elements of brand love itself, though these 14 factors subdivided some themes from Studies 1 and 2 into smaller units. Respondents from Studies 1 and 2 saw one of these 16 factors (great quality/qualities) as an antecedent of brand love. The other contained managerially relevant consequence items including willingness to engage in positive WOM, favorable repurchase intentions, questioning negative information, and brand loyalty. For brevity, we call this loyalty/WOM/resistance (L/WOM/R). (We also measured willingness to pay a higher price but excluded it because of poor reliability.)

3We eventually did not use 2 of these 59 items in the analysis (one measuring the willingness to spend time shopping for the brand and the other measuring the extent to which attitudes toward it were mixed and conflicted), because their inclusion in the measures of the constructs in the final data lowered rather than raised reliability coefficients.

Final data collection. From this preliminary analysis, we identified 59 items that loaded highest on these factors and also displayed satisfactory scale reliabilities as scalar measures of those factors.3 These were then administered in random order through an online survey to 268 college undergraduate students (approximately 50% male) in a large university. Similar to Thomson, MacInnis, and Park’s (2005; Study 1) and Escalas and Bettman’s (2005) surveys, consumers entered their own choice of a “brand I love“ for the consumer electronics product category. After answering questions about this loved brand, respondents then answered the same questions about another brand from the

Brand Love / 7

High Quality (Antecedent) Self–Brand Integration (Also Includes Attitude Strength 1) Current self-identity

TABLE 1 Brand Love Higher-Order Prototype Model: Latent Constructs, Components, and Constituent Items
Well-made, functional quality, practical. Says something about who you are, others seeing you using it get a sense of who you are, important part of self, degree of image overlap between brand and self, personal identity matches brand identity, important to be one of the people who use this brand, brand is an important part of self-identity, brand is a rewarding part of self-identity. Helps present self to others as the person you want to be, makes you look like what you want to look, makes you feel like how you want to feel. Makes life meaningful, makes life worth living, gives life purpose, is inherently important, is more than an investment in future benefit, experience feelings of desire.

Life meaning and intrinsic rewards Passion-Driven Behaviors Willingness to invest resources

Desired self-identity

Positive Emotional Connection Intuitive fit Emotional attachment Positive affect Long-Term Relationship

Things done in past (involvement)

Passionate desire to use

Attitude Strength Attitude strength 1: frequent thoughts (part of self–brand integration) Attitude strength 2: certainty and confidence

Anticipated Separation Distress Attitude Valence

Loyalty/WOM/Resistance (Consequence)

same product category toward which they felt “mostly neutral about, instead of loved.” Unless otherwise noted, all analyses refer to the data about loved brands. Results

Very often talk to others about it, very often have thoughts about it, frequently find myself thinking about it, frequently find myself thinking about using it, find that it keeps popping into my head, feelings toward it are strong, feel lots of affection toward it. Certainty of feelings/evaluations, how strongly hold feelings/evaluations, how quickly feelings/evaluations come to mind, confidence of feelings/evaluations, intensity of feelings/evaluations. Strength of loyalty, if hear something bad would question it in own mind, would buy again, would say positive things about brand to others.

Feel psychologically comfortable using it, meets needs perfectly, natural fit, what I’ve been looking for, fits tastes perfectly, felt right when first encountered it, now feels right, strength of feeling of liking. Feels like old friend, emotionally connected, feel a bond. Content, relaxed, fun, exciting, calming, helps relax, pleasurable. Will be using for a long time, will be part of life for long time to come, feel sense of long-term commitment. Anxiety, worry, fear, apprehension. Satisfaction, compares well with ideal product, like–dislike, positive–negative, meets expectations, feelings of liking toward it, good–bad, favorable–unfavorable.

Have spent lot of time making it fit my needs; willing to spend lot of money improving and fine-tuning it after buy it; willing to spend lot of time improving and fine-tuning it after buy it; have invested lot of time, energy, or money in it; was willing to spend lot of time shopping to buy it specifically; have used it often in appropriate occasions. Feel myself craving to use it, feel myself desiring it, feel a sense of longing to use it, feeling of wanting toward it, feeling of desire toward it, feeling of longing toward it. Have been involved with it in past, have done a lot of things with it in the past, have interacted a lot with it or the company that makes it.

Structural equations modeling of a first-order structural model of the brand love prototype. First, we estimated a nonhierarchical (first-order) structural equations model for the loved brands data (using LISREL 8.8, n = 187) in which we specified brand love to consist directly of the 14 core underlying factors emerging from our exploratory factor (and previous qualitative) analysis. Fit statistics for this and all other models reported next met or came very close to all 8 / Journal of Marketing, March 2012

the standard criteria (root mean square error of approximation [RMSEA] < .06, nonnormed fit index [NNFI] > .95, comparative fit index [CFI] > .95, standardized root mean square residual [SRMR] < .08; Hu and Bentler 1999). (For brevity, model fit statistics are not reported here but are available from the authors.) All composite indicators of the 14 latent constructs fit well, as hypothesized. Table 1 lists the individual latent first-order components (organized into the higher-order hierarchical structure reported subsequently) and measurement items. Within each latent first-order component, we randomly split the various indicators listed in Table 1 into two groups and then used composite indicators of each group, consisting of aver-

4Thus, in cases in which the latent higher-order construct consisted of more than one first-order component of the brand love system from Studies 1 and 2 (e.g., self–brand integration combined current self-identity, life meaning, and desired self-identity), two separate composite indicators captured each conceptual component. In cases in which the conceptual component from Studies 1 and 2 matched up with a single first-order latent factor (e.g., anticipated negative affect), its items were randomly split into two composite indicators.

