Good Reviews Go out the Door, Bad News Travels Fast?—Research on the Influence of Reference Groups and Their Language Styles in Positive Online Reviews on Consumers’ Brand Attitudes

Based on the theory of brand 
identity and social comparison theory, this study explores how positive 
comments from different reference group types affect consumers’ brand 
evaluation and purchasing intentions, and then this study discusses how the 
language style of positive online reviews can further regulate the influence of 
positive comments from different reference group types according to social 
comparison theory. The study by two scenario-based experiments found that when 
positive comments come from avoided groups compared to desired groups, 
consumers produce lower brand evaluation and willingness to buy, in which the 
psychological mechanism is that consumers are influenced by the avoided group 
and produce lower brand identity. Interestingly, the language style of positive 
reviews has a significant lytic effect on the effect of positive comments in 
the reference group. Compared with self-reinforcing language style, when 
avoiding the group’s self-deprecating language style, consumers will produce 
higher brand identity based on downward comparison, which in turn leads to 
higher brand evaluation and willingness to buy. This study examines the 
influence of reference group type on positive comment effect, and on this 
basis, for the first time verifies the adjustment effect of different language 
styles on the positive comment of the reference group, which to some extent 
makes up for the deficiency of the existing theory, and also gives the 
enterprise how to view the positive comments correctly. In the era of Internet e-commerce to make good use 
of the power of online review to build a good brand image, improve marketing 
effect provides the corresponding management inspiration.


Introduction
Many scholars and business managers believe that positive reviews can produce positive marketing effects. If the brand is spread by positive online reviews from consumers, it will greatly promote the increase of brand sales. Consumers at the same time look at some online comment which is good or bad are the basis of success of brand promotion, since the media environment electric business platform to comment online consumers in the way different from the traditional, the past is given priority to the kind of language limitations of word of mouth which is too big, not only exists in the transmission range and speed, and transmission need certain conditions has brought certain obstacles. Compared with the traditional way of communication, consumers now prefer to post comments in the form of "picture + text", which is also much faster and wider than the traditional way. Many merchants on online retail platforms such as Tmall, taobao and pinduoduo also realize the importance of positive reviews and increase product praise in various ways, such as encouraging consumers to post "favorable comments" or offering certain cash coupons to consumers who "print photos". In order to attract consumer reviews, merchants also provide external rewards represented by material rewards to consumers, in an attempt to stimulate the enthusiasm of consumer reviews and thus attract more potential consumers. What's more, some businesses will use machine software to virtual "likes" and "praises", and hire mercenaries to post positive product reviews disguised as brand consumers. The research of scholars vilnai-yavetz and Tifferet (2015) found that online reviews have the effect of "one picture is worth a thousand words". Zhang Yanhui and Lee Chongwei, 2016;Lin Shuang et al., 2017 pointed out that pictures in online word-of-mouth can enhance the usefulness of reviews and consumers' purchase intention. The review mode of "praise + print" will be more convincing to consumers, because online reviews with pictures are more vivid and can have a greater impact on consumption decisions.
Although the traditional view that a positive online reviews in the general conditions of brand with cadbury, and we observe that some consumers in the process of online publishing positive comments when the language style used may affect the other consumer's attitude toward the brand, leads to that its evaluation of the brand and purchase intention are changing. When target consumers of a brand see a series of "print pictures + praise" behaviors of non-target consumers, positive online reviews may even produce unexpected reactions, which can be called "praise goes out, bad news travels fast". For example, we found something interesting. Tmall singles' day celebration, some girls' clothing brand flagship store innovation, a wide range of goods, sales evaluation is very hot. When we looked at the customer reviews, we found that some elderly women and middle-aged women who were wearing age-inappropriate or even inappropriate clothing for girls posted a lot of positive behaviors, which made young girls hesitate to buy such clothes. In addition, some "middle-aged fat uncle" page, comments on the young man's shirt to a pile of similar "this dress looks really handsome" "feeling goes with me ah" "simple sense is good, the makings are full marks such as" compliments, then don't forget to take a few powers of handsome wearing the shirt as attached to the above comments, but the man in the strange hairstyle exaggeration, bloated figure put his shirt, the deformation of the netizens feel rather he deliberately belittle product, and even called it the "fat strange uncle" "wonderful work". Again like, recently, a man named microblog hot search list # # sellers want to delete the buyer show topic caused the attention of people, many netizens to participate in the discussion, less than two days reading quantity exceeds 300 million, discuss the amount of 50,000 person-time, this topic mainly want to sharing his own seen buyers of some of the funny show, we in turn top the sum of evaluation work, slag saw a bearded man, drying out using the armani lipstick after a "flaming lips" figure, a caption wrote: "super super nice, a very high level of feeling". Seemingly humorous behind, it is the sellers do not want to see the positive online praise, and even some sellers would rather spend money to delete the praise behavior.
Users of products post these "buyer shows" to share their shopping experiences, give brands good word-of-mouth, and provide positive online reviews, but make other consumers want to "never buy again." Many netizens have made these "buyer show" vs. "seller show" into humorous comparison pictures, which have been widely spread on weibo, WeChat, douyin and other social media.
The initial intention of enterprise personnel is to encourage consumers to "speak well of" and "print pictures" for the products through a large amount of marketing input to enhance consumers' purchase intention and bring positive marketing effects. But in the aforementioned cases, positive online reviews have a negative marketing effect, causing other potential consumers to have a lower opinion of the product, or even turn away in disgust. So, why is this the case, and why don't good reviews bring good results?
Research purpose: The fundamental purpose of this study is to reveal two different moderators: the reference group type (desire vs. avoidance) and the language style (self-improvement vs. self-mockery) that influence consumers' positive online reviews of brands, resulting in differences in brand evaluation and purchase intention. Previous studies on the boundary conditions under which positive online comments exert negative effects have paid less attention to the characteristics of comment originators. This study intends to introduce the concept of reference group proposed by American sociologist heyman in 1942 into the re-search field of positive online reviews, and discuss the differences in consumers' perception of positive reviews under different types of reference groups.
This study intends to introduce comment language style type as regulating variable, discussed under the condition of positive comments, positive online reviews on consumers' brand evaluation and purchase intention of boundary conditions, namely positive online reviews have a negative effect of boundary conditions, and introducing the theory of brand identity, social comparison theory to explain the psychological mechanism of consumers.

