Image congruence has been viewed as a multidimensional construct consisting of actual, ideal, social self-image congruence, and functional congruence (Sirgy, 1982). Consumers tend to buy products/brands when they perceive the symbolic image of the products/brands matches their actual, ideal, or social image. In addition, consumers may buy products/brands whose functional performance matches their expectations or desires (Belk, 1988; Sirgy, 1982), referring to functional image congruence. To this end, we use a holistic approach to reflect both symbolic and functional congruence by identifying three types of image congruence: self, functional, and social image congruence.

  Image congruence has been found to influence consumers’attitudinal and behavioral responses to a brand (Kressman et al., 2006; Malär et al., 2011; Srigy, Johar, Samli, & Claiborne, 1991). It is well established that con-gruence between brand image and self-concept enhances consumer’s emotional connection with the brand (Malär et al., 2011). Further evidence suggests that stronger brand attachment, in turn, leads to greater consumer loyalty, increases the company’s financial performance, and strengthens consumer- brand relationships (Park, Maclnnis, Priester, Eisingerich, & Lacobucci, 2010; Thomson & Johnson, 2006). However, there is a dearth of empirical research that examines why and how image congruence influences consumers’ affective responses such as emotional brand attachment. Moreover, little attention has been paid to the mediation process underlying the structural relationship between multidimensional image congruence and brand loyalty.

  Based on the tripartite model of consumer-brand relationships, including cognitive, emotional, and behavioral facets (Blackston, 1991), the purpose of this study is to examine how consumers’ perceived image congruence (cognitive facet) influences the formation of emotional bond with the brand (emotional facet), thereby leading to heightened levels of brand loyalty (behavioral facet). Specifically, this study aims to develop and test the integrative model involving the effects of three types of image congruence (i.e., self, functional, and social image congruence) on emotional attachment and brand loyalty. The proposed model sheds new light on understanding the effects of image congruence on emotional brand attachment and the moderating role of product type (i.e., self-expressive vs. functional product).

Theoretical background

 Self-concept and image congruence

  Central to the concept of image congruence is self-concept which refers to a cognitive referent structure in evaluating the symbolic meaning of a product or brand (Sirgy, Grewal, & Mangleburg, 2000). Research has demonstrated that image congruence occurs when consumers choose focal objects that bolster their self-concept (Hogg et al., 2000; Sirgy, 1982; Sirgy, Lee, Johar, & Tidwell, 2008). Consumers tend to prefer or purchase products/ brands that are consistent with their own self-image (Sirgy et al., 2008). For example, Hogg et al. (2000) explore the relationships among self-concept, image congruence, and product/brand evaluation. The authors demonstrate that consumers are more likely to select a product/brand congruent with their own self-concept.

  While earlier research has discussed a unidimentional perspective of self-concept (e.g., Grubb & Grathwohl, 1967; Hamm & Cundiff, 1969), Sirgy (1982) views self-concept as a multidimensional construct that is comprised of three domains of the self: actual, ideal, and social. Building on his premise, three types of image congruence are proposed in the branding context including actual self-image congruence, ideal self-image congruence, and social self-image congruence. First, actual self-image congruence is defined as the extent to which consumers actually see themselves with the brand image or personality. Second, ideal self-image congruence represents the extent to which consumers match their desired image with the brand image. When a brand’ symbolic image reflects a consumer’ actual or ideal self-image, the consumer tends to buy the brand in order to maintain and enhance his or her self-esteem. Finally, social self-image congruence is conceptualized as the extent to which consumers’beliefs about how they are viewed by others are consistent with the brand image. Given that symbolic brand image can mirror the user’ image, consumers tend to buy or use a brand as a certain way of expressing themselves to others (Aaker, 1999; Sirgy et al., 1991).