Structural equations modeling of a hierarchical, higherorder brand love prototype. In the spirit of the grounded theory–building objective underlying this project, we next explored the data using a hierarchical, higher-order structural model. We examined the factor correlations among the components estimated previously in our first-order models to determine which components might be grouped together as parts of higher-order factors and, where the empirical correlations we observed also made conceptual sense, to combine closely related conceptual components of brand love uncovered in our prior qualitative Studies 1 and 2. We observed that the following three groupings correlated with one another at levels that were (mostly) higher than with other subcomponents: (1) passion-driven behaviors (willingness to invest resources, passionate desire to use, and things done in the past [involvement]), (2) self– brand integration (life meaning, desired self-identity, current self-identity, and attitude strength 1 [frequent thoughts]), and (3) positive emotional connection (a sense of “intuitive fit” with the brand in which it feels “just right,” the extent to which the respondent feels emotionally attached or bonded to the brand, and the extent to which consumers have specific positive feelings [e.g., contentment, fun] connected with it). These correlation patterns were very similar

Comparison of mean levels. Mean differences were estimated between the high and low love samples using multiple groups measurement models. Confirming the results of Studies 1 and 2, the high-love-brand factor means were statistically higher (at p < .01) than the low-love-brand means on all components, with the smallest difference on attitude strength 2. Another such test showed that the high-lovebrand factor means were also statistically higher than the low-love-brand means on high quality and on L/WOM/R.

Nomological validity. A second structural equations model estimated the relationship between these 14 brand love prototype components, the antecedent of high quality, and the consequence of L/WOM/R. These models fit well, with each of our first-order brand love components relating positively and significantly (p < .01) to each of these two constructs (high quality and L/WOM/R) but none at levels high enough to threaten discriminant validity.

ages of the source items, in the estimation of each factor or subcomponent of it (Bagozzi and Edwards 1998).4 All 14 components possessed adequately high (> .6) levels of average variance extracted (Fornell and Larker 1981), and composite construct reliability levels > .7. The factors had adequately high discriminant validity ( coefficients < 1.0, or by the  difference test).

in the high and low brand samples.5 We modeled these three groupings as three second-order constructs. The four other (first-order) factors in this higher-order structural model were the remaining components that did not cluster in this manner. We depict these factors in Figure 1 and describe them in more detail in Table 1. Following emerging interpretations of formative versus reflective measurement and recent recommendations in consumer research and psychology (Bagozzi 2011), we represent the brand love prototype using reflective indicators of hierarchical organized factors. The three second-order factors and four first-order factors all loaded on a single third-order factor, which we labeled “brand love.” Then, we estimated this empirically driven but conceptually justifiable hierarchical structure of the brand love prototype. Relationships from the third-order brand love factor to the seven underlying brand love factors (three second-order and four first-order), and from them to their subcomponents, are all strong, positive, and significant, as hypothesized (see Figure 1 and Table 2). All composite indicators of the seven latent constructs also fit well and as hypothesized. Nomological validity. A follow-up model estimated the relationship between this higher-order “core” brand love factor and the two other constructs conceptualized as antecedents (high quality) or consequences (L/WOM/R). These models fit well for both high and low brand love. In them, the disattenuated correlation coefficients showed that our higherorder brand love factor was indeed related positively and significantly (at p < .01) to each of these two other constructs, but none so highly that it threatened discriminant validity.

5The only statistically significant difference across them was that while attitude strengths 1 and 2 correlated strongly (.56) in the high brand love sample, this correlation was much weaker (.12) in the low brand sample.

Measuring brand love as a single, unitary construct versus a multicomponent prototype. Managerially and theoretically, it should be more helpful to measure the individual components of the love prototype separately if the purpose is to understand all the options available for increasing love for a brand. Nonetheless, it could still be asked whether our multicomponent brand love prototype adds predictive value over a summatory, single-factor measure of overall brand love. In our questionnaire, we also asked respondents to rate the degree to which they loved the brand in question in an overall sense: (1) “Overall, how much do you love [Brand]?” (1 = “not at all,” and 10 = “very much”), and (2) “Describe the extent to which you feel love toward [Brand]” (1 = “not at all,” and 7 = “very much”). We used these items as separate indicators of a summatory measure of overall brand love. Confirmatory factor analyses (details omitted for brevity) show that the two indicators for overall brand love had high, positive, and significant loadings on the factor of .79 and .82 and average variance extracted and composite construct reliability statistics of .71 and .83; furthermore, the overall brand love latent factor had strong and positive (.84–.95), but significantly below 1.0, correlations with both our first-order and higher-order brand love factor models.

Brand Love / 9

FIGURE 1 Higher-Order Brand Love Factor Model
Passionate Desire to Use Passion-Driven Behaviors Willingness to Invest Resources Things Done in Past (Involvement) Desired Self-Identity Current Self-Identity Life Meaning Attitude Strength 1: Frequent Thoughts Intuitive Fit

Self–Brand Integration

Positive Emotional Connection Brand Love

Emotional Attachment Positive Affect Long-Term Relationship Anticipated Separation Distress Overall Attitude Valence Attitude Strength 2: Certainty/Confidence

Notes: The specific items for each indicator for each construct are not shown here for simplicity, but they are listed in Table 1.