Reference Group
In social psychology, reference group refers to those groups that serve as the standard or model for people to judge things. The reference group is also known as the significant others, and the attitude and standards of the reference group often become the standard for individual self-judgment and social action (merton, 2006). Since then, social psychology studies have pointed out that the reference group has the function of comparative evaluation. Individuals use the reference group as the criterion and starting point of social comparison to evaluate themselves or others. Therefore, in the context of social marketing, real consumers who use brands can serve as reference groups for brands. The reference group for a brand is the group of consumers who compare their brand with them socially when evaluating the brand (Escalas & Bettman, 2005).
Consumers with different degrees of social identification with the group can divide the reference group into ingroup and outgroup. Ingroup refers to the social group in which consumers currently perceive themselves (Malär et al., 2011;Van Veelen et al., 2015), and the social group that consumers currently perceive they do not belong to is the out-group. Outside groups contained in the desire to group (aspirational reference group) and avoid group (dissociative reference group).
Ingroup and outgroup are related to the real self of consumers, that is, consumers perceive the social group to which they belong at present (Malär et al., 2011;Van Veelen et al. 2015), desire group and avoidance group are related to consumers' ideal selves. The desire group is the social group that consumers identify with and want to be a member of. However, avoidance groups, such as "rich second generation" and "tuhao", are social groups that consumers do not identify with and avoid becoming one of them.
At present, studies on reference groups at home and abroad mainly focus on the influence of reference groups on consumer attitudes and behaviors, including consumers' purchase intention, purchase behavior, product evaluation and brand attitude. Chen jiayao et al. (2006) explored the influence of reference group on consumers' purchase intention by means of situational simulation experiment, and pointed out that the evaluation of perceived value by reference group can significantly affect consumers' purchase intention. Gong xiushuang et al. (2017) analyzed the data of more than 1000 residents' consumption and showed that different influence types (normative influence and informational influence) generated by reference groups can positively affect consumers' purchase intention. In terms of purchasing behavior, consumers tend to avoid and give negative evaluations to products used by evading groups (Chae et al., 2017). Such negative evaluations include not only symbolic products (Han et al., 2010) but also technologic-based products (Arbore et al., 2014). Similarly, in terms of product evaluation, the product use behavior of the eager group will improve consumers' evaluation of the product, while the product use behavior of the avoiding group will reduce consumers' evaluation of the product.
Reference groups have significant influence on consumer attitudes and behaviors. Previous studies have pointed out that consumers prefer ingroup and desire group (Escalas & Bettman, 2005), and there is out-group derogation, and avoid association with avoidance group. Consumers show negative attitudes and behaviors towards products used by avoiding groups (Berger & Heath, 2008;Kim et al., 2016). White and Dahl (2007)'s research on outgroup shows that avoidance group has a significant impact on consumers' self-brand connection, consumer evaluation, product evaluation and consumer choice, and consumers have different attitudes towards different outgroup.
In terms of brand attitude, Hong-yan Liu (2017) from the perspective of self motivation of target consumption group brand use behavior on the influence of the target consumer's brand attitude, the results show that different types of target consumption group brand use behavior on the influence of the target consumers' brand attitude, this influence by the adjustment of the self construction.
The brand use behavior of the desired group will improve the brand evaluation of independent self-target consumers. Avoiding the group's brand use behavior will reduce the self-dependent target consumers' brand evaluation. Feng Wenting et al. (2019) pointed out that when the "bad group", that is, avoiding the group and "saying good things" for the brand, positive word of mouth would lead to consumers' negative stereotypes of the brand.
Through reviewing the above literature, some existing studies on the negative effects of positive reviews also involve the influence of reference groups. This study will systematically study the moderating effect of the reference group on positive comments and discuss its further influencing mechanism.

