Positive and Negative Behaviors of Oppositional Loyalty in Online Communities

Based on the theories of moral identity, community engagement, emotional attachment, and brand loyalty, this study aims to explore oppositional loyalty’s driving factors and dimensional structure (the willingness to pay a price premium, oppositional referrals, schadenfreude, and trash-talking). Our samples comprised 533 members who contributed more than three posts monthly in Xiaomi’s mobile online communities. The structural equation modeling method was utilized to test direct effects, and hierarchical regression was adopted to verify the moderating relationship. The findings show that community engagement and brand attachment positively affect oppositional loyalty, and brand attachment partly mediates the influence of community engagement on oppositional loyalty. Furthermore, moral identity positively moderates the impacts of community engagement and brand attachment on oppositional referrals. On the contrary, moral identity negatively moderates the relationships between community engagement and trash-talking as well as between community engagement, brand attachment, and schadenfreude. Negative effects on competing brands are harmful to favored brands through the moderating effects of moral identity. Enterprises and online communities should discard the idea that more oppositional loyalty is ideal and correctly guide members’ positive and negative behaviors.

The associate editor coordinating the review of this manuscript and approving it for publication was Xiping Hu.
Researching various industries simultaneously, scholars have explored the confrontational characteristics and behavioral performance of oppositional loyalty. In terms of the specific objects of confrontation, competing brand usually refers to those with similar prices, products or services, and are direct rivals of customers' preferred brands [5], [16]. From the intensity of antagonistic interaction, antagonistic behavior can be divided into two forms: monologues and dialogues [17]. From the perspective of oppositional content, it involves the advantages and disadvantages of the brand, products, corporate strategy, market share, development prospects, and marketing ethics [18], [19]. From the expression of hostile emotions, there are pleasures and outrages: pleasures are manifested in online community members ''ridicules, jokes, schadenfreude'' [20], and outrages mean hatred, anger, diatribes [8], [17]. Concerning the nature of the confrontation, positivity signifies the evaluation and selection of competing brands based on objective facts [1]; negativity refers to trash-talking, insult, slander, or profanity without competitive brand experience [8], [20].
With the rapid development of new social media, antagonistic loyalty is frequently observed in online communities. Unethical behaviors against competing brands are more popular than ever before, such as schadenfreude, trash-talking, hate, and desire to harm [21], [22]. However, no scholar has studied the influence of moral identity or business ethics on oppositional loyalty, and the structural dimension of adversarial loyalty has continued to be a mystery. Besides, most of the existing literature has been qualitative research on oppositional loyalty [1], [6], [8]. There have also been problems, such as narrow research perspectives and incomplete driving variables in influencing factors.
Community engagement is one of the essential driving factors of oppositional loyalty [2], [9], [23], and is closely related to the positive psychology and behavior of online community members, such as brand loyalty, identification, trust, and satisfaction [24]. However, community relationships' transformation into a brand relationship often requires community members' brand attachment or commitment as an intermediary [25], [26]. Therefore, it is necessary to explore further the causal relationship between community engagement, brand attachment, and confrontational loyalty. Based on 533 samples, this study constructs an oppositional loyalty theoretical framework, deeply analyzes the formation mechanism of antagonistic loyalty behaviors, and contributes to the literature on brand loyalty. The practical value is to clarify the misunderstanding that companies and online communities believe more oppositional loyalty behaviors are better than few. The negative impacts on competing brands ultimately produce harmful effects on favorite brands through moral identity.
The remainder of this article is organized as follows. First, the literature review covered antagonistic loyalty, community engagement, brand attachment, and moral identity. Second, the relationship hypotheses and research model were proposed. Third, we introduced measurement development, data collection, and sample characteristics. Next, the results of the data analysis were reported and discussed. Finally, we concluded the theoretical and practical implications along with future exploration directions.

