The role of human brands in consumer attitude formation: Anthropomorphized messages and brand authenticity

Abstract The purpose of the study attempts to assess consumer reactions by classifying anthropomorphized messages according to the authenticity of a human brand with which they are associated. And also the study to investigate the causal relationship between attributions of human brand and consumer–brand relationships, brand attitudes, and purchase intention. An experimental design was used to test the hypotheses. To examine factors that can help brands achieve marketing goals through the use of human brands, this study developed research hypotheses and a structural model by focusing on the consumer information-treating process based on the authenticity of a human brand and message type. The results indicate the existence of a between-group difference with respect to anthropomorphized messages as opposed to non-anthropomorphized messages as well as levels of brand authenticity. An interaction effect was observed for the consumer–brand relationship and attitudes toward a brand. In the main effect analysis, anthropomorphized messages induced a stronger consumer–brand relationship, more positive attitudes toward a brand, and stronger purchase intention for a brand. The results of a path analysis indicated that three dimensions of human brand attributions positively affect consumer–brand relationships. This study provides a new approach to the use of anthropomorphized messages by focusing on the relationship between advertising models and messages. As a result, positive effects can be expected as identification with an advertising model increases.


PUBLIC INTEREST STATEMENT
Given the explosion of user engagement with social and mass media, Twitter, Facebook, Instagram and other social networking sites have become ubiquitous. Now, an ordinary person can develop her own human brand by increasing her intangible value through social media. As such, a human or personal brand can be seen as a broader concept that pertains to celebrities who cultivate public images or public personae to sell themselves or products and services. In particular, celebrities can propagate and deliver their own unique information more favorably than an ordinary person as they can use mass media as a base. Thus, mass media play a very important role for celebrities. Personal brand positioning can be formed by self-presentation, non-verbal tools, and verbal expression in a way that informs others of one's personal capacity, character, and abilities. This study provides a new approach to the use of anthropomorphized messages by focusing on the relationship between advertising models and messages.

Introduction
Celebrities often promote a company or a brand to establish the brand's presence in various fields (e.g., entertainment, sports, advertising). As brands struggle to differentiate themselves in a crowded marketplace and advertising becomes increasingly important, the dependence of brands on news about celebrities or their images has increased as well. As the presence of celebrities in advertising grows, it can seem natural to regard celebrities as brands in themselves (Centeno & Wang, 2017;Dwivedi & Johnson, 2013;Eren-Erdogmus et al., 2016;Freire et al., 2018;Lunardo et al., 2015;McCormick, 2016;Min et al., 2019;Muda et al., 2014). Given the explosion of user engagement with social and mass media, Twitter, Facebook, Instagram and other social networking sites have become ubiquitous. Now, an ordinary person can develop her own human brand by increasing her intangible value through social media. As such, a human or personal brand can be seen as a broader concept that pertains to celebrities who cultivate public images or public personae to sell themselves or products and services. In particular, celebrities can propagate and deliver their own unique information more favorably than an ordinary person as they can use mass media as a base. Thus, mass media play a very important role for celebrities (Min et al., 2019;Muda et al., 2014;Rindova et al., 2006). Personal brand positioning can be formed by self-presentation, non-verbal tools, and verbal expression in a way that informs others of one's personal capacity, character, and abilities (Labrecque et al., 2011;Min et al., 2019;Muda et al., 2014).
Many studies regarding the role of celebrity models in advertising have shown that celebrity endorsers have favorably influenced important advertising effectiveness measures such as consumer attitude formation (La Ferle & Choi, 2005;Min et al., 2019;Muda et al., 2014). Several studies have reported that the effects of celebrity models vary according to several factors, including product involvement, consumers' personal characteristics, and messages, rather than depending simply on harmony between a model and a product image (DeSarbo & Harshman, 1985;McCormick, 2016). The use of a celebrity model does not always produce a desirable outcome, however, and advertisements can be effective only when the products being advertised match the image of the model (DeSarbo & Harshman, 1985;McCormick, 2016). Several studies have suggested that a celebrity model's reliability and attractiveness is important in the formation of as well as changes in acceptors' attitudes (Chu et al., 2019;Delbaere et al., 2011;Fleck et al., 2014).
Studies have shown that the effects of celebrity models can vary depending on certain conditions. Considering that factors such as consumers' personal characteristics and product involvement cannot be controlled or selected in an advertisement, the model and the associated message will be the most critical factor in establishing a successful celebrity-model advertising strategy. Harmony between a celebrity model's attributes and a brand's message plays an important role in advertising. Still, such a human or personal brand is already perceived as an independent brand by consumers. The message a brand hopes to communicate can be the core factor in building an advertising strategy that implements a human or personal brand. Despite the critical relationship between a human or personal brand and a corporate brand's message, previous studies have focused mainly on such variables as attachment and performance. Some studies also address attributes associated with human brands.
Thus, the purpose of this study attempts to assess consumer reactions by classifying anthropomorphized messages according to the authenticity of the human brands with which they are associated. This paper investigates how an existing human brand as perceived by consumers affects the message an anthropomorphized symbol communicates and consumer attitudes toward the associated brand. This study also examines interactive effects between anthropomorphized/non-anthropomorphized messages and the perceived authenticity of human brands.