  On the other hand, consumers are likely to evaluate products/brands based on their functional or utilitarian attributes (Park, Macinnis, & Priester, 2006; Sirgy et al., 2000). While symbolic meaning is derived from the symbolic or experiential attributes of a brand (e.g., sexy, modern, classic and friendly), functional value is associated with the utilitarian attributes of a brand (e.g., quality, price, and performance) (Sirgy, 1982). When the functional or utilitarian attributes of products/brands reflect the users’ expectations or desires related to those attributes, functional image congruence is created (Kressmann et al., 2006; Sirgy et al., 1991). For example, a specific brand’s laptop computer may be evaluated in terms of a set of utilitarian attributes, such as CPU processor, memory, price, display size, and weight. When this product provides functional attributes that meet the consumer’s expectation, functional image congruence can be established. Based on the above theoretical posits and empirical evidence, this study integrates both symbolic (self and social) and functional aspects of consumer brand evaluations into the sub-categorization of image congruence including self-image congruence,social image congruence, and functional image congruence.

  Brand Attachment

  The concept of attachment has been originated by Bowlby (1979) who viewed it as a mutual relationship between parents and infants. Such attachment involves the basic human need to be attached to particular others. The children’s desire for attachment to their parents is continued through the adult stage via peer group relationships (Trinke & Bartholomew, 1997) and romantic relationships (Hazen & Shaver, 1987). This concept of human attachment has evolved into brand attachment in the marketing context.

  Attachment is strongly time-dependent because “ttachment develops over time as relationship between the self and the entity evolves”(Park et al., 2010, p. 4). Importantly, consumers develop a positive feeling of “oneness” with the entity (e.g., brand) since they view the entity’ resources as their own possession (Mittal, 2006). Although a brand-self link is exemplified by a cognitive memory network, attachment can be regarded as an intrinsically emotional construct (Mikulincer & Shaver, 2007). Building on the general principle of attachment theory (Mikulincer & Shaver, 2007), brand attachment is conceptualized as the strength of the emotional bond connecting the self with the brand (Thomson, Macinnis, & Park, 2005). Since consumers may have a deep emotional attachment to a specific brand, Thomson, Macinnis, and Park (2005) develop a measure of emotional brand attachment as a multidimensional variable, characterized by affection, passion, and connection. Strong brand attachment is found to affect positive consumer responses to the brand, such as brand attitude, satisfaction, and commitment to the brand (i.e., brand loyalty) (Mikulincer & Shaver, 2007; Park et al., 2006; Thomson et al., 2005).

  Some researchers argue that brand attachment should be distinguished from other constructs such as brand attitude and satisfaction (e.g., Park et al., 2010; Thomson et al., 2005). Brand attachment accounts for “ot”stimulus- induced feelings that are derived from consumer-brand relationships, whereas brand attitude involves “old”feelings because it is associated with the holistic judgment about brands (Park et al., 2010). Brand attachment also differs from satisfaction because attachment tends to develop over time and multiple interactions are required, whereas satisfaction can occur immediately (Park et al., 2010). Previous research also suggests that consumer satisfaction may not be sufficient for creating a long-term brand relationship (Charroll & Ahuvia, 2006; Vlachos, Theotokis, Pramatari, & Vrechopoulos, 2010). To illustrate, Vlachos et al. (2010) argue that satisfied consumers do not always re-patronize a brand nor build a brand loyalty through affectionate ties with a brand. Consumers who feel attached to specific products/ brands are not likely to replace them with other products/brands and experience emotional loss when they lose their attached objects (Lyons-Ruth, Dutra, Schuder, & Bianchi, 2006). This attachment is likely to occur more with durable products (e.g., laptop, designer wallet, furniture, and automobile) than with non-durable goods (e.g., medicine, toothpaste, and paper towel) whose lifetime is shorter (Ball & Tasaki, 1992).

Hypotheses development

  Based on the previous discussions, we develop a research model to postu-late that self-image congruence, functional image congruence, and social imagencongruence lead to brand loyalty through brand attachment.

 The effect of different types of image congruence on brand attachment

 Many companies attempt to find a strategic way to create strong relational bonding with their consumers because such attachment has a profound impact on attitudinal and behavioral responses to the brand (Park et al., 2010). The effects of self-image congruence on brand attachment has been explored in marketing literature. For example, consumers’self-image congruence plays an important role in building and fostering emotional brand attachment (Malär et al., 2011). Specifically, actual self-image congruence has a stronger influence on consumers' emotional brand attachment than ideal self-image congruence (Malär et al., 2011). As suggested by Mugge, Schifferstein, and Schoormans (2006), consumers tend to develop stronger emotional attachment to products with a personality that is congruence to their self-image. As the aforementioned studies indicate, brand consumption that is consistent with consumers’self-image leads to evoke positive feelings toward the brand and reinforces emotional brand attachment (e.g., Malär et al., 2011; Mugge et al., 2006). Based on the above argument, the following hypothesis is proposed:

  H1. Self-image congruence will positively impact brand attachment.