II. LITERATURE REVIEW A. OPPOSITIONAL LOYALTY
The existence of oppositional loyalty in online communities has received more attention from academics [1], [23] ever since Muniz and Hamer [5] first put forward this concept. Although the favored brand products or services are objectively inferior to competitive brands, community members are still willing to buy or use preferred brand products out of adversarial loyalty [10]. Oppositional or confrontational brand loyalty is ''a psychological phenomenon observed among members of a brand community who hold negative and opposing views about rival brands, and even exhibit antagonistic behaviors towards those brands'' [16] (p. 254). In the online community context, for competitive brands, ''opposition'' represents a circumstance that induces adversary, antagonism, rejection, hostility, and attack, which evolves via comparisons or contrasts. Obviously, the essence of oppositional loyalty includes the confrontation against competitive brands and the loyalty with favored brands.
Oppositional loyalty has two aspects. From one perspective, the equity of the preferred brand increases additional cash flow, in large quantities or repeated purchases, positive word of mouth, and energetical recommendations. From another perspective, it is a liability for the competing brands, including rejected purchases, negative word of mouth, boycotting products, and warning consumers against the brands [6]. Similarly, Kuo and Hou [16] and Liang et al. [7] found that oppositional loyalty is demonstrated in two facets: consumers or online community members purchased favored-brand products tautologically, in large numbers and at a price premium; on the contrary, they rejected to try, recommend, and buy any products of competing brands, even though these products were considered better.
Previous research indicates close relationships among brand commitment, word-of-mouth, willingness to pay a price premium, and oppositional brand loyalty [27]. As a critical measure of brand loyalty and value, the willingness to pay a price premium signifies that those consumers are volunteers to pay more for their favored brand, resulting in sustainable advantage and higher profitability [28]- [31]. Moreover, oppositional referrals, like oppositional loyalty, indicate negative attitudes and behaviors toward brands that customers pour attention into the competition with their favored brand [32]. By the same token, the focal brand loyalty strikingly impacts oppositional referrals (e.g., desire to harm or trash-talking) against the competitive brand [21].
From the negative effects' angle, oppositional loyalty comprises lots of antagonizing activities of competitive brands, for instance, disturbing the work in online communities, defaming its products or customers, and demonizing rival brand as an ''enemy brand'' [6], [7], [33]. In detail, Japutra et al. [34] deemed that oppositional loyalty consisted of schadenfreude and trash-talking. In online communities, these two dimensions (and intergroup stereotypes) represent the dark side of confrontational loyalty [20]. Trash-talking is chosen by customers who slander one brand to benefit another brand without having experienced it. Schadenfreude indicates a kind of malicious joviality when rival brands or their consumers have misfortunes or failures [22]. The prior study reveals that adversarial loyalty jeopardizes the rival teams' sponsors and fans, and schadenfreude mediates the influences of fan identification and brand community identification on oppositional loyalty [12].
Apart from the above findings, knowledge about antagonistic loyalty is lacking in the following respects. First, the structural dimension of oppositional loyalty remains unknown. Relevant literature has divided the concept into individual and group antagonistic loyalty or four aspects: absolute oppositional loyal, defender, hidden oppositional loyal, and skeptical loyal [35]. Nevertheless, previous studies have not fully illuminated the essence of ''antagonism'' concerning competitive brands or reflected on rival brands' negative effects. Second, many studies have described confrontational loyalty as a qualitative concept [18], [34]. Other shortcomings include limited research perspectives and incomplete driving variables as influencing factors [9]. Finally, during the short term, confrontational brand behavior can improve the preferred brand's financial performance. However, in the long term, the adverse influence of the competing brand will negatively affect the favorite brand through incompatibility with marketing morals.
In summary, we define antagonistic loyalty as the attitudes and behaviors of consumers who express loyalty to their favored brand by opposing the products, customers, fans, online communities, companies, and even sponsors of the competing brands. In online communities, the term comprises four dimensions that affect the preferred brand and the rival brand. The willingness to pay a price premium and oppositional referrals benefit the favored brand, while schadenfreude and trash-talking negatively influence the rival brands.

B. COMMUNITY ENGAGEMENT
Kahn [36] proposed the term ''engagement'' when he was studying the working environment. Since then, scholars have conducted deeply relevant explorations in various fields, including pedagogy, psychology, and organizational behavior. Patterson et al. [37] first defined customer engagement as the degree of customer involvement in perception, emotion, and behavior in the relationship between customers and a brand or a specific organization. The term is multidimensional, dynamic, and complex in the connotation and essence and is widely used in the online brand community. Furthermore, customer engagement comprises all aspects of customer psychology and behavior: loyalty, commitment, trust, satisfaction, involvement, positive word of mouth, emotional bonding, purchase intention, value co-creation, etc. [24], [25], [38]. In addition, it is a critical variable in relationship marketing, and profoundly impacts a company and stakeholders in financial performance, market penetration, product creativity, competitive advantage, brand equity, and company reputation [39]- [41].
Community engagement is the Holy Grail of establishing customer relationships in online communities [42] because it expresses the close relationship between customers and brands, products, companies, and other clients. From the angles of customers' psychology, motivation and attitude, community engagement reflects the customer interaction, sharing, and cooperation with other members of the online community, and their active participation in community activities to create value for themselves, other members, and the online community [43]- [45]. Based on the self-efficacy theory, Han et al. [46] found that community engagement posi-tively affected collective efficacy and significantly impacted community members' social happiness and loyalty through collective efficacy.