Human brands
The largest difference between human brands and ordinary product or service brands is that the subject of the brand is a human being, an individual person. In existing studies on brand theory, inanimate objects that comprise brand images are often personified and given personalities or identities that make it possible to form relationships with consumers. Meanwhile, a human or personal brand is perceived as a single human being or personality who does need to be anthropomorphized (Lunardo et al., 2015;McCracken, 1989). Thomson (2006) was the first to propose an academic approach to the concept and characteristics of human brands. He argued that a person can become a single independent brand and defined human brands as celebrities whose influence is differentiated from that of an ordinary person based on popularity. Celebrities, as subjects of marketing, interpersonal, or inter-organizational communications, have been referred to as human brands or person brands (Close et al., 2011;Lunardo et al., 2015;Parmentier et al., 2013).
From the perspective of brand management, providing consumers with information about a product is essential to building a single product brand. This process uses diverse factors associated with a brand, such as the product's function, consumer interest in the product, and devising a brand name that captures the characteristics of the product. Keller (2008) included human beings within the range of branding in a theory of strategic brand management on the assumption that people can themselves become brands. According to Keller, a human brand can project a clear brand image much as a product can be associated with a brand image insofar as image and awareness are very important success factors for human brands.
Some regard the 1952 presidential election in the United States as the first election that involved the application of modern marketing and branding strategies. Reeves (1961) positioned the presidential candidate Dwight Eisenhower as a straightforward, strong, and familiar leader based on his own experience as an observer of Eisenhower's leadership during World War II, which arguably led to Eisenhower's winning the vote. It was a brand-creation success story that described the American president as a legendary figure. The phenomenon of celebrity branding only grew in prominence afterwards. In particular, the phenomenon became prominent in entertainment with the onset of the era of mass media (Chu et al., 2019;Rein et al., 2005). Woischwill (2003) and Behrendt and Panetta (2003) investigated how traditional brand theory can be applied to the images of the German writer Goethe and the pop musician David Bowie (Ilies, 2017). Beyond those examples, branding studies involving people from a wide range of occupations have included soccer players (Centeno & Wang, 2017;Chadwick & Burton, 2008;Dwivedi & Johnson, 2013), corporate chief executive officers (Casanova, 2004), marketing experts (Shepherd, 2005), and newly employed college professors as subjects (Bendisch et al., 2007;Close et al., 2011;Lunardo et al., 2015;Parmentier et al., 2013). The phenomenon of celebrity branding has therefore involved a variety of fields including politics, sports, entertainment, and economics (Centeno & Wang, 2017;Dwivedi & Johnson, 2013;Eren-Erdogmus et al., 2016;Freire et al., 2018).