Social Identity and Brand Identity
Social identity theory was proposed by Tajfel and Turner et al. gave a new explanation of group behavior and became the most influential theory in the study of group relations. In 1978, Tajfel defined social identity as: "an individual recognizes that he or she belongs to a particular social group, and at the same time recognizes the emotional and value significance of being a member of the group" Social identity theory involves four conceptual levels: social-categorization, social identity, social comparison and positive distinction. According to social identity theory, individuals have social identity to their own group through social classification, and have in-group preference and out-group prejudice. When individuals are overenthusiastic about their own group, believe that their group is better than other groups, and experience inter-group differences in seeking positive social identity and self-esteem, inter-group prejudice and inter-group conflict are likely to arise. Tajfel believes that the pursuit of social identity is the root cause of inter-group conflict and discrimination, that is, the consciousness of belonging to a certain group will strongly influence our perception, attitude and behavior.
Tajfel and Turner's social identity theory reveals the inner psychological mechanism of intergroup behavior. Since the theory was put forward, it has received extensive attention from psychologists. The concept of "social identity" has been widely accepted by social psychology and has inspired a lot of related researches. For the first time, social identity theory distinguishes interpersonal and inter-group behavior, and divides identity into two levels of self-perception: individual identity and social identity. Social identity theory understands "social" as being under the background of group relations. Identity is generated in group relations. The identity of individuals to groups is put at the core, which reveals the essence of social psychology more deeply.
In the field of marketing, marketing scholars use this theory to analyze and explain consumer behavior. For example, Reed et al. (2003) constructed an integrated theoretical model of the influence of social identity on consumer behavior. He believes that the influence of social identity on consumer behavior must be accessible and diagnosable. Reed also defined accessibility and diagnosability.
He believed that since social identity is a part of self-concept and occupies a certain position in self-concept, its ability to be motivated and the degree of motivation is accessibility. Whether consumers can identify the group behavior to which they belong, and the degree of identification is diagnosable. In 2009, Guo Yi and hanley model of Reed is deep step development, put forward the need to put consumers in specific consumer scenarios, by giving a certain marketing system shock, to achieve the purpose of stimulate the social identity of internalization, so that the social identity from belong to self concept, so as to constantly influence consumer buying behavior.
Social identity theory has also been applied to the study of word of mouth.
Domestic scholars Wang Xinxin and Chen Runqi (2010) believe that cultural identity is the core content of social identity. Cultural identity is a kind of relationship between individuals and a certain region. Individuals share the same identity. Similar cultures and values are social identities based on the consensus of the regional members and external observers on the salient features of the regional culture. Driven by cultural identity, in-group members have a certain degree of preference towards regional brands, so they are willing to spread positive word of mouth of regional brands. Scholars Wang Xinxin and Chen Runqi analyzed the influence of cultural identity on the word-of-mouth communication of regional brands among regional consumers, and made up the shortage of researches on regional brand word-of-mouth from the perspective of consumers.
The concept of brand identity comes from the theory of social identity. Social identity theory is very useful for understanding consumer behavior, because consumers can still think of themselves as part of a social group without openly acting with other consumers (or even knowing them).
At present, researches have defined brand identity from multiple dimensions. Consumers often use the symbolic value of a brand to create and display their self-concept. DelRio (2001) distinguished brand identity into "personal identity" and "social identity" for the first time, and the concept of "personal identity" was recognized by subsequent studies. Therefore, the combination of consumer self-concept and brand will profoundly affect consumers' brand identity.
Studies at home and abroad have shown that brand identity is based on the fit between consumers and brands in the dimensions of value perception, self-image, self-perception and self-concept. Nicola et al. (2012) regard consumers' brand identity mainly as a cognitive expression. Brand identity is the company's implicit commitment to its customers and contains a unique set of internal expectations of the brand association. Kapferer (1992) developed the prism of brand identity, which holds that brand identity has six components: brand image, relationship, reflection, personality, culture, and self-image This paper defines brand identity as: brand identity is consumers' consistent evaluation of self-concept and brand. The fit relationship between consumers and brands lays a theoretical foundation for the following brand identity as a positive online review mediating consumers' purchase decisions under the adjustment of reference groups.
After the brand identity theory was proposed, scholars at home and abroad have carried out researches on brand identity theory, mainly from the perspective of the relationship between brand identity and brand equity, loyalty and brand transformation. Moreover, many scholars have conducted empirical research on it, but the research on the relationship between brand identity and each of the above mentioned is less and more scattered.
From the perspective of brand equity, brand management is increasingly regarded as a useful means to make full use of organizational assets and create value. Positioning the brand at the core of the organization represents a shift away from seeing the brand as a unique peripheral asset. In this sense, brand building goes beyond creating image and popularity; this is a way to build the "soul" of the company and communicate within and outside the company. The practical significance of brand identity is increasingly prominent.

Social Comparison Theory
In most cases, social comparison is also called interpersonal comparison. As one of the most basic and universal social psychological phenomena, social comparison has been attracting researchers for the last half century. Social comparison means that people compare themselves with others and compare themselves from the perspectives of personal ability, physical health status and situation, so as to determine their own social attributes or construct self-evaluation. According to the social comparison theory, people can make upward or downward social comparisons. Upward comparison refers to "the comparison of a person against his superior person", and downward comparison refers to "the comparison of a person who is inferior to him".
Reviewing the previous studies, the research perspective of social comparison mostly focuses on the motivation and strategy of social comparison. Researchers believe that self-evaluation, self-improvement and self-satisfaction are the three main sources of motivation for social comparison. According to Collins (2000), the results of upward social comparison constitute an important source of individuals' self-evaluation, in which individuals' expectation of the results is the key. That is to say, before the upward social comparison, individuals expect that there is a certain gap between themselves and the comparison object, resulting in feelings of inferiority, depression and so on. Based on the downward comparison proposed by Hakmiller (1966), Wills (1981) pointed out in his research that when an individual experiences any negative life event such as failure or loss, the individual's self-esteem and mental health level will decline. At this point, the individual tends to compare with people who are worse off than himself, so as to maintain his self-esteem and subjective well-being. At the same time, the maintenance model of self-evaluation emphasizes that in addition to the need for self-evaluation, social comparison of individuals actually implies the desire for positive results to maintain positive evaluation of themselves (Tesser, 1998).
Social comparison studies also link social comparisons to changes in self-esteem.
Although self-esteem is defined in a variety of ways, its common core is a sense of goodness in liking oneself or oneself. High self-esteem is associated with positive emotions, while low self-esteem is associated with negative emotions. When a person compares himself with a taller person and notices that he/she is lower than he/she is, his/her self-esteem is threatened or reduced. Some scholars use social comparison theory to evaluate how exposure to attractive public images, such as attractive models in advertisements, can lead to upward comparisons.
The results suggest that this can lead to a range of adverse consequences for women, such as negative emotions, low self-esteem, and body dissatisfaction. In social comparison theory, making downward comparisons tends to boost temporary self-esteem because they think they are better at certain areas than their comparators. Studies have shown that higher self-esteem increases positive emotions, while lower self-esteem leads to negative emotions.
Previous studies on social comparison have shown that people often make social comparisons in a "spontaneous, effortless, unintentional and relatively automatic" way. And according to the meta-analytical evidence from Want (2009), while social comparisons are automatic processes, their effects can be altered by conscious processing. How to adjust consumers' social comparison process through marketing clues will have important research significance.
With the deepening of the research on social comparison, researchers now pay more attention to the application of social comparison in other fields. This study will focus on the mechanism of social comparison online shopping on consumers' purchasing decision behavior.