More Abstract

More Concrete

6We also estimated predictive models using our first-order factor model. However, because of extensive collinearity across its 14 components, the individual path coefficients predicting brand loyalty/WOM were not usable. Therefore, we do not discuss these first-order predictive models further.

Therefore, we tested the comparative predictive power of this overall brand love factor, versus our brand love prototype higher-order factor, in explaining and predicting L/WOM/ R.6 Table 3 presents the models estimated here, with the key

Comparative Predictive Models

R-square statistics (in explaining brand loyalty/WOM) and fit statistics. First, we estimate Model 1 as a baseline, which is typical of conventional attitude models in marketing that rely on the brand’s perceived high quality to drive L/WOM/R. High quality alone significantly predicts brand loyalty/ WOM (standardized coefficient .41, p < .01, R2 =.17). Next, we estimate Models 2a and 2b, in which brand love, and nothing else (not even high quality), directly predict L/ WOM/R. In Model 2a, overall brand love significantly predicts L/WOM/R (standardized coefficient = .65, p < .01, R2 =

10 / Journal of Marketing, March 2012

TABLE 2 Path Coefficients of Higher-Order Structural Models
Brand love Æ Enduring passion Brand love Æ Self-brand integration Brand love Æ Long-term relationship Brand love Æ Positive emotional connection Brand love Æ Anticipated separation distress Brand love Æ Overall attitude valence Brand love Æ Attitude strength 2 (certainty/confidence) Enduring passion Æ Willingness to invest resources Enduring passion Æ Desire to use Enduring passion Æ Things done in past Self-brand integration Æ Life meaning and intrinsic rewards Self-brand integration Æ Desired self-identity Self-brand integration Æ Current self-identity Self-brand integration Æ Attitude strength 1 (talk/think frequently) Positive emotional connection Æ Intuitive fit Positive emotional connection Æ Emotional attachment Positive emotional connection Æ Positive affect Model Fit Statistics 2 (333) p< RMSEA NNFI CFI SRMR .96 .83 .59 .99 .55 .60 .62 .70 .79 .65 .80 .74 .90 .85 .84 .82 .78

High-Brand-Love Sample (n = 187)

Low-Brand-Love Sample (n = 187) .99 .90 .72 1.00 .54 .65 .19 .79 .83 .69 .76 .66 .92 .92 .90 .90 .63

TABLE 3 Predictive Model Statistics

640.67 .01 .075 .97 .98 .081

847.14 .01 .094 .96 .97 .069

3b High quality

3a High quality

2b Higher-order factor brand love

2a Overall brand love

1

High quality

.42). In Model 2b, our higher-order brand love factor also significantly predicts loyalty/WOM/resistance, but at a much higher level (standardized coefficient = .78, p < .01, R2 = .61). Note that because we are making comparisons only between models with the same number of variables, it is appropriate to compare these raw statistics with each other. Next, Models 3a and 3b use both high quality and brand love together, to predict L/WOM/R, with high quality predicting brand love as well (see Table 3). Model 3a uses overall brand love and significantly predicts L/WOM/R (standardized coefficient = .55, p < .01, R2 = .47). However, again, when we use the higher-order brand love factor

Notes: Model Fit Statistics: Model 1: 2(1) = .041 (1.01), p = .84 (.32), RMSEA = .002 (.004), NNFI = 1.00 (1.00), CFI = 1.00 (1.00), SRMR = .003 (.001). Model 2a: 2(1) = 2.15 (.31), p = .14 (.58), RMSEA = .078 (.00), NNFI = .97 (1.00), CFI = 1.00 (1.00), SRMR = .02 (.01). Model 2b: 2(387) = 752.96 (1020.60), p = .00 (.00), RMSEA = .076 (.097), NNFI = .97 (.96), CFI = .98 (.96), SRMR = .084 (.074). Model 3a: 2(6) = 4.82 (6.47), p = .57 (.37), RMSEA = .0003 (.02), NNFI = .99 (1.00), CFI = 1.00 (1.00), SRMR = .022 (.026). Model 3b: 2 (444) = 882.29 (1152.55), p = .00 (.00), RMSEA = .078 (.096), NNFI = .97 (.96), CFI = .97 (.96), SRMR = .089 (.078).

Higher-order factor brand love

Overall brand love

Loyalty/WOM/Resistance

Loyalty/WOM/Resistance

Loyalty/WOM/resistance

Loyalty/WOM/resistance

Loyalty/WOM/resistance

High Love (n = 187) .61 .42 .17

R2 for Brand Loyalty

Low Love (n = 187) .63 .52 .17

.63

.47

.65

.55

instead of the overall brand love factor (Model 3b), it not only significantly predicts L/WOM/R (standardized coefficient = .73, p < .01) but also produces a notably higher Rsquare of .63 (compared with .47 for Model 3a). To summarize, these comparative model tests show a noticeable improvement in the ability to predict brand loyalty, WOM, and resistance to negative information, by using the multiplecomponent higher-order brand love prototype as opposed to a summatory overall brand love measure. High- versus low-brand-love structure and relationships. Previously, we analyzed the data about brands that each respondent reported either loving or coming as close to love Brand Love / 11