Messages using Anthropomorphized symbols
Many studies of consumer decision-making have found that consumer evaluations of and decisions regarding the same information can differ depending on the types of message involved (Tversky & Kahneman, 1981). Reviewing existing studies related to anthropomorphizing products and services of the associated brands (Kim & McGill, 2011;Puzakova et al., 2013), Aggarwal and McGill (2007) examined the effects of personifying automobiles on consumer product evaluation. In their study, subjects were first exposed to a humanized message ("Hi! I am Lexus") or a nonhuman message ("You will now see a picture of a Lexus") and were shown a front view of a car with a human-like expression. Those who were given the humanized message evaluated the car differently based on the particular expression they were shown. Those who were given a nonhumanized message largely agreed in their evaluations. Miesler et al. (2011) conducted an experiment that involved presenting a front view of an actual car or the shape of the front view depicted as a baby to two subject groups and reported that the group who encountered the babyshaped car reacted more favorably to that image than did those who were shown the standard image. Kim and McGill (2011) reported that the risks involved in playing slot machines were perceived more clearly by subjects who were shown a personified slot machine than by subjects who were shown a non-personified slot machine (Puzakova et al., 2013). Connell (2013) studied anthropomorphized animals, including bears and lions, and reported that an anthropomorphized bear, which is closer in form to a human, induced more favorable reactions than a lion, which represented the non-anthropomorphized condition. Based on the effects of anthropomorphizing such diverse objects as cars, mobile phones, animals, and slot machines, or the delivery of messages within environments that are manipulated to seem homelike or personal, personified products or brands induce more favorable evaluations from consumers than nonanthropomorphized products or brands (Tam et al., 2013). Many previous studies in the marketing literature have shown that perceiving a brand as a human-like entity affects consumers' attitudes and evaluations (Aggarwal & McGill, 2007;Kwak et al., 2015;Landwehr, McGill, & Herrmann, 2011;Puzakova et al., 2013;Wan et al., 2017).
In particular, consumers who receive messages exhibit varying attitudes based on the diverse attributes, such as attractiveness and reliability, of a model who appears in an advertisement. Consumers tend to trust advice from celebrities and believe that celebrities have the authority to discuss advertised products. In general, companies prefer celebrity advertising models because they can play the role of drawing consumers' attention in an environment that often overloads consumers with information (Erdogan et al., 2001;Kwak et al., 2015Kwak et al., , 2017Landwehr, McGill, & Herrmann, 2011;Puzakova et al., 2013;Wan et al., 2017).
The effects of a message can differ depending on the type of message delivered and the level of influence a given model has on public opinion. A previous study on message types found that a message has stronger or more intense effects when the message coincides with the situation that consumers face (Aggarwal & McGill, 2007;Kim & McGill, 2011;Kwak et al., 2017;Puzakova et al., 2013;Tversky & Kahneman, 1981). This study therefore assumes that delivering a message through anthropomorphizing a product or symbol-implying that a given product has a human side-can increase the congruence between the model and the message. Humanization affects not only consumer perceptions of an advertising model; it also affects the perception of a corporate brand's assets. This study therefore suggests the following hypothesis: H1: All else remaining equal, being exposed to an anthropomorphized message will have a greater impact on consumer-brand relationships, consumer attitudes toward a brand, and purchase intention for a brand than when the message is not anthropomorphized.
Consumers who form strong relationships with human brands exercise enormous influence on the larger society and tend to form positive intentions to purchase products or services that human brands advertise or support. As a result, human brands have been recognized as an important differentiating tool in messaging to consumers (Aaker, 1997). The characteristics of a human brand can be transferred to the brand that the human brand endorses. In terms of the authenticity of a human brand, consumer perceptions of a company or brand that the human brand supports can become more positive, an effect that grows when the message type coincides with consumer dispositions (Aaker & Lee, 2000). (2007) conducted an experiment to gauge how a human brand affects consumer evaluation of a product. Using the human message "Hi! I am Lexus" in the experiment, the study found that the group who were exposed to the human message evaluated the product more positively than did those exposed to an ordinary message. Where a human brand is used as a marketing medium, congruence between the human brand and a marketing message will increase if the message delivered is personified with humanlike attributes. Some scholars have argued that personalized or anthropomorphized brands produce enhanced brand likability and favorable consumer responses as well as stronger positive emotions and more favorable attitudes toward the brands (Aggarwal & McGill, 2012;Chu et al., 2019;Puzakova et al., 2013).

Aggarwal and McGill
Following a recent trend that emphasizes corporate social responsibility and morality, brand authenticity has become an important marketing factor (Morhart et al., 2015;Moulard et al., 2016;Napoli et al. 2014;Spiggle, Nguyen, & Caravella, 2012). Brand authenticity stems from the persistent pursuit of consumer value rather than purely commercial motivations. Interpreting brand authenticity as it relates to the broader concept of authenticity, brand authenticity can be defined as the ability to deliver on business promises (Beverland & Farrelly, 2010;Dickinson, 2006;Morhart et al., 2015;Moulard et al., 2016). Brand authenticity can also be defined as a combination of internal consistency-maintaining the truthfulness and core of the brand itself-and external consistency-not presenting an exaggerated or false appearance to the public (Brown et al., 2003;Grayson & Martinec, 2004).
Brand authenticity has become a criterion consumer can use to understand and select brands, which some have argued has overtaken quality as the prevailing purchasing criterion (Gilmore & Pine, 2007;Hernandez-Fernandez & Lewis, 2019), while for marketers it has become a factor that can secure competitive advantage to enhance a company's value as well as profits. Some scholars have emphasized that authenticity has begun to interest marketers who are keen to analyze consumer preferences for authentic offerings ( Everyone searches for the meaning of life and pursues authenticity related to personal goals. People prefer brands that can strengthen the authenticity of their lives and their ideal selves (Beverland & Farrelly, 2010). Once consumers form an attachment to a certain brand, positive information is spread by word of mouth and consumers support a favored brand more strongly to maintain the positive relationship (Fournier, 1998;Yeung & Wyer, 2005). Fedorikhin et al. (2008) argued that brand preference and brand authenticity increase at the same time when consumers develop emotional fondness for a certain brand. Thus, if consumer attitudes towards a brand differ with the level of consumer attachment to the brand, it stands to reason that a brand that projects human attributes will be able to create such attachment. Consumers perceive higher brand authenticity when a brand is endorsed by a human brand to which they feel attached (Beverland & Farrelly, 2010;Fedorikhin et al., 2008;Morhart et al., 2015). Beverland and Farrelly (2010) argued that heroic celebrities who provide elements such as life value, meaning, trust, and courage influence perceptions of authenticity more strongly than celebrities whose careers depend to a greater extent on values such as fun and fantasy.
Some researchers have found that consumers with strong perceptions of authenticity reported significantly stronger perceptions of brand trust than those with weaker perceptions of authenticity, in investigating the relationship between brand authenticity and brand trust as part of a conceptual framework (Coary, 2013;Hernandez-Fernandez & Lewis, 2019). This finding suggests that consumer perceptions of a brand's authenticity are strongly associated with brand relationships as well as attitude formation. Consumer attitudes towards a company and its brand are therefore more likely to be positive as well, because consumers form positive and strong perceptions of a company or brand when they are exposed to a human brand with great authenticity and an anthropomorphized message. Hence, this suggests the following hypothesis: H2: An interactive effect between message types and the degree of human brand authenticity will affect consumer-brand relationships, attitudes toward a brand, and purchase intention for a brand.