C. BRAND ATTACHMENT
The attachment theory was put forward by Bowlby [47] when analyzing the relationship between mothers and their infants. He pointed out that mother-infant attachment behaviors had four prominent characteristics: seeking intimacy, safe harbor, reliable foundation, and separation grief. People could generate psychological and emotional attachments to various objects, e.g., pets [48], gifts [49], and collectibles [50]. In the field of marketing, attachment is transformed into a concept related to customer relationships, such as place attachment [51], exhibition attachment [52], and brand attachment [26]. In online communities, attachment positively affects e-loyalty (online stickiness and referral loyalty) [53]. Attachment is the core of customer-brand relationships [54]. If we cannot first understand the relationship between the customers and the nature of possession attachment, it is impossible to expect to comprehend customer behaviors [55].
Brand attachment describes the strength of the connection, which relates to the emotional and cognitive bond between the brand and consumers [56], and the bond has a positive effect on brand equity, customer value, and company financial results [57], [58]. Brand prominence and brandself connection are two of the most significant conceptual properties of brand attachment. The brand-self connection refers to customers' perceptions and emotions about a brand, while brand prominence measures a brand's level of arousal in the mind of the customers [59]. Strong brand attachment causes customers to sacrifice specific resources to maintain a better relationship, such as premium purchases, word of mouth recommendations, defending acts, and even forgiving brand negligence and scandals [58].

D. MORAL IDENTITY
Blasi [60] first studied moral identity as a professional term. He described it as the orientation and expectation of an individual to grow into a moral person out of his moral vision. It is also viewed as a social self-concept schema regarding an individual's moral status or image and, therefore, a potential link between moral cognition and behavior. Many scholars divide moral identity into internalization and symbolization [61]- [64]. Internalization is rooted in the individual's heart and comes from the mind's judgment and acceptance of morality. Symbolization reflects the individual's actions and comes from the external moral performance of the individual's social life.
Moreover, as influencers, moral identity is both a predictor and mediator of individual prosocial behavior [65], [66]. Individual moral characteristics are reflected or expressed by their actions and moderate ethical emotion, psychology, motivation, and decision [67], [68]. Notably, the essential components of social relationality include ethical communication and commitment [69].
We believe that moral identity is a self-recognition scheme that integrates individual moral characteristics, standards, and ideals into a society's moral image, behavior, and goal. It is also the moral orientation and position of ''who I am, how I want to do, and where I am going.'' The literature presents moral identity as a relatively mature empirical paradigm with fruitful research results. However, in the context of the Web 3.0 era, studies about the complicated relationships between moral identity and netizens' cognition, psychology, emotions, and actions are insufficient.

III. RESEARCH MODEL AND HYPOTHESES A. COMMUNITY ENGAGEMENT AS ANTECEDENT
Community engagement is a crucial factor of online communities' success, and its influences can be summarized into four categories. First, community engagement can help companies improve financial performance by increasing sales, product innovation, market penetration, and competitive advantage [70], [71]. Second, community engagement promotes brand-related results, comprising brand loyalty, attachment, commitment, satisfaction, trust, evaluation, positive attitude, and self-brand connection [72]- [75]. Third, community engagement plays an essential role in the growth and sustainable development of online communities, including value cocreation, positive word-of-mouth, stickiness intention, community participation, continuous use, online comments, identification, and feedback [44], [76], [77]. Finally, community engagement significantly improves customer value, such as economic benefits, knowledge benefits, social benefits, self-esteem, satisfaction, happiness, sharing, and social wellbeing [9], [78].
Online communities have developed into a strategic marketing tool for building relationships with customers, which explains why world-renowned companies have invested in online community construction [79]. However, the mechanism of the brand relationship generated by the community relationship is still not fully explored [80]. Previous research has shown that community engagement can enhance brand loyalty and equity [43], [81]. This relationship's transformation from the community to the brand highlights community engagement's marketing function, and more attention is still needed to comprehend the mediating effect of brand attachment [25], [80].
Based on psychological and sociological theories, Baldus [23] found that community engagement significantly affects oppositional loyalty, willingness to pay a price premium, word-of-mouth, and other supportive behaviors through psychological sense. Antagonistic loyalty reflects the relationship between preferred brands and competitive brands simultaneously. Through the mediating role of perceived benefits and community commitment, community engagement has a significant positive effect on adversarial loyalty: oppositional referrals and refusing to buy products from competing brands [9]. Hence, we propose the following hypotheses: H1a: Community engagement has a positive impact on the willingness to pay a price premium.

H1b: Community engagement has a positive impact on oppositional referrals.
H1c: Community engagement has a positive impact on schadenfreude.
H1d: Community engagement has a positive impact on trash-talking.
H2: Community engagement has a positive impact on brand attachment.