Language Style
Language style is a comprehensive expression of different characteristics in the use of language, and also covers its own appropriate characteristics. Style refers to the atmosphere or style of speech and its means of expression formed by language in special communicative occasions in order to adapt to special communicative purposes. Language style is a comprehensive expression of the characteristics of language system itself and the various characteristics of language application. Language style is not the language system itself, but a way to express the meaning understood by the speaker through subjective and objective conditions.
The expression styles of interpersonal communication can be divided into self-deprecating and self-strengthening (Martin et al., 2003). Self-strengthening endorsements mainly enhance their authority and social influence by exaggerating their own attractiveness, professional level and sense of trust, or use the halo effect of celebrities to improve the marketing effect (Newman et al., 2011).
Self-deprecating expression style is to create an equal interpersonal relationship by making fun of oneself (Gong & Li, 2017). Self-deprecating endorsement is a way of self-deprecating to narrow the psychological distance with the affected person (Gong & Li, 2017), creating a close and equal form of expression, allowing the affected person to focus on the product, thus promoting brand promotion.
Self-strengthening and self-deprecating communication styles are also widely used in the study of humor. Humor is a common advertising communication strategy (Eisend, 2009) in practice, many enterprises use humorous language to respond to consumers' complaints. Martin et al. (2003), and such as Vaillant (1977) the individual good humor communication style is divided into two categories: type self-strengthening humor (Self-enhancing humor), refers to the individual in the face of the pressure by self-reinforcing humorously emphasize their own advantages, is a defense or enhancement mechanism of courage. Self-deprecating humor: self-deprecating humor refers to the way that an individual makes others feel more lovable (or less threatening) by joking about himself or herself, or describes a third party who has nothing to do with him or her (such as telling jokes) to shorten the psychological distance with him or her, so as to reduce the tension in interpersonal relationship (Lefcourt & Martin, 1986). Since brands have human-like attributes, enterprises can also use these two types of humor to communicate with consumers. Corporate humor (compared with non-humor) complaint response strategy is more conducive to improving con-DOI: 10.4236/jss.2020.82022 280 Open Journal of Social Sciences sumer brand attitude; In addition, different types of humorous responses have different effects. Self-strengthening humorous complaint response is more effective in improving consumers' brand attitude under the co-ownership relationship paradigm, while self-deprecating humorous complaint response is conducive to improving consumers' brand attitude under the co-ownership and transactional relationship paradigm.
In social media, the expression style of celebrity endorsement also has similar characteristics (Kozinets, Valck, Wojnicki, & Wilner, 2010). This paper reviews previous literature studies and finds that the two language styles of celebrity endorsement (self-improvement vs. self-mockery) have important effects on consumers' purchase intention and word-of-mouth recommendation. Studies have shown that self-strengthening celebrity endorsement can create huge economic value and social value for enterprises (Chung, Derdenger, & Srinivasan, 2013).
Some scholars also believe that celebrity self-mockery can bring favorable word-of-mouth communication value (Chen, Chang, Besherat, & Baack, 2013;Chung & Cho, 2017). Huang minxue (2018) explored the influence of two types of celebrity endorsement on consumers' word-of-mouth recommendation under different product scenarios based on the social influence theory and product difference theory. The research found that the self-strengthening celebrity endorsement mainly through strengthening its normative influence, and thus enhance consumers' word-of-mouth recommendation of hedonic products; Self-deprecating celebrity endorsements enhance consumers' word-of-mouth recommendations for practical products by reinforcing their informational influence. Self-strengthening celebrity endorsement is more conducive to improving consumers' word-of-mouth recommendation of hedonic products; Self-deprecating celebrity endorsements are more conducive to promoting consumers' word-of-mouth recommendations for practical products.
To sum up, linguistic styles (self-deprecating vs. self-strengthening) are increasingly being applied to areas of branding and word-of-mouth recommendation, influencing consumers' brand attitudes by narrowing or alienating the psychological distance. As one of the most extensive forms of online public praise, the above literatures lay a good theoretical foundation for this study to explore the language style of online reviews on online consumers' brand evaluation and purchase intention. This study will further explore the moderating effect of the interaction between publishers and language styles on consumers' brand identity.