  Extant literature has suggested that functional image congruence is anstrong predictor of consumer behavior such as brand preference and brandnloyalty. According to Sirgy and Samli (1985), the impact of functional imagencongruence on store loyalty is greater than that of self-imagencongruence. Sirgy et al. (1991) claim that the greater the congruence betweennthe brand’'s utilitarian/functional attributes and the referent’'s beliefs,nthe more likely the consumer is going to hold a favorable attitude towardnthe brand. If there is a match between brand utilitarian attributes and consumers’'nexpectations of its performance, favorable attitudes toward thenbrand are formed, which leads to stronger attachment to the brand. In sum,nthe greater the functional congruence between consumers and brands, thenmore likely that brand attachment is formed. These arguments lead us tonpropose the following hypothesis:

  H2. Functional image congruence will positively impact brand attachment.

  Elliott (1997) suggests that self-symbolism and social symbolism can con-tribute to the creation of symbolic brand meanings. This is somewhat con-sistent with the findings of Grubb and Grathwohl (1967) who report that consumers tend to use a product/brand as a way of enhancing self-concept (as an individual and social experience process) because they want to transfer socially attributed meanings of the product/brand to themselves.

  The function of social image congruence in consumer behavior can be explained by two motivations: social consistency motive and social approval motive (Johar & Sirgy, 1991; Sirgy et al., 1991). Social consistency motive suggests that consumers are motivated to keep their self- image in a way that they want to be seen by others, while social approval motive denotes that consumers have a motivation to earn approval from others (Sirgy et al., 2000). Therefore, we propose that consumers use brands that are consistent with their social image to gain social approval from others.

  Although previous literature finds the effect of self-image congruence (i.e., actual and ideal self) on consumers’ behavioral responses to brands, there have been inconsistent findings on the social image congruence relationship. Some studies fail to show the influence of social image congruence on brand preference, purchase intention, and store loyalty (Maheshwari, 1974; Samli & Sirgy, 1981), while others report the positive effect of social image congruence on consumer behaviors such as product preference and ownership (Ericksen & Sirgy, 1989), store loyalty (Samli & Sirgy, 1981), and brand loyalty (Back, 2005). Specifically, Han and Back (2008) point out that social image congruence is a significant predictor of customer emotion and loyalty in the lodging industry. Given the prevailing notion that consumers purchase products/brands as a way of expressing their self-image to others (Aaker, 1999), we argue that the congruence between a consumer’ social image and the brand’ image yields positive emotional responses such as emotional attachment. Based on this argument, we propose the following hypothesis:

  H3. Social image congruence will positively impact brand attachment.

 The effect of brand attachment on brand loyalty

  Brand attachment predicts consumers’ commitment to a brand because highly attached consumers are willing to preserve interaction with the brand, thereby leading to greater brand loyalty and inclination to pay price premium (Fedorikhin, Park, & Thomson, 2008; Thomson et al., 2005). Similarly, Park et al. (2006) indicate that brand attachment is a significant predictor of brand loyalty because it enhances consumers’ actual purchase behaviors. Furthermore, Fedorikhin et al. (2008) state that consumers who are highly attached to a brand tend to view the brand as part of themselves and sustain their relationship with the brand, and therefore, are less likely to use competing brands. Therefore, the following hypothesis is proposed.

 H4. Brand attachment will positively impact brand loyalty.

 The moderating role of product type

  It is important to note that the effect of image congruence on brand at-tachment may not be equally pronounced for all consumer product types. Thus, this study attempts to examine whether image congruence (i.e., self, social, or functional) operates differently in the formation of brand attachment, according to product type (expressive versus functional). According to Baek, Kim, and Yu (2010), expressive products (e.g., clothing, handbag, and dress shoes) typically provide consumers with symbolic and affective benefits, while functional products (e.g., laser printer, toothpaste, and pain reliever) deliver functional and cognitively-oriented benefits to consumers.