B. MEDIATOR OF BRAND ATTACHMENT
Interestingly, the formation mechanism of brand attachment is mainly related to three theories: self-congruity, resources view, and customer engagement [57], [82], [83]. Selfcongruence includes the actual self and ideal self [84]. When the brand image is consistent with the customer's actual or ideal self-concept, it is possible to form and strengthen brand attachment [85]. Related to brand attachment highly, the three types of resources are hedonism, symbolism, and function [83], [86], corresponding to the three abilities of a brand: enticing, enabling, and enriching the self [87]. Customer engagement establishes emotional bonds and self-connection with the brand [24], [88], and brand attachment enhances the relationship between customer engagement and brand loyalty [25], [26].
Customers firmly attached to a particular brand may sacrifice their image, time, money, energy, and other resources to demonstrate brand loyalty. These loyalty behaviors comprise brand commitment, repeat purchases, willingness to pay a price premium, blog writing, and community participation [56], [58]. Besides having a direct positive effect on loyalty [89], brand attachment also acts as a mediator between consumer engagement and loyalty; furthermore, brand attachment has a greater mediating impact on the customer engagement-brand loyalty relationship than customer trust [25].
In addition to motivating positive consumer behaviors, brand attachment can also trigger negative behaviors such as anti-brand actions, trash-talking, schadenfreude, compulsive purchase, and impulsive consumption [22], [90]. Referring to the theory of antagonistic loyalty, Marticotte et al. [21] showed that focal brand loyalty and self-connection could jeopardize or impair the rival brands, and these kinds of negative consequences were termed oppositional referrals. According to Pournaris [27], brand attachment positively affects word-of-mouth, willingness to pay a price premium, and oppositional brand loyalty via consumer commitment. Consequently, we posit: H3a: Brand attachment has a positive impact on the willingness to pay a price premium.
H3b: Brand attachment has a positive impact on oppositional referrals.
H3c: Brand attachment has a positive impact on schadenfreude.
H3d: Brand attachment has a positive impact on trashtalking. VOLUME 10, 2022 H4a: Brand attachment mediates the relationship between community engagement and the willingness to pay a price premium.
H4b: Brand attachment mediates the relationship between community engagement and oppositional referrals.
H4c: Brand attachment mediates the relationship between community engagement and schadenfreude.
H4d: Brand attachment mediates the relationship between community engagement and trash-talking.

C. MODERATING ROLE OF MORAL IDENTITY
More and more brand moral behaviors are being supervised, inspected, and judged by the public. Brand ethics has increased in importance as a function of customer loyalty, and moral considerations develop into business norms, especially when consumers punish or retaliate against certain brands for contravening the rules. Boosting brand loyalty requires a strong sense of morality, including honesty, fairness, integrity, civilization, social responsibility, and other facets [91]- [94]. A company with stricter ethical standards is more likely to win its clients' commitment to corporate social responsibility. This perceived commitment enhances higher customer trust and satisfaction, thereby reinforcing brand loyalty [95].
With growing ethical consumption, brand ethics is increasingly essential for consumers to fulfill their desire for selfidentification or self-expression and substantially affects brand commitment, passion, and experience [96]. Notably, taking ethical issues into account from the perspective of brand norms promotes marketing activities on the right course and improves the relationship between brands and customers [97].
Ethical judgment does not always lead to ethical behavior. Nevertheless, moral identity moderates the relationship between moral decisions and activities [98], [99]. Through the impacts of moral reasoning choice, distrust, moral judgment, dissatisfaction, or moral identity, corporations' violations of societal and human behavioral standards may result in lousy word-of-mouth, schadenfreude, brand avoidance, boycott, and anti-brand activity [100], [101]. Regarding moral cognition, brand attachment ethically affects customers' consciousness and behavior [102] and strongly relates to communication-based fairness, value-based fairness, and loyalty inclination [94]. Accordingly, the following hypotheses are proposed.
H5a: Moral identity positively moderates the relationship between community engagement and the willingness to pay a price premium.
H5b: Moral identity positively moderates the relationship between community engagement and oppositional referrals.
H5c: Moral identity negatively moderates the relationship between community engagement and schadenfreude.
H5d: Moral identity negatively moderates the relationship between community engagement and trash-talking.

H6a: Moral identity positively moderates the relationship between brand attachment and the willingness to pay a price premium.
H6b: moral identity positively moderates the relationship between brand attachment and oppositional referrals.
H6c: Moral identity negatively moderates the relationship between brand attachment and schadenfreude.
H6d: Moral identity negatively moderates the relationship between brand attachment and trash-talking.
We constructed a research model of oppositional loyalty based on the above hypotheses, as shown in Figure 1.  Table 1.