Research Hypotheses and Theoretical Models
According to the literature review of reference groups, the influence of reference groups on consumer attitudes and behaviors has been pointed out at home and abroad. As for the evasive group, consumers generally try to avoid keeping a connection with it, resulting in avoidance behavior. In contrast, consumers tend to be more eager groups. In terms of product evaluation, consumers tend to avoid products used by evading groups and give negative evaluations (Berger & Heath, 2008;Chae et al., 2017;Kim et al., 2016), this negative evaluation not only applies to symbolic products (Han et al., 2010) but also includes technology-based products (Arbore et al., 2014). The product use behavior of evading groups will reduce consumers' evaluation of products.
In recent years, some scholars have also introduced reference groups into the study of online comments. Feng Wenting et al. (2019) pointed out that when the "bad group" (avoidance group) "speaks well of the brand," positive word of mouth will lead to consumers' negative stereotypes of the brand. Combined with the phenomenon that "aunt print" affects product sales in marketing practice, this study believes that when positive comments come from different reference groups, consumers' brand evaluation and purchase intention will be significantly different. Specifically, we propose the following hypothesis: H1: when positive online reviews come from avoidant groups, consumers generate lower brand ratings than when they come from desire groups.
H2: compared with the desire group, when positive online comments come from the avoidance group, consumers will have lower purchase intention.
Many studies in social psychology have shown that consumers prefer ingroup and desire group (Escalas & Bettman, 2005) while exist out-group derogation, and avoid association with avoidance group. There are a number of explanatory mechanisms. According to social identity theory, group preference originates from individual social classification and group identity generated on this basis.
In the study of consumer behavior, consumers can, hidden under the condition of contact with other consumers through brand building and even form a social group, a group preference and disadvantageous group psychological basis is between the consumer and the brand of self-brand coupling degree of identity and consumer brands for groups. When consumers believe that the brand can represent them and desire that the group is using the brand, they will strengthen the connection between themselves and the brand, showing stronger brand love and brand attachment. When the evading group is using the brand, consumers will choose to weaken the connection between themselves and the brand if they cannot weaken the connection between the evading group and the brand, or even cause brand switching behavior. (2007) on outgroup shows that avoidance group has significant influence on consumers' self-brand connection, consumer evaluation, product evaluation and consumer choice. Combined with the existing theory, we believe that the internal mechanism is consumers' brand identity. Specifically, when consumers see positive comments from the desire group, they will improve the consistency perception and brand identity between themselves and the brand, and thus have higher brand evaluation and purchase intention. When consumers see positive comments from the avoidance group, they will perceive the strong conflict between the brand image (their own image) and the avoidance group, thus reducing their own brand identity, leading to lower brand evaluation and purchase intention, so as to maintain the consistency level of self-concept. Therefore, we propose the following research hypotheses: H3: brand identity plays an intermediary role in the influence of reference group types on consumers' brand evaluation. Compared with the desire group, when the positive online comments come from the avoidance group, consumers will have lower brand recognition, which will lead to lower brand evaluation.

The research of White and Dahl
H4: brand identity plays an intermediary role in the influence of reference group types on consumers' purchase intention. Compared with the desire group, when the positive online comments come from the avoidance group, consumers will have lower brand identification and thus lower purchase intention.
In the field of sociology, the influence of language style on relationship perception in interpersonal communication has been widely demonstrated (Gong & Li, 2017). In the study of consumer behavior, language style is often used to improve advertising effect and product reputation (Chen et al., 2013;Chung & Cho, 2017). According to existing studies, the use of self-strengthening language style can bring stronger attraction and professional level by exaggerating oneself, so as to improve the topic's authority, social influence and trust degree, mainly with the help of normative influence. The use of self-deprecating language can narrow the psychological distance with the affected person by means of self-deprecation (Gong & Li, 2017), thus achieving the effect of establishing an equal relationship (Gong & Li, 2017), mainly by means of informational influence.
As a kind of language, the application of the language style of positive comments will also bring different influences on the recipients of comments, especially by different types of communicators. In this study, when the avoidant group USES the language style of self-strengthening to make positive comments, its focus is more on its own normative influence, which tends to make the recipient pay more attention to the reviewer (the avoidant group), so the focus is more on the relationship between themselves and the avoidant group. Under the influence of social comparison, consumers at this time will more strengthen the differences and distance between themselves and the avoidance group, and are more likely to depreciate and stay away from the avoidance group. When the avoidance group USES self-deprecating language style to make positive comments, it will produce a certain effect of self-deprecation, and its informational influence will be enhanced. At this point, there are two effects: first, the influence of information is enhanced, consumers will pay more attention to the content of the review itself rather than the reviewer, thus highlighting the "positive" of the positive review; The second is that consumers will generate the perception that they are comparing with the people in a worse situation, thus raising their expectations for positive results and maintaining their positive evaluation (Tesser, 1998), and generating higher self-esteem and positive emotions, consumers will have higher brand evaluation and purchase intention. Therefore, we DOI: 10.4236/jss.2020.82022 283 Open Journal of Social Sciences propose the following hypothesis: H5: the language style of the review moderates the influence of the reference group type on brand evaluation. Compared with the self-reliant language style, when the self-deprecating language style is used in the positive online comments, consumers will have higher brand evaluation.
H6: the language style of the review moderates the influence of the reference group type on brand evaluation. Compared with the self-reliant language style, when the self-deprecating language style is used in the positive online comments, consumers will have higher brand evaluation.
Positive comments from different types of reference groups and their language styles can affect consumers' perception of the brand. In the theoretical derivation above, this study has explained that the use of self-reliant language style and self-deprecating language style by evading groups will have different effects on the brand evaluation and purchase intention of the comment recipients, but the psychological mechanism has not been explained. According to the perspective of social comparison theory, people will determine their own social attributes or construct self-evaluation based on the results of comparison between themselves and others. People also need to use certain social comparison to divide their own ethnic groups, so as to form and strengthen their own group preferences, which is also the basis for the formation of social identity and even brand identity. When the avoidance group USES the self-strengthening language style, its normative influence is strengthened, and consumers will reduce their sense of brand identity for the purpose of weakening their own connection with the avoidance group. When the avoidant group USES self-deprecating language style, its informational influence is strengthened, and consumers will also conclude that there is a relative distance between the avoidant group and the brand based on the downward comparison, which is "not worthy" of the brand. Thus, self-evaluation and positive emotions are maintained, which will generate high brand recognition. Based on this, the following hypotheses are proposed in this study: H7: the mediating effect of brand identity between reference group type and brand evaluation is moderated by the language style of comments. Compared with the self-strengthening language style, when the self-deprecating language style is used in the positive online comments to avoid the group, consumers will have higher brand recognition and thus higher brand evaluation.
H8: the mediating effect of brand identity between reference group type and purchase intention is moderated by the language style of comments. Compared with the self-strengthening language style, when the self-deprecating language style is used in the positive online comments to avoid the group, consumers will have higher brand identification and thus have higher purchase intention.