The relationship between attributions of human brand and brand performance
The concept of brand personality was introduced and defined by Plummer in 1985 as a function of humanlike characteristics that consumers attribute to a brand (Aaker, 1997). Brand personality has been stressed as a differentiating strategy a brand can use to market a product because consumers can use it as a tool with which to express themselves. (Aaker, 1997;Swaminathan et al., 2007). When a company establishes a strong brand personality for a product or service, consumers can form and remember unique and amicable associations with the brand. This in turn creates strong emotional bonds, which in turn increase brand loyalty and trust (Johnson et al., 2000;Keller, 1993;Siguaw et al., 1999).
As brand personality is formed by personifying a brand, consumers can more easily relate a brand to a particular celebrity when the brand has a strong brand personality (Aaker, 1997;Lau & Phau, 2007). Studies have implemented Aaker's (1997) Brand Personality Scale (BPS), which measures brand personality along five dimensions, including strength, sophistication, competency, earnestness, and interestingness (Lee & Ra, 2002;Yang & Cho, 2002). Other scholars have investigated relationships between five brand personality dimensions (sincerity, ruggedness, excitement, sophistication, and competence) and attitude formation toward a brand (Sung & Kim, 2010). Another study suggests that the brand personality dimensions of sincerity and ruggedness significantly influence brand-attitude formation (Hernandez-Fernandez & Lewis, 2019). In other words, consumers can more easily identify a brand with its advertising models or celebrities when the brand personality is stronger (Chu et al., 2019). A human or personal brand, which as noted acknowledges a person as a single independent brand, turns real individuals into brands, not simply abstract entities with humanlike characteristics (Aguirre-Rodriguez, 2014; Fournier & Alvarez, 2012;Puzakova et al., 2013). Hence, the concept of a human brand differs from that of a brand personality.
From the perspective of brand communication, positive images of human brands can be transferred to corporate brands or products. In particular, once consumers become attached to a certain human brand, their positive feelings toward the brand can be transferred easily to a company and its products or services. Chu et al. (2019) emphasized that consumers may describe and assess a brand using personalities and brand character attributes and then associate the brand with human-like traits to retain the brand's image in their minds (Aguirre-Rodriguez, 2014;Fournier & Alvarez, 2012;Ingenhoff & Fuhrer., 2010). An established image is based mainly on brand personality, becoming an anthropomorphic image that consumers associate with the brand. Consumer attitudes towards a corporate brand will vary according to the degree of authenticity consumers perceive to be associated with an associated human brand that is used to market the corporate brand. Hence, this study suggests the following hypothesis: H3: The attributions of human brand attributes will affect the consumer-brand relationships, attitudes toward a brand, and purchase intention for a brand.

Pretest
The experimental stimulus used in this study was chosen by the author and two experts in business administration, who were consulted to determine the human brand that best fit the purposes of this study. A product was then chosen by considering primarily congruence with the human brand. First, the classification of the human brand was considered by expanding on the various types proposed in the literature (Sachlecht, 2003). In this study, a pretest was conducted to find human brands that fit contexts involving sports, artists, politics, and companies. The study examined the characteristics required of a human brand by referring to an existing study that suggested 20 survey questions pertaining to brand personality and chose four human brands with appropriate congruence with the four abovementioned fields. The human brands selected were then subjected to an additional pretest with undergraduate students as subjects and a sports star was finally selected for the experiment. Petrochemical and energy-sector products were chosen after considering whether questionnaire items pertaining to the human brand personality could be expressed effectively and whether the image of the product line is congruent with the chosen human brand. The experimental stimulus for the control group was an advertising message delivered by a non-celebrity model, in contrast with the celebrity of the human brand model.
A total of 70 undergraduate students who were enrolled in a major course in a department of business administration participated in the pretest. Prior to the actual experiment, pretests checked the experimental procedure, the composition of the questionnaire, and the print and color condition of the questionnaire. Among the human brands aligned with the four fields, the experimental participants chose the human brand that best matched the product where multiple responses were possible. Park Ji-sung (a famous soccer star) was finally selected as the experimental human brand stimulus. In the operational test of the human brand model and the noncelebrity model (a non-famous person), participants first read operational questions regarding the company model and product image and responded on a 5-point scale ranging from 1 (very negative) to 5 (very positive). In the t-test results, the experimental stimulus proved to have statistically significant effects. The mean value of participants' attitudes towards the human brand model was 3.71 (S.D. = .407) and the mean value of participants' attitudes towards the noncelebrity model was 2.60 (S.D. = .473), indicating statistical significance (t = 25.25, p < .001).
Because a real company was selected as the experimental subject for this study, sample bias was a possibility. To confirm the equivalency of the experiment stimuli across groups, a pretest of brand loyalty was conducted and test subjects who exhibited extreme brand loyalty were removed from the sample. Because loyalty to a focal company can affect experimental results, that exogenous variable was removed. In the pretest, three questions pertaining to brand loyalty were measured on a 5-point scale. Test subjects with a lowest score of 1 and a highest score of 5 were removed from the sample.