  Katz’ (1960) functional theory of attitudes alludes that the concepts of self-image, social image, and functional image congruence make differential impacts on consumer attitude and behavior by product type. Katz (1960) identifies four psychological functions that attitudes serve: (1) an ego-defensive function, (2) a knowledge function, (3) a utilitarian function, and (4) a value-expressive function. Among these functions, the value-expressive function, which represents a means for establishing and maintaining norms of social appropriateness, is formed through self-image and social image congruence; and the utilitarian function, which represents a means for maximizing reward experiences and minimizing negative ones, is driven by functional image congruence (Johar & Sirgy, 1991). This argument leads to the assumption that consumers are strongly influenced by the expressive attributes of a product/brand when self-image or social image congruence is high, while they are strongly influenced by the utilitarian attributes of a product/brand when functional image congruence is high (Johar & Sirgy, 1991).

  Thus, we propose that the effect of self-image congruence or social image congruence on brand attachment is greater for an expressive product than for a functional product. On the other hand, we expect that the effect of functional congruence on brand attachment is stronger for a functional product than for an expressive product. Accordingly, the following hypotheses are proposed.

  H5. The impact of self-image congruence on brand attachment will be stronger for an expressive product than for a functional product.

  H6. The impact of functional image congruence on brand attachment will be stronger for a functional product than for an expressive product.

  H7. The impact of social image congruence on brand attachment will be stronger for an expressive product than for a functional product.

 Pilot test: Classification of functional/expressive products

 Prior to data collection, we conducted a pretest to identify two product types: a highly functional product and a highly expressive product. This is to generate variations in the participants’ responses to brand images and attributes and also to test the moderator effect with these two product types. First, an initial pool of eight product categories was selected from the MRI’s syndicated market research database (www.mriplus.com) on the basis of frequency: designer jeans, designer wallet, watch, perfume, laptop, pillow, laser printer, and ski jacket. Next, in order to identify functional and expressive products, we recruited faculty, staff members, and students in one of the major southeastern universities in the United States. We asked the participants (n = 60) to evaluate the extent to which they would characterize the product as primarily for a functional use or for an expressive use. The scale for the functional versus expressive value was adopted from Kempf (1999) who used a 7-point rating scale (1 = Primarily for a functional use and 7 = primarily for an expressive use). Based on the mean score, laptop (M = 2.94) was selected as a functional product and designer wallet (M = 6.11) was selected as an expressive product. The price ranges of these two products were not significantly different because laptops (e.g., Dell and HP) were mostly in the $250 - $1,000 range and designer wallets (e.g., Coach and Gucci) were mostly in the price range of $200 - $1,000.

 Main study: Sample and measurement instruments

 The data for the main analyses were collected using a web survey. Consumer panelists of a commercial online survey company received an e-mail invitation to complete an online survey hosted by a major university in the southeast region of the United States. This survey was conducted for a larger-scale study, for which questions on different products were solicited.

 For this study, a total number of 673 cases (laptop = 352 and designer wallet = 321) were used. A manipulation check question was asked to verify that designer wallet or laptop was perceived as an expressive product or a functional product, respectively. The question was “Would you characterize the given product as primarily for functional or for self-expressive use?” and it was rated using the same rating scale as the one used in the pretest. We selected only moderately or highly functional (ratings of 1, 2, and 3) and expressive (ratings of 5, 6, and 7) products, resulting in 403 cases (229 cases for laptop and 174 cases for designer wallet).

 The age of participants ranged from 18 to 79 years, with an average age of 38.6. Gender was evenly disturbed with 49.9% male and 50.1% female. The majority of participants were Caucasian (73.7%), followed by Asian American (9.2%), African American (7.9%), and Hispanic American (7.2%). The largest number (35.7%) of the participants completed some college or vocational school, followed by bachelor’s degree (35.0%), and high school (15.6%). Household income was widely distributed with the median income of $45,000-$59,999.

  At the beginning of survey, we asked the respondents to select one of the given product categories (i.e., designer wallet and laptop) that they purchased within the past five years. Next, they were asked to provide the name of their favorite brand for the product category that they selected. The brand names that were most frequently mentioned for each product category were: Coach (n = 54), Gucci (n = 20), and Louis Vuitton (n = 15) for designer wallet; and Dell (n = 65), HP (n = 50) and Toshiba (n = 25) for laptop. After respondents selected their brand, they responded to ques-tions on self-image congruence, functional image congruence, social image congruence, brand attachment, and brand loyalty.