Theoretical Model
Showing in Figure 1.

Sample and Data Collection
Based on the reference group theory, this study aims to explore the differences in consumers' brand evaluation and purchase intention when consumers see different reference group types (avoidance group vs. desire group) in the face of positive reviews, and introduces brand identity as an intermediary variable to explain the mechanism. There are three problems to be solved: 1) the influence of reference group type on consumer brand evaluation; 2) influence of reference group type on consumers' purchase intention; 3) whether brand identity plays an intermediary role in the influence of reference group types on brand evaluation and purchase intention.
In this study, the inter-group experimental design of reference group type (avoidance group vs. desire group) was adopted, and the experiment was carried out by the method of scenario simulation. All subjects were randomly assigned to one of two groups. This study refers to the research design of Tu Rongting et al. (2018), and manipulates the reference group type by using consumers' shopping situation on a well-known e-commerce platform. Specifically, the subjects were directed into the online shopping interface for choosing dresses, and viewed the relevant reviews after browsing the products. In the desire group (hereinafter referred to as the desire group), participants saw a picture of a young woman and commented on it. In the avoidance group, participants were shown pictures of a middle-aged woman and asked to comment. For the purpose of manipulation, the comments remain the same except for the picture content. After viewing the reviews, the participants were asked to answer related questions about measuring the reference group, brand evaluation, and purchase intention, as well as related questions about demographics. The specific questionnaire is shown in the appendix.
The first study was conducted in a comprehensive university in Guangzhou. Due to the research needs, the subjects of this study were all young women. A total of 100 questionnaires were issued in this study, and 85 valid questionnaires were finally recovered, with a recovery rate of 85%. Among them, there were 39 subjects (45.88%) in the avoidance group and 46 subjects (54.12%) in the desire group.

1) Reliability test
In this study, SPSS22.0 was used to test the reliability of each part of the questionnaire. According to the analysis results, the Cronbach's coefficient of three parts of reference group, brand identity and purchase intention reached 0.742, 0.840 and 0.91, respectively. All constructs had higher than 0.7, and the overall reliability was acceptable.

2) Control inspection
The mean value of the measured term of the reference group is taken as the value of the reference group. In this study, the independent sample T test was used to test whether the difference of the reference group perception between the two groups was significant. As shown in the table, the difference between the desire group and the avoidance group in the perception of the reference group was significant (M desire group = 2.7257, M avoidance group = 2.2156, t (83) = −2.237, p = 0.015), indicating that the manipulation of the reference group was successful.
3) Influence of reference group type on brand evaluation and purchase intention First, the influence of reference group type on brand evaluation. In order to explore whether the positive comments given by different reference groups have a significant impact on consumers' brand evaluation, the study was tested using one-way analysis of variance (ANOVA). In the homogeneity test of variance, p = 0.185 > 0.05, consistent with the premise of anova (see appendix ii). The analysis results show that the reference group type has a significant impact on brand evaluation (F (5, 79) = 3.447, p = 0.001). As shown in Figure 2 in the face of negative and positive online comments, brand evaluation of evading group relationship consumers (M evading group = 4.4853, SD evading group = 1.0038) is significantly higher than that of craving relationship consumers (M evading group = 5.3030, SD evading group = 0.6952), H1 holds. Showing in Figure 2.
Second, the influence of group type on purchase intention. The study was still tested using a single factor analysis of variance (ANOVA). In the homogeneity test of variance, p = 0.210 > 0.05, which conforms to the premise of anova (see appendix ii for details). Variance analysis results showed that brand relationship norms had a significant impact on consumers' purchase intention (F (5, 79) = 11.214, p < 0.001). In the face of positive online comments, the purchase intention of consumers in the avoidance group (M avoidance group = 4.3971, SD avoidance group = 0.9517) was significantly lower than that of consumers in the desire group (M desire group = 5.2727, SD desire group = 0.8110), and H2 was established.
In order to test the mediating effect of brand identity on the influence of reference group types on brand evaluation, this study referred to the study of Hayes (2013) and conducted the mediating effect test using Bootstrap program. Choose model 4 sample size to 5000, under the 95% confidence interval, the mediation effect of brand recognition test results do not contain 0 (LLCI = 0.2974, ULCI = 0.5228), at the same time.
Showing in Figure 3.   Reference group type direct effect on brand evaluation test results did not include 0 (LLCI = 0.1880, ULCI = 0.4714), shows that brand recognition is part of the intermediary role, its effect is 0.4034 similarly, we examine the mediating effect of brand identity on the purchase intention of the reference group type. Results show that under the 95% confidence interval, the mediation effect of brand recognition test results do not contain 0 (LLCI = 0.2777, ULCI = 0.4734), at the same time reference group types of purchase intention directly effect the inspection results also does not contain 0 (LLCI = 0.1536, ULCI = 0.4171), shows that brand recognition is part of the intermediary role, its effect is 0.3640. Furthermore, we use the one-way analysis of variance to explore how brand identity mediates the influence of reference group type on brand evaluation and purchase intention, that is, take reference group type as the factor and brand identity as the dependent variable for testing. The homogeneity test results showed that the variables were suitable for anova (p = 0.649 > 0.05, see appendix ii for details). According to the variance analysis results, the reference group type Open Journal of Social Sciences has a significant impact on brand identity (F (5, 79) = 6.494, p < 0.001). As shown in Figure 3, compared to see positive comments from desire for groups of consumers (=5.0303 M eager to brand identity, SD is eager to brand recognition = 1.132), and see positive comments from avoidance groups of consumers have a lower brand recognition (=4.4706 M to dodge brand identity, SD to dodge brand recognition = 0.992).