Relationships. In almost all our models, the structural relationships among the brand love prototype components, the antecedent of high quality, and the consequence of L/WOM/R were very similar, with three exceptions. First, when we modeled the relationships between the antecedent of high quality and the brand love components in the firstorder model, we found a significant difference in the relationship of high quality with passionate desire to use, across the high- (low-) brand-love samples, which was .30 for low brand love but .12 for high brand love. That is, a passionate desire to use the brand was related to high quality much more for less-loved brands than for highly loved ones. Second, in the third predictive model (see Table 3), in which we used both high quality and brand love together to predict brand loyalty/WOM, we found that the direct path from high quality to L/WOM/R was not significant for high brand love but was significant for low brand love. High quality had a standardized coefficient into brand love of .46 for high brand love (and .40 for low brand love) and into L/WOM/R of .12 (n.s.) (direct = .13, p < .05). Importantly, while these last two path coefficients appear to be very close, formal chi-square tests revealed that they were actually significantly different at p < .05. Thus, for high brand love, the higher-order brand love prototype mediates all the effects of high quality on L/WOM/R; for low brand love, there are both direct and indirect effects. Finally, for Model 3b results (also Table 3), which compares the path coefficients in the model for high versus low

Structure. For both our first-order and hierarchical models, the structural models that fit our high-brand-love data also fit our low-brand-love data (n = 187) well, and all but two latent construct indicators again fit well and as hypothesized in the low-brand-love data. The exceptions were one item of positive affect and one of attitude strength (No. 2). Multiple-group tests showed that the model testing the equality of factor patterns across the high- and low-brand data samples—but allowing for just these two indicators to vary in their loadings—could not be rejected. Thus, an important finding from our modeling is that the relationships from the brand love factor to the underlying factors and subcomponents for low brand love are very similar (with the exception of two loadings) to those obtained for high brand love and that the structure of the brand love prototype does not appear to be different across high and low love brands.

as possible (hereinafter, “high love brands”). In this section, we compare these high love brands with low love brands, which are brands that each respondent provided in response to a question asking for a brand toward which they felt “mostly neutral about, instead of loved.” In our data, while respondents reported significantly lower levels of love for this second brand than they did for the first brand (on a tenpoint “how much you love it” item, the first had a mean of 7.39 and the second brand a mean of only 3.76, p < .001), even for this second brand, 80% of respondents reported levels of love greater than 1 (“not at all”). Thus, we refer to the second group as brands that are low in brand love rather than neutral brands.

brand love, while these path coefficients were mostly similar, they differed for the path from brand love to attitude strength 2 (certainty/confidence) at .66 high (.18 low). However, we could not perform a formal test of this difference, because two factor loadings for these two factors are significantly different for the two samples. We can claim statistically that the loading of attitude strength 2 on the higher-order brand love factor is significantly higher in the high-brand-love sample than in the low-brand-love sample. We return to this result in the “General Discussion” section. Theoretical Contributions

We began this article by pointing out that although brand love had recently emerged as an important concept among both practitioners and academics, theoretical progress has been hampered by (1) the substitution of a literature review on interpersonal love for the basic exploratory research on brand love needed to lay a solid foundation for future work, (2) the failure to distinguish between the love emotion and the longer-lasting and more complex love relationship, and (3) the nonutilization in the brand love domain of the nowaccepted prototype approach to identifying and defining different types of love. Therefore, using a grounded theory approach, we conducted two qualitative studies that uncovered ten consumer-experienced features of the brand love prototype. Then, we sorted these features into an antecedent (perceptions about great quality/qualities), the core of brand love, and consequences of brand love (brand loyalty, positive WOM, resistance to negative information, and willingness to pay a price premium) (see Table 1). Thus, our work builds on prior research by taking constructs that had previously been studied independently and showing that brand love can function as an integrated framework for investigating how they work together. Our qualitative studies also show that research on brand love that is derived directly from theories of interpersonal love tends to overlook the crucial issues of how loved brands become part of the consumer’s identity and provide intrinsic benefits. However, this does not mean that it is inappropriate to use the interpersonal relationship literature as a source of hypotheses, or even as supporting evidence, for research on consumer– brand relationships. In Study 3, we continued in the spirit of the grounded theory study of the brand love prototype to a quantitative measurement of these brand love prototype features, followed by a structural equations modeling analysis of these data. We estimated both first-order and higher-order representations of brand love that are useful for managerial analysis and action. Predictive models using the higherorder model showed that it predicts brand loyalty, WOM, and resistance to negative information better than a simple overall measure of brand love. Thus, through this programmatic set of studies, we show that in terms of maximum theoretical and explanatory power, the phenomenon of consumer love for a brand is