Experimental procedure
To examine factors that can help a brand achieve marketing goals through the use of human brands, this study developed research hypotheses and a structural model by focusing on the consumer information-treating process based on the authenticity of a human brand and the type of the associated message. The study conducted to evaluate a human brand image. Students who participated in the experiment were randomly divided into two groups and entered separate lecture rooms. A total of 308 students in a department of business administration participated in this experiment.
First of all, all participants saw two types of experimental stimulus. The stimuli with a famous soccer star who was finally selected as the experimental human brand stimulus. Another one is the non-celebrity model (a non-famous person) with same condition. The experimental participants chose the human brand that best matched the product where multiple responses were possible. In the operational test of the human brand model and the non-celebrity model (a nonfamous person), participants first read operational questions regarding the company model and product image The experiment followed a two-by-two experimental design. To prevent leakage of information about the previous experimental stimulus, two experiments were implemented simultaneously in separate spaces. There was no previous exposure to the experimental stimulus before the experiment. Random allocation was applied only in the case of message type (anthropomorphized/nonanthropomorphized) because answers to the question about the authenticity of a human brand should reflect subjects' personal characteristics. For example: An anthropomorphized message such as "I am your energy" was presented on the main page while general information about the product and company (e.g., a brief introduction to the company and a description of the product's components) was presented in the non-anthropomorphized message. A between-group design based on corporate message type (anthropomorphized vs. non-anthropomorphized) and the authenticity of the human brand (strong vs. weak) was used.

Measurement of variables
The questionnaire used in this study was divided into two versions based on corporate message type. An image representing the message type (anthropomorphized or non-anthropomorphized) was inserted at the top of each questionnaire. The questions were arranged in the order of human brand attributes, consumer-brand relationships, attitudes toward the brand, and purchase intention for the brand.
Referring to the adjectives used in the extant literature, several items regarding human brand personality were composed based on 42 specific image adjectives sorted along five dimensions that measure brand personality, as suggested by Aaker (1997). These dimensions were reliability, potency, sincerity, excitement, and sophistication. Question items for the study were composed by combining items taken from Aaker (1997) with items used in other studies (Gwinner, 1997;Lee & Ra, 2002;Yang & Cho, 2002;Sung & Kim, 2010). From a total of 57 items, two experts first removed adjectives with redundant and unnecessary meanings. A total of 20 adjectives that correspond to the image that the participants in the pretest formed of the experimental brand were finally chosen.
This study referred to a previous study that used eight items to investigate consumer-brand relationships. The questions were modified to fit this study (Fournier, 1998). The scale ranged from 1 "Strongly Disagree" to 5 "Strongly agree". Three items used in MacInnis and Park (1991) were applied to measure attitudes toward the brand. Separate scales were used to measure the three items: for the first item, 1 represents "Bad" and 5 represents "Good"; for the second item, 1 represents "Unfavorable" and 5 represents "Favorable"; for the third item, 1 represents "Unlikeable" and 5 represents "Likeable". To measure extended intention to purchase, the study recomposed questionnaire items used in the literature to fit this study and a total of four items were used (Chaudhuri & Holbrook, 2001).
Finally, the question regarding the authenticity of the human brand included a total of three items. The dictionary definition of authenticity is "true and genuine property", "based on truth", "being truthful in nature", "originality", or "genuineness". Authenticity contrasts with "duplication", "copy", or "similarity". In socio-psychology, authenticity consists of the coincidence of a subject's inner state and her behavior and is related to truth about oneself (Hernandez-Fernandez & Lewis, 2019; Jin, 2018). The questions were recomposed to fit this study and three questions were measured using a 5-point Likert scale. Items used in this study are described in the Appendix A.