 Table 1 illustrates the measurement items for each construct. The measurement items of self-image congruence were adopted from Sirgy and Johar (1999) to capture both actual and ideal self-image congruence. The items of functional image congruence were adopted from Sirgy et al. (1991); social image congruence, from Carroll and Aluvia (2006); brand attachment, from Thomson et al. (2005); and brand loyalty, from Chaudhuri and Holbrook (2001). Brand attachment items were measured using a 7-point rating scale (1 = not at all, 7 = very well); and the other scales were measured on a 7-point Likert-type scale (1 = strongly disagree, 7 = strongly agree).

Table 1. Summary of measurement model statistics

Notes: CR = critical ratio; the first item for each measure was set to a value of 1.00; *p<.05

Results

  To test the hypothesized relationships among latent constructs (i.e., self-image congruence, functional image congruence, social image congruence, brand attachment, and brand loyalty), a structural equation modeling (SEM) was run via Amos 16. Based on the recommendation of Anderson and Gerbing (1988), we validated the measurement model using a confirmatory factor analysis (CFA) prior to running SEM. Parameters were estimated using Maximum Likelihood method. In terms of the model fit, we used comparative fit index (CFI), Tucker-Lewis index (TLI), normed fit index (NFI), root mean square error of approximation (RMSEA), and standardized root mean square residual (SRMR).

  The CFA results provided a good fit of the data: χ²(987) = 2622.2, CFI = .93, TLI = .92, NFI = .92, RMSEA = .05, and SRMR = .05. Construct validity was checked with both convergent and discriminant validity. Convergent validity was confirmed in three different ways. First, the standardized factor loadings, ranging from .69 to .98, were statistically significant (p < .05). Second, all construct reliability values of the latent constructs ranged from .91 to .97, well above the threshold value (> .70) (Hair, Anderson, Tatham, & Black, 1998). Third, the average variance extracted (AVE) ranged from .63 to .86, exceeding the recommended level of .50 (Forenell & Larcker, 1981). Discriminant validity was confirmed by AVEs larger than the shared variances (i.e., squared correlation coefficients) between all possible pairs of constructs (Fornell & Larcker, 1981) (Table 2). Thus, the construct validities of all latent constructs were satisfactory.

Table 2. Factor correlation matrix of constructs

Notes: CR = composite reliability; AVE = average variance extracted; *p<.05

 The structural model also provided satisfactory model fit: χ²(996) = 3008.9, CFI = .92, TLI = .91, NFI = .91, RMSEA = .05, SRMR = .07. As illustrated in Figure 1, the positive effect of self-image congruence on brand attachment was significant (β = .46, p < .001), supporting H1. Functional image congruence was also found to increase brand attachment (β = .15, p < .01), supporting H2. Furthermore, there was a positive impact of social image congruence on brand attachment (β = .40, p < .001), supporting H3. Finally, brand attachment positively affected brand loyalty (β = .46, p < .001), supporting H4.

Figure 1. Final model

Notes: χ² (996) = 3008.9 (p<.001), CFI = .92, TLI = .91; NFI = .91; RMSEA = .05; SRMR= .07; *p<.05; ***p<.001

 Testing for mediation

 The current investigation further examined whether the effects of self-im-age, functional image, and social image congruence on brand loyalty were mediated by emotional brand attachment. In order to test the indirect effects of image congruence dimensions, we conducted a bootstrapping analysis with 5,000 re-samples drawn from the original data that were used to estimate the bias-corrected 95% confidence intervals (Zhao, Lynch, & Chen, 2010). As suggested by Zhao et al. (2010), the indirect effects should be significant when the confidence interval (CI) does not contain zero.