B. MEASURES
Measurement items in our study were derived from scales found in authoritative literature and were adapted to mirror scenarios about Xiaomi's online communities. Community engagement items chiefly referred to the scale of Baldus [23] and Algesheimer [45]. The items of brand attachment were mainly based on Lacoeuilhe [103] scale and Louis and Lombart [104]. The moral identity was measured using ten items developed by Aquino and Reed [64]. The measurement of willingness to pay a price premium was based on Pournaris [27]. The scale of Kuo and Hou [16] was mainly used for the questions of oppositional referrals. Items used to measure schadenfreude were adapted from Baren [105] and Japutra [22]. The trash-talking items were primarily referred to the scale of Bansal and Clelland [106] and Marticotte [21]. Appendix A lists the items in detail.
An associate professor of marketing and a Ph.D. in English adopted back translation to reduce the translation deviation of the English questionnaire to ensure the consistency and accuracy of the Chinese scale in expression. We optimized and modified 13 items after the Delphi method, two team meetings, and other procedures to ensure that the items were reliable and valid. Following the pretest of 153 valid data and the application of exploratory factor analysis, we removed two items from the trash-talking, and willingness to pay a price premium. Answer options were given on a seven-point Likert scale from ''completely disagree'' to ''completely agree.'' Respondent requirements for the first two items of the questionnaire included participating in Xiaomi mobile communities and writing at least three posts each month.

C. PROCEDURE
This study mainly collected data through the online survey site Sojump.com (a famous and professional Chinese survey site that released over 100 million questionnaires). In the questionnaire introduction, we elucidated the academic goals and core constructs. A gift of CNY 5.00 was offered to respondents to complete the questionnaire, and the investigation was anonymous. The questionnaires were emailed privately to the members of Xiaomi's mobile online communities (such as the Xiaomi community, Xiaomi's official website, and Xiaomi's microblog), then we befriended the members and asked them to complete the survey. In cases where the respondents were not online, we procured their email addresses together with a link to the questionnaire. The samples were collected from July to September 2021.

A. NORMALITY, RELIABILITY, AND VALIDITY
The mean value of the seven variables ranged from 4.143 to 6.069, and the standard deviation was between 0.699 and 1.698. The absolute values of the maximum skewness and kurtosis of the variables were 1.379 and 1.368, respectively; 533 data met the normal distribution requirements of empiri-cal analysis [107]. The sample's overall Cronbach's alpha was 0.925, conforming to the reliability standard from an overall perspective.
Moreover, according to Table 2, all seven variables had alpha levels greater than 0.8, suggesting that each variable had high internal consistency. Over 0.6 was the standardization factor loading for all measurement items, and the average variance extracted (AVE) of each variable was higher than 0.50, implying excellent convergent validity. Lastly, the minimum value of the square root of the AVE score for each construct was 0.715, and the maximum value of the correlation coefficients among the constructs was 0.494 (Table 3). In each row and column, the square root of each variable AVE was indubitably greater than the correlation coefficients, demonstrating that the differentiation between the variables is valid [108].

B. ANALYSIS OF COMMON METHOD BIAS
Seven components with eigenvalues greater than one were extracted, and the proportion of total variance explained by component 1 was 27.838%, which was less than the critical value of 50% [109]. Moreover, the maximum correlation coefficient among the variables was 0.545, less than the critical value of 0.75 (Table 3). Therefore, the sample data's common method bias was not serious and could not affect this study's conclusion.

C. TESTING OF DIRECT EFFECTS
The testing results were as follows: absolute fit indices:  Community engagement was associated with the willingness to pay a price premium, oppositional referrals, schadenfreude, trash-talking, and brand attachment, with standardized coefficients being 0.131, 0.197, 0.294, 0.187, and 0.332, respectively. In addition, t-values exceeded 2.680, and P-values were less than 0.01 (Table 4). This meant that community engagement had a significant positive impact on brand attachment and the four dimensions of oppositional loyalty. The hypotheses H1a-H1d, and H2 were supported by empirical data. Meanwhile, the brand attachment affected the willingness to pay a price premium, oppositional referrals, schadenfreude, and trash-talking, and the standardized coefficients were 0.366, 0.423, 0.496, and 0.501, respectively, with t-values greater than 6.921 and P-values less than 0.001. It showed that community commitment played a significant role in four dimensions of confrontational loyalty, and the hypotheses H3a-H3d were supported by empirical evidence.