Manipulation check
The study conducted t-tests to check the significance of the mean values for each group by investigating whether the experimental stimuli that consisted of exposing participants in one group to the celebrity human brand the celebrity and participants in the other group to the noncelebrity human brand were appropriately manipulated. The mean of the experimental stimulus for each group proved to be significant. The operational tests for the human brand and the noncelebrity brand yielded the following results: very positive/very negative: 3.50/1.60, t = 30.2, p < .001. The results of the operational test for anthropomorphized and non-anthropomorphized messages about the company and the brand image are as follows: anthropomorphized/nonanthropomorphized: 3.69/2.16, t = 35.1, p < .001.

Results for degree of authenticity of human brands and Anthropomorphized messages
For hypotheses 1 and 2, the study then analyzed the effectiveness of the sub-factors comprising each variable. And also the study conducted Multivariate Analysis of Covariance (MANOVA) using pre-test scores as the covariate. A MANCOVA tests differences in the values of the dependent variable. Analysis of the between-group differences between the dependent variables based on exposure to the anthropomorphized message yielded a Wilks's λ = .901 (F = 13.753, p = .000). Analysis of between-group differences according to level of authenticity yielded a Wilks's λ = 821 (F = 27.161, p = .000). Interaction analysis yielded a Wilks's λ = .876 (F = 3.122, p = .026). The dependent variables, including consumer-brand relationships, attitudes toward the brand, and purchase intention for the brand, showed significant differences at the .05 level (see Table 1). Table 2 below presents the results of the main effect analysis. In the analysis the effects of exposure to the anthropomorphized message, a significant between-group difference was observed for consumer-brand relationships (F = 18.10, p = .000), attitudes towards the corporate brand (F = 22.34, p = .000), and intention to purchase the brand (F = 20.91, p = .000). The results of the analysis of the level of brand authenticity showed a significant between-group difference for consumer-brand relationships (F = 71.28, p = .000), attitudes toward the brand (F = 21.33, p = .000), and purchase intention for the brand (F = 34.37, p = .000). An interaction effect was observed between the consumer-brand relationship (F = 3.784, p = .052) and attitudes toward the brand (F = 3.050, p = .082), but no interaction effect was observed with purchase intention (F = .743, p = .389).
That is, the dependent variables-consumer-brand relationships, attitudes toward the brand, and purchase intention for the brand-showed significant differences at the .05 significance level. However, analysis of interactions showed insignificant results, the dependent variables-purchase intention for the brand-showed insignificant differences at the .05 significance level. Thus, hypotheses 1 and 2 were supported. However, results pertaining to interaction between variables indicated that only intention to purchase (F = .743, p = .389) has an interaction effect. The remaining variables showed insignificant interaction effects.

Exploratory Factor Analysis of Human Brand Attributes
For hypotheses 3, the study then analyzed the causal relationship between attributions of human brand and brand performance such as consumer-brand relationship, attitude toward the brand and purchase intention by using SEM (structural equation modeling). To test the structural relationships in the model, the hypothesized causal paths were estimated.
It is necessary to check the dimensions of the attributes of a human brand for reliability. In this study, reliability was checked by implementing exploratory factor analysis (EFA) using the measured items, which is known as common method variance. The extent of spurious covariance was shared among the variables because of the common method used in collecting data. Thus, this study checked for possible common method variance using Harman's single-factor test (Liang et al., 2007;Podsakoff et al., 2003) for (in)dependent variables. According to this approach, common method variance is present if a single factor accounts for most of the covariance in the dependent and independent variables. This study found no dominant factor emerging from factor analysis, implying that common method variance was not a serious problem. After collecting and cleaning the data, verification tests were conducted to determine the measurement model's validity.
Cronbach's alpha based on internal consistency was computed to assess the reliability of individual items and total items. As seen in Table 3, the Cronbach's alpha value was above 0.7 for all factors, satisfying the reliability of the items that belong to each factor. Multicollinearity can be measured by tolerance and the variance inflation factor (VIF). In general, the criterion for determining collinearity is tolerance below .10 and a VIF over 10. In the test results for this study, tolerance was .519 and the VIF was 1.695, indicating the absence of a multicollinearity problem.
To test the validity of the sub-question items for the brand image attributes, factor analysis was implemented using the VARIMAX rotation method. The criterion for removal was having an eigenvalue over 1 and standardized regression weights of measurement items below 0.5. The results of a Kaiser-Meter-Olkin Test was .789 and of a Bartlett's sphericity test was χ 2 = 681.5, indicating significance. This confirms that there is no problem in conducting a factor analysis. In the results for brand personality, a total of three factors were extracted, which explained 57.3 % of the variance. As shown in Table 2, the factor loadings of the items in the measures range from 0.602 to 0.878, demonstrating convergent validity at the item level. Note: Interaction: anthropomorphized*authenticity; ** p < .05, ***p < .001 Among the brand image attributes, the reliability of a total of 20 question items was .881, which is a satisfactory level. EFA revealed three factors with eigenvalues greater than 1.00. This analysis suggested that the data sample used in this study was unlikely to have been contaminated by common method bias (Podsakoff et al., 2003). In the results of the EFA, human brand personality exhibited three dimensions. Specifically, for factor 1, positivity, youth, coolness, extroversion, trend, passion, activeness, trust, familiarity, affectionateness, and interest constituted a "dynamics" dimension. For factor 2, happiness, softness, attractiveness, success, and honesty constituted an "attractiveness" dimension. For factor 3, boldness, masculinity, leadership, and technology constituted a "leadership" dimension (see Table 3).
A confirmatory factor analysis (CFA) was conducted to test whether validity was close to the standard for CFA, where x 2 is 513.53(df = 432), p = .000, CFI = .952, GFI = .890, AGFI = .847, NFI = .890, and RMSEA = .057. As shown in Table 4, discriminant validity was assessed by comparing the correlations of components with average variance extracted (AVE). The final indicator of convergent validity is AVE, which measures the extent of variance captured by a construct in relation to the extent of variance that is attributable to measurement error. The AVE for this study also satisfies the standard of 0.5, which means that the measurement indexes exhibit convergent validity. The AVE falls between .519 and .687, and the means of the squares of the correlation coefficients fall between .018 and .212, which indicates that the AVE is higher than the means of the squares of the correlation coefficients (r 2 ). This also satisfies the requirements for establishing the discriminant and convergent validity for the purposes of research hypothesis model verification.