General Discussion

12 / Journal of Marketing, March 2012

modeled with much more richness and diagnostic insight, for both highly loved and less loved brands, when using a prototype conceptualization that includes seven distinct elements: (1) passion-driven behaviors reflecting strong desires to use it, to invest resources into it, and a history of having done so; (2) self–brand integration, including a brand’s ability to express consumers’ actual and desired identities, its ability to connect to life’s deeper meanings and provide intrinsic rewards, and frequent thoughts about it; (3) positive emotional connection that is broader than just positive feelings, including a sense of positive attachment and having an intuitive feeling of “rightness”; (4) anticipated separation distress if the brand were to go away; (5) long-term relationship, which includes predicting extensive future use and a long-term commitment to it; (6) positive attitude valence; and (7) attitudes held with high certainty and confidence. We believe that our higher-order prototype model adds value over the individual study of its included components in several ways. First, it leads to a much more comprehensive and integrated understanding of how consumers actually experience brand love than the prior academic study of its individual components, as stand-alone theoretical constructs, would suggest. Second, by showing how survey data on these prototype elements can be collected and modeled structurally, it demonstrates how more lower-level, concrete subcomponents can be used to influence higherlevel and more abstract consumer perceptions. For example, it shows three pathways through which consumers might develop a stronger emotional connection toward a brand: the association of the brand with various types of positive affect, the sensing of an intuitive fit with it, and the development of an “old friend”–like bond with it. Third, it allows for the situationally varying assessment of which component or subcomponent might have the strongest impact (weight) on the overall strength of felt brand love. For example, in our data on consumer electronics, structural model coefficients were highest for positive emotional connection (.99), passion-driven behaviors (.96), and self– brand integration (.83), and much lower for separation distress (.55), long-term relationships (.59), overall attitudes (.60), and attitude strength 2 (.62). Such diagnostic insight is useful theoretically and managerially. It is important to note here that our structural and measurement models for highly loved and less-loved brands did not differ significantly (only two measurement loadings were not invariant), with two significant exceptions. We found that attitude strength 2 (attitudes being held with greater certainty and confidence) had significantly higher loadings on higher-order brand love for more-loved brands than less-loved ones. This is theoretically reassuring, because more-loved brands with which consumers have more, and more direct, interactions indeed should have stronger (not just more positive) attitudes (e.g., Krosnick et al. 1993). We also found that beliefs about high quality of the brand were a significantly weaker contributor to L/WOM/R for more-loved brands than for less-loved brands. Again, this makes theoretical sense because for more-loved brands, aspects of brand love such as self–

brand integration and positive emotional connection should have relatively more influence. Managerial Implications The key managerial question is how brand managers can turn merely liked brands into loved brands and maintain that relationship over time. To repeat, our hierarchical model can assist managers in showing how more lowerlevel, concrete subcomponents can be targeted—through product and service design and marketing communications—to influence the higher-level and more abstract consumer perceptions that shape a consumer’s feeling of brand love. We discuss some of these ideas next. Furthermore, a manager with budget constraints needs to know which pathways and mechanisms to emphasize to maximize the return on investment of brand-love-increasing efforts. The path coefficients of the components in our higher-order model (Table 2) are higher for some (i.e., enduring passion, positive emotional connection, and self-brand integration) than for others (i.e., anticipated separation distress and long-term relationship), indicating the likely greater salience and importance of the former in creating brand love. However, our model estimates also showed that highly loved brands had higher means than less-loved brands on all the elements of the brand prototype factors. This suggests that all the elements represent potential pathways to build brand love, subject to category- and brand-specific opportunities and constraints, some of which managers might not have adequately appreciated before:

1. Facilitate passion-driven behaviors (reflecting strong desires to use the brand, to invest resources into it, and to interact frequently with it). The need to create a strong desire to use the brand suggests the need to employ design and packaging techniques that have been shown to create a strong hunger-like, visceral sense of desire (e.g., Belk, Ger, and Askegaard 2003; Norman 2004). Creating the need to invest resources (of both time and money) into, and to interact frequently with, the brand suggest the use of ways to encourage (or even require) accessorization and personalization, in enjoyable ways. Examples include Scion cars, which encourage owners to choose not only colors and sound systems but also armrests, interior lights, steering wheels, wheels and wheel covers, sport mufflers, and so on. From a product/service design perspective, it suggests creating more modular mix-and-match type product platforms and/or more software or service layers on top of the hardware, which invite such personalization. 2. Build brands that symbolize or facilitate self–brand integration, including not only the usual aspects, such as a brand’s ability to express the consumers’ actual and desired identities, but also the ability of brands to connect to what we refer to as life’s deeper meanings and important values. Furthermore, respondents told us that while their loved items provided them with a wide range of benefits, the delivery of intrinsic rewards was a key factor that set strongly loved brands apart. Previous results from the interpersonal love domain (Seligman, Fazio, and Zanna 1980) also suggest that love differs from liking by having more of the rewards tied to intrinsic rather than extrinsic benefits. If this is an important way in which love for a brand differs from liking for it, it seems that more-loved brands are those that are especially successful in linking themselves to a