The Moderating Effect of Language Style
Based on the conclusion of study 1, this study further explores the moderating effect of comment language style (self-strengthening/self-deprecating) on brand evaluation and purchase intention of reference group type.
The inter-group experiment design of 2 (reference group type: desire group/avoidance group) × 2 (language style: self-strengthening type/self-mockery type) was adopted in the study, and the experiment was carried out by the method of scenario simulation. Based on the experiment of study 1, this study, referring to the study of Huang Minxue et al. (2018), divides the language expression in the positive comment scene into self-strengthening type and self-deprecating type, and other contents are basically consistent with study 1.
The second study was conducted in a comprehensive university in Guangzhou, a total of 240 questionnaires were issued, and 186 valid questionnaires were finally recovered, with a recovery rate of 93%. Among them, 38 people (20.43%) in the avoidance * group, 53 people (28.49%) in the avoidance * self-improvement group, 38 people (20.43%) in the desire * self-mockery group, and 56 people (30.11%) in the desire * self-improvement group. The procedure of this study is basically the same as that of study 1. The four groups were first asked to read a simulation of the corresponding type of manipulation. The difference from study 1 was that the relevant comments under the two pictures were different (as shown in Figure 4). The focus of the comments the self-improvement group saw was "I'm beautiful. This dress really suits me." The self-deprecating group saw comments that centered on "the dress is beautiful. I don't really deserve it." This

Data Analysis (Study 2)
1) Reliability test Similar to study 1, this study used SPSS22.0 to test the reliability of each part of the questionnaire. According to the analysis results, Cronbach's coefficient of reference group type, language style type, brand identity and purchase intention reached 0.710, 0.873, 0.886 and 0.942, respectively. All constructs had higher than 0.7, and the overall reliability was acceptable. 2

) Control inspection
In order to ensure the accuracy of the research results, this study conducted manipulation test on the reference population type again. The mean value of the measured term of the reference group is taken as the value of the reference group. In this study, the independent sample T test was used to test whether the difference in the perception of the reference group was significant. the difference between the desire group and the avoidance group in the perception of the ref- Showing in Figure 5: Similarly, we used anova to test the moderating effect of language style on purchase intention of reference group types. First, homogeneity test of variance was conducted, and the results showed that the data were suitable for anova (F (3, 182) = 0.571, p = 0.635 > 0.05) (see appendix ii for details). Then, this study conducted multivariate anova with reference group type and language style as factors and purchase intention as dependent variables (see appendix ii for details). The results showed that the main effect of reference group type on purchase intention was significant (F (3, 182) = 11.708, p = 0.013), which again supported the conclusion of study 1. And language style is have a significant effect on purchase intention (F (3, 182) = 10.828, p = 0.024), and the types of reference groups and the interaction effect of language style significantly influence on purchase intention (F (3, 182) = 15.063, p = 0.007), explain language style for reference group types had regulation impact on purchase intention (as shown in Figure 6), H6 was established.
Showing in Figure 6.

Moderated Mediation Effect Test
The above studies have explored the main effect of reference group types on Figure 5. The moderating effect of language style on the influence of reference group type on brand evaluation. brand evaluation and purchase intention, the mediating effect of brand identity and the moderating effect of language style. Finally, this study used the Bootstrap program to examine the moderating effect of language style on the mediating effect of brand identity between reference group type and brand evaluation/purchase intention. First, we examine the moderating effect of language style on the mediating effect of brand identity between reference group type and brand evaluation. Mod-el7 was selected and the sample size was 5000. Within the 95% confidence interval, we found that when the avoidance group used the self-reliant language style, the mediating effect of brand identity on the brand evaluation of the reference group was not significant (LLCI = −0.2137, ULCI = 0.1937, the interval included 0). When the avoidance group USES self-deprecating language style, brand identity has a significant mediating effect on the brand evaluation of the reference group type (LLCI = 0.0454, ULCI = 0.4653, the interval does not include 0). The confidence interval of direct effect of reference group on brand evaluation does not include 0 (LLCI = 0.0761, ULCI = 0.6877), indicating that brand identification plays a partial intermediary role in the influence of reference group on brand evaluation at this time, with effect size of 0.217 (see appendix ii for details).
Secondly, we examine the moderating effect of language style on the mediating effect of brand identity between reference group type and purchase intention. The inspection procedure is the same as above. The results show that when the avoidance group USES the self-strengthening language style, the mediating effect of brand identity on the purchase intention of the reference group is not significant (LLCI = −0.2413, ULCI = 0.2017, the interval includes 0). When the avoidance group USES self-deprecating language style, the mediating effect of brand identity on the purchase intention of the reference group type is significant (LLCI = 0.0423, ULCI = 0.4877, the interval does not include 0). The confidence interval of direct effect of reference group type on purchase intention does not include 0 (LLCI = 0.1576, ULCI = 0.7973), indicating that brand identity plays a partial intermediary role in the influence of reference group on purchase intention at this time, and the effect size is 0.2279 (see appendix ii for details).