D. TESTING MEDIATION
Brand attachment is a mediator between community engagement and the willingness to pay a price premium, oppositional referrals, schadenfreude, and trash-talking ( Table 5). The standardized effect values of the coefficient product were 0.121, 0.140, 0.165, and 0.166, respectively. All P-values were less than 0.001. Simultaneously, within the 95% confidence interval, zero was not included between the upper and lower limits of the bias-corrected and percentile. Therefore, brand attachment played a significant mediating role between community engagement and the four dimensions of oppositional loyalty. The hypotheses H4a-H4d were supported by empirical data.  variables after standardization: community engagement and brand attachment. Model 2 was the addition of a standardized moderator variable based on Model 1: moral identity. Model 3 was based on Model 2 by adding standardized interactive items of community engagement × moral identity and brand attachment × moral identity. The results are shown in Tables 6 and 7. The largest VIF value was 1.192 in all multilayer regression tests, and the DW values ranged between 1.868 and 1.990. The R2 of Model 3 was higher than that of Model 2, and the R2 of Model 2 was higher than that of Model 1. Each of the F-values was significant at a level less than 0.001. The normal distribution was observed in standardized regression residuals. Taken together, the model fitted well, and the overall effect was statistically significant.
In the regression test of moral identity moderating the relationship between community engagement, brand attachment, and oppositional referrals, the interaction coefficients were 0.136 and 0.227, respectively, and the t-values were 3.748 and 6.227, respectively, indicating that the positive moderating effects were significant. Hence, Hypotheses H5b and H6b were supported.
We examined moral identity's moderating effects on community engagement, brand attachment, and schadenfreude through multiple regressions. As a result, the interaction coefficients were −0.181, −0.133, and the corresponding t-values were −5.173, −3.800, respectively. It was evident that the negative moderating effects were significant. Hence, Hypotheses H5c and H6c were supported. At the same time,   moral identity moderated the effect of brand engagement on schadenfreude, with an interaction coefficient of −0.197 and a t-value of −5.8117. A significant negative moderating effect could be observed. Accordingly, Hypothesis H5d was supported.
Tables 6 and 7 present regression tests of moral identity regulating the relationship between community engagement, brand attachment, and the willingness to pay a price premium. The coefficients of the interaction terms (0.002; 0.035) and the t-values (0.053; 0.849) express that the positive regulating effects were insignificant, assuming rejection of H5a and H6a. Meanwhile, we looked at how moral identity moderated brand attachment and trash talk. There was an interaction coefficient of −0.056, a t-value of −1.439. Consequently, hypothesis H6d was rejected.

VI. DISCUSSION AND IMPLICATIONS A. DISCUSSION
The empirical result showed that community engagement and brand attachment had significant positive effects on confrontational loyalty. Relevant to our research, Wirtz et al. [44] believed that community engagement significantly affected customer commitment, loyalty, and willingness to continue participating in the community. Similarly, the existing literature has demonstrated the positive effects of community engagement on antagonistic loyalty [2], [9], [23]. Furthermore, emotional attachment is an essential predictor of brand attitude, engagement, and value. As an emotional link between the brand and self-concept, brand attachment prompts customers to pay a price premium, repeat purchases, commitment, recommendations, and other loyal behaviors [26], [58]. Meanwhile, under the influence of attachment style (anxiety and avoidance), prior studies have confirmed that solid brand attachment leads to unfavorable customer behaviors, such as anti-brand actions, schadenfreude and trash-talking [34]. Comparing previous scholars with the research dimensions, we expand the comprehensive academic understanding of oppositional loyalty: the willingness to pay a price premium, oppositional referrals, schadenfreude, and trash-talking.
In terms of moral identity moderating brand relationships, our empirical results are consistent with previous studies. Yand and Yen [110] believed that both the symbolization and internalization angles of moral identity positively adjusted the impact of self-construction on purchase intention. Huang et al. [94] emphasized that customers' perceptions of fairness played a significant role in influencing brand loyalty. Apparently, antagonistic loyalty may benefit favored VOLUME 10, 2022 brands financially. However, schadenfreude and trash-talking against rival brands, moderated by moral identity, damage the favorite brand's reputation and equity.
We conducted this study to clarify the perception among companies and online communities that increased antagonistic loyalty behaviors were desired. Our results highlight that the adverse effects on competing brands are harmful to favored brands through the impact of moral identity. Concerning moral identity, our findings showed that it did not positively moderate the effects of community engagement and brand attachment on willingness to pay a price premium significantly. This may be because 70.92% of respondents reported an income beneath 4,000 CNY per month. Therefore, economic conditions limited customers' willingness to pay a price premium, and moral identity played a less prominent role in regulating actions. Besides, those who possess a higher moral identity reckon to purchase Xiaomi's phone at a price premium to benefit the company financially and harm consumers' interests.
The hypothesis that moral identity negatively moderates the effect of brand attachment on trash-talking (negative oppositional loyalty behavior) was also rejected. The influence of other variables on moral identity adjustment (including company scandals, product recalls, immoral speeches or actions of company leaders, and negative information) induced stereotypes among community members. The company's unethical and social responsibility violations generate customer reactions such as negative word-of-mouth, anger, contempt, punishment, discrediting, anti-brand, protests, boycotts, and even deal-breaking [111]- [114]. Furthermore, moral identity is greatly influenced by personal preference, cultural background, education, living environment, and social customs. Therefore, some community members may indulge in trash-talking towards competitive brands.