Brand Love / 13

Taken together, our multiple uncovered-and-modeled features of the brand love prototype build on and extend prior marketing research focusing on attachment (Thomson, MacInnis, and Park 2005) and actual, or aspired-to, selfidentity (Escalas and Bettman 2005). We believe it is no coincidence that the Apple iPod was the brand most often rated by our respondents as their highest-loved consumer electronics brand: It exemplifies many of the components of brand love identified in this research. The iPod (and iPhone) product is all about personalizing the consumer’s music and other selections such as applications. This personalization requires substantial investments of time and effort and frequent interactions. And what could be more reflective of a person’s authentic self-identity, and arousing of passion, than one’s choice of music, which reflects deeply held life values and connects the user with others with similar musical tastes (Larsen, Lawson, and Todd

consumer’s sense of self-identity and giving life “meaning,” by connecting to some life aspects considered inherently important—possibly by linking credibly to social betterment (through corporate social responsibility campaigns) or to deeply held individual values (e.g., individual creativity, frugality and simplicity). Marketing communications could attempt to emphasize intrinsic benefits (e.g., happiness) over extrinsic ones (instrumental) or even to reframe the latter as the former. Other activities and domains with which the targeted consumer segment feels a strong sense of identityrelated involvement (e.g., sports, as is currently being leveraged by Samsung in its NFL partnership; the environment) can also strongly contribute to a consumer’s sense of selfidentity, and brands that credibly come to symbolize such self-identifying values can become strongly loved brands. 3. Create positive emotional connections with the brand. This is broader than just positive feelings; it includes a sense of attachment (“old friend,” bond) and an intuitive feeling of rightness about the brand. This may be achieved by endowing the brand with a sense of authenticity from its origin and history, the vision of its founders, and its corporate culture, so that the brand buyer feels a sense of kinship about it. In the interviews, brands that seemed to come from the heart of their producers had a much easier time finding a place in the hearts of their consumers. Examples include Patagonia (with founder Yvon Chuinard) and Body Shop (by Anita Roddick). The needed sense of emotional attachment— the brand feeling like an old friend and the building of a bond with it—could also be developed through the creation of active and tightly knit brand communities, such as that of Harley-Davidson, and brand presence at emotionally meaningful events such as music festivals or NASCAR races. 4. Brands that become valuable and trusted resources, as sources of expertise and advice, are possibly most able to create and leverage a feeling of anticipated separation distress if the brand were to go away. Examples include Tide’s stain-fighting website or Kraft’s recipe website. 5. A sense of long-term relationship with the brand going forward could be created by loyalty programs—though these should emphasize intrinsic over extrinsic motives and rewards—and by marketing programs for brands that require frequent and ongoing, rather than one-shot and initial, updating and interaction. Examples might include “brand community” Facebook pages for brands (e.g., those of Coca-Cola) that incentivize frequent visits, dialog, frequent reading and input of content, and so on.

2009)? It is also noteworthy that the iPod’s industrial design (the simplicity of its click-wheel mechanism and menu navigation) is likely to promote a sense of natural and intuitive fit. Almost all (89%) respondents in Study 2 reported truly loving at least one brand. However, the brands consumers really love are likely to be a small minority of the total brands they purchase. Can our new understanding of brand love also be relevant for more mundane brands? Although this question requires further study, our findings suggest that brand love may indeed be a useful construct in many situations. Even after directing respondents to select a brand they felt “mostly neutral about, instead of loved” and allowing respondents to choose any consumer electronics brand they wished that met this criterion, 80% of respondents still reported feeling some love for these brands. This suggests that love is a continuous variable with a broad positive range and that even brands that not usually thought of as loved may potentially possess enough features of the love prototype for our brand love model to be applied to them. This conclusion is strongly supported by the finding that our brand love models were similar for both high and low love brands and effectively predicted brand loyalty, WOM, and resistance to negative information for both low and high love brands. More work is needed to determine how the components of brand love identified here interact with one another. In particular, further conceptual and empirical work is needed to more definitively assess the role of attitude valence and attitude strength within the brand love prototype.7 Furthermore, just as there are different types of interpersonal love (e.g., parental, romantic, sexual), there might be different types of brand love. Experimental research manipulating the features of the brand love prototype would complement our cross-sectional research, helping to more unambiguously establish causal directions. Longitudinal research on the temporal development—and possible waning—of brand love would also be very useful. Our candidate measures for willingness to pay higher prices, as an outcome of brand love, were not effective in Study 3, suggesting the need for better measures. Research is needed to broaden the generalizability of our findings to other types of consumers and categories, particularly durables and services. Although Study 1 examines love for consumption activities, not just brands, further study is needed of the extent to which our findings apply to other branded objects and possessions. Given the increasing importance of brand love to marketing theory and practice, our current contributions are only a beginning, and a great deal of work remains.
7Although not reported in this article for reasons of space, we also estimated structural equation models treating attitude valence as a covariate and attitude strength as a consequence. Our conclusions regarding the explanation and prediction of brand loyalty/ WOM by the other brand love prototype components remained the same.