 As shown in Table 3, there was a significant indirect effect of social image congruence → brand attachment → brand loyalty (β = .18; 95% CI = .10 to .26; p < .001). The direct path of social image congruence to brand loyalty was rendered insignificant through the strong mediating role of brand attachment. This indirect-only mediation supports the role of brand attachment as a mediator of the relationship between social image congruence and brand loyalty. The mediation analysis demonstrated significant indirect effects of self-image congruence → brand attachment → brand loyalty (β = .21; 95% CI = .09 to .37; p < .001) and functional image congruence → brand attachment → brand loyalty (β = .07; 95% CI = .01 to .13; p < .01). Also, direct paths from self-image congruence to brand loyalty and functional image congruence to brand loyalty were statistically significant (p < .001). These partial mediations indicate that the mediating role of brand attachment is supported by existence of other potential mediators in the direct paths of self-image congruence and functional image congruence to brand loyalty.

Table 3. Testing for mediation

Notes: 5,000 bootstrap samples; All estimates are standardized; CI = confidence interval; **p<.01;***p<.001

 Multiple-group analyses

 Hypotheses 5, 6, and 7 posit that product type (functional versus ex-pressive) moderates the relationship between image congruence and brand attachment. To examine this moderating effect, we divided the pooled data into separate covariance matrices for functional (i.e., laptop computer, n = 229) and expressive (i.e., designer wallet, n = 174) product types. We conducted a multi-group approach for testing moderating effects within SEM estimates (Hair et al., 1998). We first generated the base model where all paths are free to vary across two groups. While the base model was run simultaneously, each constrained estimate of gamma (i.e., path from an exogenous variable to an endogenous variable) and beta (i.e., path between endogenous variables) was tested individually for equivalency by fixing each path coefficient in one group to be equal to the other one by one (Hair et al., 1998).

 Next, we conducted a chi-square difference test to examine the path coefficient differences between functional and expressive product types. If a χ²difference (Δχ² with 1 degree of freedom is greater than the critical value being ±3.84 at the .05 significant level, then the oderating relationship is supported (Shanahan, Christopher, Carlson, & Raymond, 2012). As shown in Table 4, there were significant differences (Δχ²(df =1) > 3.84) in two path coefficients: self-image congruence → brand attachment, and functional image congruence → brand attachment. Therefore, the moderating effect of product type exists for both self-image congruence and functional image congruence, supporting H5 and H6. It should be noted that the path coefficient from self-image congruence to brand attachment increased from .37 (laptop computer, t = 2.77, p < .05) to .66 (designer wallet, t = 4.20, p < .05), while the path coefficient from functional image congruence to brand attachment decreased from .20 (laptop computer, t = 3.08, p < .05) to -.01 (designer wallet, t = -.16, p = .88). However, the Δχ² value associated with the relationship between social image congruence and brand attachment (Δχ²= 2.9) was not statistically significant, not supporting H7.

Table 4. Moderated relationships

Notes: 5,000 bootstrap samples; All estimates are standardized; CI = confidence interval; **p<.01;***p<.001

Discussion and implications

  The theoretical importance of this study lies in examining the image congruence framework in a fine-grained manner. First, this study provides empirical support for the image congruence research by examining a multidimensional approach (i.e., self, functional, and social image congruence) and their impacts on brand attachment and loyalty. Particularly, while prior studies show inconsistent results on the relationship between social image congruence and attitudinal/behavioral responses such as product preference and store loyalty (e.g., Han & Back, 2008; Maheshwari, 1974; Samli & Sirgy, 1981), this study finds positive relationships among image-congruence constructs (including social congruity), brand attachment, and brand loyalty.

  Second, while previous research has mainly focused on how self-image congruence works in creating brand loyalty (e.g., Kressmann et al., 2006; Sirgy et al., 2008), this study expands on consumers’responses to brands by examining both emotional responses (i.e., brand attachment) and behavioral responses (i.e., brand loyalty). Specifically, this study examines whether the emotional attachment mediates the relationship between image congruence and brand loyalty. The results of hypotheses and mediation testing confirmed the mediating role of brand attachment on the influence of image congruence on brand loyalty.

 Third, the current study finds that a product type (functional versus expressive) moderates the relationship between image congruence and brand attachment. In prior studies, moderators of the relationship between self-image congruence and brand loyalty have included consumers’individual characteristics (Malär et al., 2011); customer involvement and customer awareness (Sirgy et al., 2008); and knowledge, prior experience, involvement, and time pressure (Sirgy et al., 2000). Given the level of difficulty in incorporating a wide range of consumer characteristics in building mass branding strategies, this finding provides helpful advice to brand managers for developing strategies geared toward specific product types, which are more practical and tangible than varied consumer characteristics.