Additional Analysis
The study explored the differences between the brand image variable attributes according to the two types of exposure. And then paired t-tests were conducted to compare the difference. The average value of the brand image attributes decreased for 11 items with exposure to the human brand compared with the case of exposure to the non-celebrity brand. Among the components of the brand, affectionateness, happiness, interest, youth, positivity, technology, attractiveness, passion, extroversion, activeness, and masculinity decreased statistically significantly. Meanwhile, honesty, familiarity, trend, trust, leadership, success, coolness, softness, and boldness showed statistically insignificant differences (see Table 5).

Discussions and conclusion
The study examined the effects of human brand attributes on consumer-brand relationships, attitudes toward a brand, and purchase intention for a brand. This study also examined interaction effects between anthropomorphized or non-anthropomorphized messages and the authenticity of a human brand. Of the 20 items comprising the brand attributes construct, the average values of 11 of the items decreased when exposed to the human brand but did not decrease when exposed to the non-celebrity brand. Among the components of brand image, the average values for affectionateness, happiness, interest, youth, positivity, technology, attractiveness, passion, extroversion, activeness, and masculinity were higher when exposed to the human brand. These results indicate that three dimensions of human brand attributes are closely related to consumer perceptions of a brand and attitude formation, namely consumer-brand relationships, attitudes toward a brand, and purchase intention for a brand. The influence power of the attractiveness dimension is slightly higher than that of the dynamic and leadership dimensions for building consumer-brand relationships. The attractiveness dimension included happiness, softness, attractiveness, success, and honesty. Based on the study results, it appears that consumers consider the attractiveness dimension more deeply when they are building relationships with a brand. The MANCOVA results indicate the existence of a between-group difference with respect to the anthropomorphized as opposed to the non-anthropomorphized message as well as the level of brand authenticity. An interaction effect was observed for consumer-brand relationships and attitudes toward the brand. In the main effect analysis, the anthropomorphized message induced stronger consumer-brand relationships, attitudes toward the brand, and purchase intention for the brand. The group with high brand authenticity formed stronger consumer-brand relationships, attitudes toward the brand, and purchase intention for the brand. An interaction effect of anthropomorphized messages and brand authenticity on consumer-brand relationships and attitudes toward the brand was observed, but no such effect was observed on purchase intention for the brand.
This study's results imply that consumer-brand relationships, attitudes toward a brand, and purchase intention for a brand are stronger when consumers are exposed to anthropomorphized messages. The study also shows that perceptions of a corporate brand vary according to whether a message is anthropomorphized. The study's results also suggest that anthropomorphized products or messages elicit more favorable evaluations from consumers than nonanthropomorphized products or messages. This result can be interpreted to imply that consumer attitudes towards human brands as well as their attitudes towards corporate brands and the intention to purchase such brands are more positive when consumers are exposed to anthropomorphized messages than when they are exposed to non-anthropomorphized messages. This result indicates that consumers evaluate a brand and product more positively when presented with a positive message like "I am your energy" than when presented with a non-celebrity brand or a non-anthropomorphized message.
The study examined consumer-brand relationships when exposed to human brands as well as consumer attitudes toward a brand and their purchase intention for a brand when the authenticity of the associated human brand is high. As is the case with attachment, a human brand with high authenticity strengthened the relationship with the human brand and it also played a positive role in shaping attitudes towards the brand and intention to purchase the brand. These results indicate that consumers form positive relationships with a human brand that has high authenticity and they develop an attachment to it that helps to sustain the consumer-brand relationship. The most important attributes of the human brand image found in this study are affectionateness, happiness, interest, youth, positivity, technology, attractiveness, passion, extroversion, activeness, and masculinity. Consumers perceived the human brand as having an image that was simultaneously strong and soft.