B. THEORETICAL IMPLICATIONS
This article extends the research about traditional moral identity into the mobile internet context and narrows the research gap concerning moral identity with antagonistic loyalty in online communities. Our empirical analysis shows that community engagement and brand attachment significantly affect the four dimensions of confrontational loyalty. Furthermore, brand attachment plays a mediating role where community engagement affects oppositional loyalty. Moral identity has significant positive and negative moderating effects of community engagement and brand attachment on the portion of oppositional loyalty.
On a more detailed level, we make the following contributions to research. First, we define oppositional loyalty in the online community and divide it into four structural dimensions: willingness to pay a price premium, oppositional referrals as positive influences on preferred brands, and schadenfreude and trash-talking with unfavorable influences of competing brands. The empirical results are consistent with previous studies concerning antagonistic loyalty [5], [16], [22] but supplement the deficiencies in the sub-division of its structural dimensions. We enhance the existing antagonistic loyalty theory by systematically presenting the structure of oppositional loyalty, expanding related knowledge, and addressing the deficiencies of the brand loyalty theory.
Second, the theoretical model of antagonistic loyalty is constructed and verified by data. We find that brand attachment is an essential driver for adversarial loyalty, thus responding to the call for further investigation into the influencing factors of antagonistic loyalty [2], [9]. Attachment promotes the willingness to pay a price premium and oppositional referrals with favored brands and produces schadenfreude and trash-talking with competitive brands. These new paths provide theoretical support for future oppositional loyalty research and enrich the engagement and attachment theories in influencing results.
Third, moral identity plays a significant positive moderating role between community engagement, brand attachment, and oppositional referrals and has adverse moderating effects of community engagement and brand attachment on schadenfreude. Moreover, moral identity negatively moderates the relationship between community engagement and trash-talking significantly. Previous studies show that the perceived fairness of morality is closely related to brand attachment and loyalty intention [94]. In the event of ethical violations, customers may retaliate against the brand, schadenfreude, and anti-brand activism [22], [101]. Our perspective of confrontational loyalty helps academia understand the moderating role of moral identity [98], [99].

C. MANAGERIAL IMPLICATIONS
Combined with the theoretical review and empirical results of brand relationships, antagonistic loyalty is a double-edged sword for online communities, enterprises, and stakeholders. Without a doubt, online communities are the Holy Grail for cultivating brand loyalty, but these communities are not a panacea for all marketing effects. Enterprises should discard the idea that more oppositional loyalty is ideal. From the perspective of sustainable long-term development, the adverse impact of negative antagonistic loyalty behaviors on the favorite brand through moral identity should not be underestimated -it will inevitably damage favored brand equity and market value and stunt its growth. Therefore, we propose the following suggestions for managing online communities.
First, enterprises should correctly guide oppositional loyalty behavior and increase resource investment in the online community to enhance engagement. For example, if members publish posts about their willingness to pay a price premium and oppositional referrals to the preferred brand, they should be awarded bonus points, medals, titles, privileges, upgrades, and prize exchanges. In contrast, community members who post extensive schadenfreude or trash-talking aimed at competitive brands should be removed, demoted, blocked, or banned. Furthermore, multiple online communities should be established. These communities can disperse the impact of negative antagonistic loyalty behaviors and reduce the probability of severing the brand relationship. Moreover, it can agglomerate the role of positive confrontational loyalty behaviors and meet the individualized demands of community members. Finally, opinion leaders need to be valued as role models. Our findings showed significant active relationships between the positive and negative behaviors of confrontational loyalty and the time of joining the community, weekly participation time, and the number of monthly posts. Companies should make full use of these direct and close relationships and opinion leaders' discourse power in the online community to correctly induce the positive and negative behaviors of antagonistic loyalty.
Second, corporations need to improve brand attachment strength because it significantly impacts the four dimensions of confrontational loyalty. Enterprises should prevent, guide, and balance the occurrence and spread of negative antagonistic loyalty behaviors but support and encourage the formation and spread of positive oppositional loyalty behaviors. On one side, online community managers should pour more attention to shaping brand personality and urge it to match community members' actual or ideal self, thereby increasing the positive actions of confrontational loyalty. For those who like to attack rival brands, a forum like an ''emotional venting room'' can be set up in the community. Members who often post schadenfreude and anti-brand actions feel love and freedom but limit the negative influence in a small circle. On the other side, the quality of information will directly affect the attachment emotions of community members. Online community managers should timely release more accurate, reliable, and popular information to promote the community and the brand. At the same time, let more community members perceive the justice and fairness of the community.
Third, firms should promote moral identities and strengthen the education of community members and guide them in setting up correct values. The community's moral culture and communication norms should be established, for example, by adding moral identity aspects to brand stories, eliminating vulgar elements in community symbols, or motivating community members with outstanding moral performance in community activities. A daily public opinion monitoring mechanism to supervise unethical tendencies and posts of community members can reduce the adverse effects of negative behaviors, including schadenfreude and trash-talking. It imperceptibly influences the psychology and behavior of oppositional loyalty, promotes the right actions, and limits the damaging posts.