 Our findings provide important implications to marketing practitioners as well. The finding that self, functional, and social image congruence positively influence brand loyalty through brand attachment provides insightful tips on brand positioning strategies. As recommended by Bhat and Reddy (1998), brands can be positioned on both symbolic and practical prob-lem-solving features. Marketers need to find strategic ways to enable consumers to feel a greater similarity between their image and the brand’ symbolic or functional image. In particular, strong overall effects of self-image congruence and social image congruence on brand attachment support the suggestion of Malär et al. (2011) that a consumer-centric approach in building brand personality is superior to a corporate-oriented approach. The consumer- centric approach may include the use of product endorsers or user imagery (e.g., cognitive age, appearance, personality, or social standing) that reflects target consumers’self- or social image; the strategic use of social media to gather information on target consumers’lifestyle and their reference groups to better understand their self and social image; and the use of storytelling about the brand with social media advertising and lifestyle marketing to better communicate the brand personality to target consumers.

  Another important managerial implication can be drawn from the moderating effect of product type (functional versus expressive) on the relationship between image congruence and brand attachment. For the expressive product, emotional brand attachment was more influenced by self-image congruence and social image congruence than by functional image congruence. This finding reinforces the importance of the consumer-centric approach in building brand personality for expressive products. For functional products such as laptop computer, all three dimensions of self-image congruence constructs have important effects on consumers’ emotional brand attachment: self-image congruence, functional image congruence, and social image congruence. Contrary to previous research suggesting that functional image congruence is more predictive of brand attitude than self-image congruence (Johar & Sirgy, 1991; Sirgy & Johar, 1999), the current study finds that self-image congruence and social image congruence are stronger predictors of brand attachment and loyalty than functional image congruence. Moreover, a non-significant effect of product type as a moderator on the relationship between social image congruence and brand attachment in the study illuminates that the influential power of social image congruence on brand attachment is similar between functional and expressive products. These findings provide valuable insights to brand managers of functional products. Although marketers of functional products still need to communicate with their consumers through utilitarian benefits such as price, quality, performance, and promotion, understanding how these products match social image of their target consumers is equally important. This means that marketers of functional products should communicate intangible attributes such as symbolic meaning and experiential value to their target consumers and promote sense of belongingness through the brand community with innovative products, dedicated service, and customer relationship (e.g., Apple®, Nespresso®).

Limitations and suggestions for future research

 Several limitations can be addressed from this study. Use of online con-sumer panelists in this study may raise concern regarding sampling bias. Although previous studies (Duffy, Smith, & Terhanian, 2005) indicate that online panel and traditional methodologies show equivalent results, caution must be taken in generalizing these findings to the population of all consumers. Another limitation is that we used only one product for each category to test the moderating effect. Thus, the results cannot be generalized to all products in the category. Further research can attempt to use a wide range of products for each category. Our findings also indicate that self-image congruence and social image congruence generally have stronger effects than functional image congruence on brand attachment and that the role of social image congruence for functional products is as important as for expressive products. However, one might argue that the impact of social image congruence can be influenced by other product/brand characteristics such as social visibility (i.e., publically visible/privately used products), brand reputation, and brand prominence (i.e., prominence of a brand’s identifying marks on the products). Thus, further research might investigate how these variables moderate the relationship between image congruence and brand attachment.

 Another future research interest is related to mediation in the relationship between image congruence and brand loyalty. According to the results of this study, brand attachment fully mediated the impact of social image congruence on brand loyalty. However, the partial mediation of brand attachment in the impact of self-image congruence and functional image congruence on brand loyalty suggests that there might be other mediators such as brand love and brand relationship quality. Identifying and examining other potential mediators in the relationship between image congruence and brand loyalty would help us comprehend the relationships among image congruence, emotional responses, and behavioral responses to brands. Lastly, “self-congruity bias” (Sirgy et al., 2000, p. 135) is another future research interest. Sirgy et al. (2000) suggest that consumers who experience high self-congruity tend to favorably process utilitarian product attributes. Thus, investigation of the direct and indirect effects of self-image congruence on brand attachment and loyalty through functional image congruence can further expand our knowledge on image congruence research.