Limitations and Further Research

14 / Journal of Marketing, March 2012

Aggarwal, Pankaj (2004), “The Effects of Brand Relationship Norms on Consumer Attitudes and Behavior,” Journal of Consumer Research, 31 (1), 87–101. Ahearne, Michael, C.B. Bhattacharya, and Thomas Green (2005), “Antecedents and Consequences of Customer-Company Identification: Expanding the Role of Relationship Marketing,” Journal of Applied Psychology, 90 (3), 574–85. Ahuvia, A.C. (1993), “I Love It! Towards a Unifying Theory of Love Across Diverse Love Objects,” doctoral dissertation, University of Michigan. — — (2005), “Beyond the Extended Self: Loved Objects and — Consumers’ Identity Narratives,” Journal of Consumer Research, 22 (1), 171–84. — —, Rajeev Batra, and Richard Bagozzi (2009), “Love, Desire — and Identity: A Conditional Integration Theory of the Love of Things,” in The Handbook of Brand Relationships, Deborah J. MacInnis, C. Whan Park, and Joseph R. Priester, eds. Armonk, NY: M.E. Sharpe, 342–57. Albert, Noel, Dwight Merunka, and Pierre Valette-Florence (2008), “When Consumers Love Their Brands: Exploring the Concept and its Dimensions,” Journal of Business Research, 61 (10), 1062–75. Aron, Elaine N. and Arthur Aron (1996), “Love and Expansion of the Self: The State of the Model,” Personal Relationships, 3 (1), 45–58. — — — and Lori Westbay (1996), “Dimensions of the Prototype of Love,” Journal of Personality and Social Psychology, 70 (3), 535–51. Babin, Barry J., William R. Darden, and Mitch Griffin (1994), “Work and/or Fun: Measuring Hedonic and Utilitarian Shopping Value,” Journal of Consumer Research, 20 (4), 644–56. Bagozzi, Richard P. (2011), “Measurement and Meaning in Information Systems and Organizational Research: Methodological and Philosophical Foundations,” MIS Quarterly, 35 (2), 261–92. — — and Jeffrey R. Edwards (1998), “A General Approach for — Representing Constructs in Organizational Research,” Organizational Research Methods, 1 (1), 45–87. — — and Kyu-Hyun Lee (2002), “Multiple Routes for Social — Influence: The Role of Compliance, Internalization, and Social Identity,” Social Psychological Quarterly, 65 (3), 226–47. Barsalou, Lawrence W. (1991), “Deriving Categories to Achieve Goals,” in The Psychology of Learning and Motivation: Advances in Research and Theory, Gordon H. Bower, ed. San Diego: Academic Press, 1–64. Bauer, Hans, Daniel Heinrich, and Carmen-Maria Albrecht (2009), “All You Need Is Love: Assessing Consumers’ Brand Love,” in Proceedings of the American Marketing Association Summer Educators Conference, Michael Kamin and Ingrid M. Martin, eds. Chicago: American Marketing Association, 252–53. Belk, Russell W. (1988), “Possessions and the Extended Self,” Journal of Consumer Research, 15 (2), 139–68. — —, Guliz Ger, and Søren Askegaard (2003), “The Fire of — Desire: A Multisited Inquiry into Consumer Passion,” Journal of Consumer Research, 30 (3), 326–51. Bowlby, John (1979), The Making and Breaking of Affectional Bonds. London: Tavistock. Capps, Brooke (2007), “How to Succeed in Second Life,” Advertising Age, (May 28), 6. Carroll, Barbara A. and Aaron C. Ahuvia (2006), “Some Antecedents and Outcomes of Brand Love,” Marketing Letters, 17 (2), 79–90. Escalas, Jennifer Edson and James R. Bettman (2003), “You Are What They Eat: The Influence of Reference Groups on Con-

REFERENCES

sumer Connections to Brands,” Journal of Consumer Psychology, 13 (3), 339–48. — — and — — (2005), “Self-Construal, Reference Groups, and — — Brand Meaning,” Journal of Consumer Research, 32 (3), 378–89. Fehr, Beverley (1988), “Prototype Analysis of the Concepts of Love and Commitment,” Journal of Personality and Social Psychology, 55 (4), 557–79. — — (2006), “A Prototype Approach to Studying Love,” in The — New Psychology of Love, R.J. Sternberg and K. Weis, eds. New Haven, CT: Yale University Press, 225–46. — — — (2009), “Love, Prototype Approach,” in Encyclopedia of Human Relationships, Harry T. Reis and Susan Sprecher, eds. Thousand Oaks, CA: Sage Publications. — — and James A. Russell (1991), “The Concept of Love — Viewed from a Prototype Perspective,” Journal of Personality and Social Psychology, 60 (3), 425–38. Fornell, Claes and David F. Larcker (1981), “Evaluating Structural Equation Models with Unobservable Variables and Measurement Error,” Journal of Marketing Research, 18 (February), 39–50. Fournier, Susan (1998), “Consumers and Their Brands: Developing Relationship Theory in Consumer Research,” Journal of Consumer Research, 24 (4), 343–73. Guadagni, Peter M. and John Little (1983), “A Logit Model of Brand Choice Calibrated on Scanner Data,” Marketing Science, 2 (3), 203–238. Harding, Lora Mitchell and Ashlee Humphreys (2011), “Do Brands Like Us? The Perceived Liking of the Brand for the Self and Brand Attraction,” working paper, Northwestern University. Hatfield, Elaine (1988), “Passionate and Compassionate Love,” in The Psychology of Love, Robert J. Sternberg and Michael L. Barnes, eds. New Haven, CT: Yale University Press, 191–217. Hazan, Cindy and Debra Zeifman (1999), “Pair Bonds as Attachments,” in Handbook of Attachment, Jude Cassidy and Phillip R. Shaver, eds. New York: Guilford, 336–54. Hendrick, Clyde and Susan S. Hendrick (2006), “Styles of Romantic Love,” in The New Psychology of Love, Robert Sternberg and Karin Sternberg, eds. New Haven, CT: Yale University Press, 149–70. Holt, Douglas B. (1997), “Poststructuralist Lifestyle Analysis: Conceptualizing the Social Patterning of Consumption in Postmodernity,” Journal of Consumer Research, 23 (4), 326–50. Hu, Li-tze and Peter M. Bentler (1999), “Cutoff Criteria for Fit Indexes in Covariance Structure Analysis: Conventional Criteria Versus New Alternatives,” Structural Equation Modeling, 6 (1), 1–55. Kleine, Susan S., Robert E. Kleine III, and Chris T. Allen (1995), “How Is a Possession Me or Not Me? Characterizing Types and an Antecedent of Material Possession Attachment,” Journal of Consumer Research, 22 (3), 327–43. Krosnick, Jon A., David S. Boninger, Yao C. Chuang, Matthew K. Berent, and Catherine G. Carnot (1993), “Attitude Strength: One Construct or Many Related Constructs?” Journal of Personality and Social Psychology, 65 (6), 1132–51. Larsen, Gretchen, Rob Lawson, and Sarah Todd (2009), “The Consumption of Music as Self-Representation in Social Interaction,” Australasian Marketing Journal, 17 (1), 16–26. Lincoln, Yvonna S. and Egon G. Guba (1985), Naturalistic Inquiry. Newbury Park, CA: Sage Publications. McAlexander, James H., John W. Schouten, and Harold F. Koenig (2002), “Building Brand Community,” Journal of Marketing, 66 (January), 38–54. McCracken, Grant (1988), The Long Interview. Newbury Park, CA: Sage Publications.