D. LIMITATIONS AND FUTURE RESEARCH
This study has some limitations, which provide directions for further research and improvements. First, macroenvironments, including politics, economy, laws, science, technology, social culture, customs, and habits, may have a particular influence on confrontational loyalty. The degree to which negative antagonistic-loyalty behaviors influence positive behaviors by moral identity, together with spillover effects on the whole industry, needs to be further investi-gated. Second, this study's sample source may be unbalanced, affecting multiple factors such as capacity, resources, time, location, and technology. Finally, the moderating variable in this study only selects the factor of moral identity. A range of moderating variables, including community identity, brand trust, psychological ownership, brand satisfaction, promotion, or defensive orientation, may also influence oppositional loyalty.

MEASUREMENT ITEMS COMMUNITY ENGAGEMENT
CE1: In Xiaomi's mobile online community, I could help improve its products.
CE2: In Xiaomi's mobile online community, I could support other members.
CE3: In Xiaomi's mobile online community, I could enjoy a lot of fun.
CE4: In Xiaomi's mobile online community, I could acquire more knowledge.
CE5: In Xiaomi's mobile online community, I could get financial benefits.
CE6: In Xiaomi's mobile online community, I could make many friends.
CE7: In Xiaomi's mobile online community, I could express my opinions and ideas freely.

MORAL IDENTITY
Consider the following traits: caring, compassion, fairness, friendliness, generosity, helpfulness, hardworking, honesty, kindness MI1: It would make me feel good to be a person who has these characteristics.
MI2: Being someone who has these characteristics is an essential part of who I am.
MI3: I would be ashamed to be a person who has these characteristics. (R) MI4: Having these characteristics is not important to me. (R) MI5: I strongly desire to have these characteristics.
MI6: I often use mobile phones that identify me as having these characteristics.
MI7: The types of things I do in my spare time (e.g., participating in the online community activities of Xiaomi's mobile) clearly identify me as having these characteristics.
MI8: The kinds of books and magazines that I read identify me as having these characteristics. VOLUME 10, 2022 MI9: The fact that I have these characteristics is communicated to others by my membership in specific organizations (e.g., the online community of Xiaomi's mobile).
MI10: I am actively involved in activities that communicate to others that I have these characteristics.
WILLINGNESS-TO-PAY a PRICE PREMIUM PP1: Although the price of Xiaomi's mobile with the same configuration is slightly higher than that of competitive brands, I would continue buying it.
PP2: Comparing the products of competitive brands, Although Xiaomi's mobile charges extra service fees (such as maintenance and value-added services), I would continue to buy from it.
PP3: Although Xiaomi's mobile at the same price has a slightly lower configuration than competitive brands, I would continue to buy from it.
PP4: Comparing with the products of the competitive brand, although the price of Xiaomi's mobile accessories (e.g., smartwatches) is slightly higher, I would continue to buy from it.

OPPOSITIONAL REFERRALS
OR1: Although many people appreciate competitive mobile brands, I still give negative comments based on product experience.
OR2: Even if competitive mobile brands launch new and better products, I still try to prevent others from buying based on product experience.
OR3: Even if competitive mobile brands meet specific needs, I still express negative opinions based on product experience.
OR4: I like to spread negative reviews of competitive mobile brands based on product experience in the online community.
OR5: When friends and relatives consulted, I told them not to buy mobiles from competitive brands based on product experience. SCHADENFREUDE SF1: I feel happy when I browse the company scandals of rival mobile brands. SF2: I am delighted when I hear negative news about rival mobile brands.
SF3: I enjoy it that fans of rival mobile brands are abused in the community.
SF4: I could not resist a slight smile upon discussing that those mobiles of the rival brand were recalled.

TRASH-TALKING
TT1: Although I have never used a rival brand's mobile, I also think its quality is inferior.
TT2: Although I have never used a rival brand's mobile, I still question its consumers and fans.
TT3: I sometimes misinterpret the rival brand's image.