Implications of the study
This empirical study of the effects of human brands on consumers provides several implications. First, the study tested the effects of anthropomorphized messages by expanding the range of application to advertising or marketing by relating the practice of anthropomorphizing messages to human brands. This study provides a new approach to the use of anthropomorphized messages by focusing on the relationship between advertising models and messages. As a result, a positive effect can be expected as identification with an advertising model increases. The results of this study can be interpreted to imply that an anthropomorphized message is more effective than a general message when companies deliver advertising messages to consumers. The study also indicates that the combined use of human brands and anthropomorphized messages can be effective in forming consumer relationships with a brand. This study result should be widely used in advertising as marketers realize the necessity of effectively using anthropomorphized messages.
The empirical findings that anthropomorphized brand messages and celebrity human brands positively affect consumer brand attitudes and evaluation, including consumer-brand relationship building, brand attitudes, and purchase intention for a brand, suggest that adopting strategies that facilitate the perception of a brand as a human-like entity should inform brand communication strategies. Specifically, human brand attributes are directly related to consumer-brand relationships, which in turn contribute significantly to brand awareness and brand loyalty. By deploying anthropomorphized messages and celebrity humans in their brand positioning and communications, brand communication practitioners can enrich consumer-brand interactions while enhancing brand-attitude formation.
Previous studies of human brands have focused on attachment to such brands. Studies of advertising models mainly stress the maximization of the effects of such advertising by investigating such attributes of models as reliability, attractiveness, and professionalism. On the other hand, this study is meaningful as a new variable, authenticity, was found to be an important attribute of a human brand. The study implies that sentimental and emotional relationships can be formed between human brands and consumers. As such, authenticity should be considered an important factor when building a strategy for human brand management.
The study identifies attributes of a human brand image through an empirical study by applying the concept of a human brand to brand personality theory. The attributes of a human brand image fall along three dimensions: dynamics, attractiveness, and leadership. Specifically, dynamics can be decomposed into positivity, youth, passion, affectionateness, and interest. Attractiveness can be decomposed into happiness, softness, attractiveness, and honesty. Finally, leadership can be decomposed into boldness, masculinity, and leadership. The attributes of the human brand image applied in this paper can contribute to building a theoretical basis for research on human brand personality. The study results could be used as important basic data in future research on human brand personality.
The study finds that the authenticity of a human brand and an anthropomorphized message positively influence the relationship between a human brand and consumers, the formation of attitudes toward the brand being advertised, and purchase intention for the advertised brand. This study result suggests that a consumer's image of a certain brand not only determines the level of satisfaction with the consumerbrand relationship but also encourages the formation of a positive attitude towards the brand and the intention to buy products from the company. This study confirmed that a human brand can have a powerful impact on consumers through its own value and personality, rather than by simply playing a role as an advertising model. In that way this study provides theoretical implications by proposing a new direction for research on human brands.
The study's analysis of the consumer information-treating process in a structural equation model using a human brand as the medium also implies that a human brand can be a very important brand asset. The study also provides practical guidelines for establishing a strategy aimed at increasing the value of a human brand. The study suggests that finding a human brand that can express a company's image in a positive and favorable way can be a very important component of a successful marketing strategy.

Limitations
Among the limitations of this study is that the research subject, which operates in the petrochemical and energy sector, was chosen primarily by considering congruence with the human brand image that was used. Here, engagement with the experimental participants was not applied in the selection process, but it is known that evaluation or attitudes differ according with the level of engagement subjects have with a company or brand. Future studies should choose products by applying the personal characteristics of experimental participants or using product groups that are divided according to levels of engagement among research subjects.
Another limitation is that diverse types of human brands were not applied. The human brand used in this study was chosen from the pretest and other types of human brands could not be investigated. To be able to generalize findings pertaining to the effects of a human brand, more diverse types of human brands should be included in future studies.
Finally, the experimental group used in this study included only college students, inevitably resulting in bias in terms of age and occupation. To generalize the results of the study, experimental subjects representing more diverse ranges of occupations and ages should be incorporated in future studies. Moreover, the personal characteristics of experimental participants were not considered in this study. Consumers these days are becoming more and more individualized, suggesting that we are living in an era that stresses individuality. In this situation, consumer attitudes or product and brand evaluations can differ according to the level of engagement a consumer experiences with a company or brand. In future studies, reactions to and evaluations of a stimulus should be compared by reference to personal characteristics such as personality and psychological characteristics.