Brand Love / 15

Murstein, B.I. (1988), “A Taxonomy of Love,” in The Psychology of Love, R.J. Sternberg and M.L. Barnes, eds. New Haven, CT: Yale University Press, 13–37. Norman, Donald A. (2004), Emotional Design. New York: Basic Books. Park, C. Whan, Deborah J. MacInnis, Joseph Priester, Andreas B. Eisengerich, and Dawn Iacabucci (2010), “Brand Attachment and Brand Attitude Strength: Conceptual and Empirical Differentiation of Two Critical Brand Equity Drivers,” Journal of Marketing, 74 (November), 1–17. Richins, Marsha L. (1994), “Valuing Things: The Public and Private Meanings of Possessions,” Journal of Consumer Research, 21 (3), 504–521. — — (1997), “Measuring Emotions in the Consumption Experi— ence,” Journal of Consumer Research, 24 (2), 127–46. Roberts, Kevin (2004), Lovemarks: the Future Beyond Brands. New York: Power House Books. Rosch, Eleanor (1975), “Cognitive Representations of Semantic Categories,” Journal of Experimental Psychology: General, 104 (3), 192–233. Rubin, Zick (1970), “Measurement of Romantic Love,” Journal of Personality and Social Psychology, 16 (2), 265–73. Rust, Roland T. and Burt Cooil (1994), “Reliability Measures for Qualitative Data: Theory and Implications,” Journal of Marketing Research, 31 (February), 1–14.

Seligman, Clive, Russell H. Fazio, and Mark P. Zanna (1980), “Effects of Salience of Extrinsic Rewards on Liking and Loving,” Journal of Personality and Social Psychology, 38 (3), 453–60. Shaver, Phillip, Judith Schwartz, Donald Kirson, and Cary O’Connor (1987), “Emotion Knowledge: Further Exploration of a Prototype Approach,” Journal of Personality and Social Psychology, 52 (6), 1061–86. Shimp, Terrence A. and Thomas J. Madden (1988), “Consumer– Object Relations: A Conceptual Framework Based Analogously on Sternberg’s Triangular Theory of Love,” in Advances in Consumer Research, Michael J. Houston, ed. Provo, UT: Association for Consumer Research, 163–68. Sternberg, Robert J. (1986), “A Triangular Theory of Love,” Psychological Review, 93 (2), 119–35. — — (1997), “Construct Validation of a Triangular Love Scale,” — European Journal of Social Psychology, 27 (3), 313–35. Strauss, Anselm and Juliet Corbin (1994), “Grounded Theory Methodology: An Overview,” in Handbook of Qualitative Research, Norman K. Denzin and Yvonna S. Lincoln, eds. Thousand Oaks, CA: Sage Publications, 273–85. Thomson, Matthew, Deborah J. MacInnis, and C.W. Park (2005), “The Ties That Bind: Measuring the Strength of Consumers’ Emotional Attachments to Brands,” Journal of Consumer Psychology, 15 (1), 77–91.

16 / Journal of Marketing, March 2012

This article has been cited by: 1. Sandra Maria Correia LoureiroConsumer-Brand Relationship 414-434. [CrossRef] 2. Brian S. Fugate, Chad W. Autry, Beth Davis-Sramek, Richard N. Germain. 2012. Does knowledge management facilitate logistics-based differentiation? the effect of global manufacturing reach. International Journal of Production Economics 139:2, 496-509. [CrossRef] 3. John R. Rossiter. 2012. A new C-OAR-SE-based content-valid and predictively valid measure that distinguishes brand love from brand liking. Marketing Letters . [CrossRef] 4. Susan Fournier, Claudio Alvarez. 2012. Brands as relationship partners: Warmth, competence, and in-between. Journal of Consumer Psychology 22:2, 177-185. [CrossRef] 5. Kevin Lane Keller. 2012. Understanding the richness of brand relationships: Research dialogue on brands as intentional agents. Journal of Consumer Psychology 22:2, 186-190. [CrossRef]