Conclusion
The essay examined the influence of different types of reference groups on consumers' brand evaluation and purchase intention under positive review conditions. The results show that when positive online reviews come from avoidant groups, consumers will have lower brand evaluation and purchase intention compared to desire groups. Among them, brand identity plays a part of intermediary role in the middle. That is to say, when positive online comments come from avoidant groups, consumers will have lower brand recognition, which will lead to lower brand evaluation and purchase intention. On the basis of study 1, experiment 2 introduced the language style type of comments as the moderating variable, and tested that the language style of comments can adjust the influence Open Journal of Social Sciences of reference group type on brand evaluation and purchase intention. That is, compared with self-strengthening language style, consumers will have higher brand evaluation and purchase intention when they avoid self-deprecating language style in positive online comments. Results from more than one factor analysis of variance (MANOVA), brand recognition in the types of reference groups and the intermediary effect between the brand and purchase intention are comment on the language style of adjustment, that is, a language style compared to self-improvement, when positive online comments to evade groups use self-deprecating language styles, consumers will have higher brand recognition, which has a higher brand evaluation and purchase intention.

Management Implications
This study provides some management references for enterprises to manage positive online reviews, and enterprises need to realize that positive online reviews do not always bring positive effects. From the perspective of the boundary conditions under which positive online comments exert negative effects, enterprises should pay more attention to the group identity characteristics of the comment publishers themselves and the matching degree with the evaluated objects (products or services), while the consumer characteristics of published comments are often ignored by enterprises. Especially in the actual marketing practice, when the e-commerce promotion node, many platform merchants often adopt "praise reward plan" in order to get more positive online reviews. For example, taobao merchants encourage consumers to give favorable comments after buying products by means of cash return or coupons. From the perspective of enterprises, marketers go to great lengths to seek favorable comments and invest a lot of human and material resources to increase positive online reviews from consumers, which often backfire. At the same time, this study puts forward that it is necessary for enterprises to understand the mechanism behind the phenomenon of "favorable comment is not necessarily favorable comment" and the boundary conditions of its influence, and puts forward the following Suggestions for e-commerce platform merchants to play the correct role of favorable comment: 1) Make use of big data algorithm technology to distribute online comments accurately The big data algorithm function of the e-commerce platform can help enterprises distinguish the comment publishers. Today, companies often indiscriminately encourage all consumers to post positive online reviews of their products, but sometimes they create negative marketing effects. Therefore, choosing the right word of mouth publisher is extremely important. On the one hand, with the help of machine algorithm analysis, enterprises can choose consumers with a high matching degree with the brand image. After the purchase of target consumers, they can use reward red packets to encourage them to publish positive online reviews, and give priority to "print pictures + praise". The technology can Open Journal of Social Sciences also be used to screen the online comments of consumers to focus on promoting the positive online comments that are easy for consumers to identify with. On the other hand, enterprises can conduct accurate personalized marketing according to the individual preferences of different target groups of consumers, and display differentiated positive online comments for different groups of consumers. To be specific, if the brand focuses on young people and sells girls' clothes, it should show more favorable comments from the young female consumers + print comments, and weaken and reduce the print comments of middle-aged women.
2) Systematically implement the management of users' online comments based on ERP (enterprise resource planning) Many studies have pointed out that positive online reviews can build a good brand image for enterprises, enhance brand evaluation, help merchants to improve product sales and other positive effects. However, in order to avoid negative effects of positive online reviews due to reference groups, enterprises should pay attention to effective management of existing reviews. For example, for large enterprises, managers can set up a special comment management team to timely control online comments posted by consumers, so as to achieve the best marketing effect of positive word of mouth. For small and medium enterprises, the scope is limited in resources, is not necessary to encourage all consumers print hair high praise, sometimes "nothing more than" the more the better, marketing value is bigger, often as far as possible in the limited within the scope of management to manage the existing high quality reviews, reduce unnecessary" negative "praise.
3) Conducive to "brand community" for social marketing The simplest, yet most valuable thing a business can get from a community is a highly engaged user or consumer. For example, the successful enterprise marketing practice shows that through the brand community, xiaomi collects user needs and allows users to participate in the preliminary research, functions, style design and other aspects of the project or product. Users will feel that they are part of it, and they will be proud of it and feel that they are special. When we feel we belong to a "community," that sense of belonging makes us happy. The same applies to our brand. When the brand community around the product is gradually formed, the long-term activities of users will give them a sense of brand identity and sense of belonging, and even make them close friends. For example, the identification of xiaomi community members with the trend and symbol represented by Xiaomi brand will make users feel a sense of belonging, which will make users feel warm and united. And, more importantly, in the brand community, people's social group is mainly derived from the reality of friends or strangers are consistent with their own interests, consumers can borrow by sharing, with pictures and words to express emotion, comment, thumb up and forward, consumers in the share and participate in the process of strengthening brand identity to be fond of, each other to strengthen consumer loyalty and vis-cosity.
Second, in the practice of marketing management, this study provides a reference for businesses to recommend users to post positive online comments.
Specifically, positive online comments posted, the enterprise shall, according to comments language style type (self-strengthening vs self-deprecation) comment on the difference of control, comment on self-improvement type of target consumers should be placed at the top, on the target consumers to purchase products after the release of self-renewal type style "printing + high praise" is after, even have the ability to businesses can take a red envelope reward for non target group delete self-strengthening "praise". From the perspective of e-commerce trading platforms, under the circumstance that merchants have no right to delete the online comments of non-target groups, they should encourage more non-target groups to publish self-deprecating online comments, just as the saying goes: "either don't speak, speak well".

Research Limitations and Future Research Prospects
Starting from the marketing phenomenon, this study explores the negative effect of positive online comments and the psychological mechanism behind it from the perspective of brand identity of review publishers. However, there are still some limitations in this study, which need to be further explored. Experiment material limitations: main types in exploring reference group for the positive effect of online reviews adjustment, in order to close to the marketing practices, to the more real and let the participants a real group of perceptual desire and avoidance, this study selected the apparel products (prone to compare the experimental materials), the future can continue to test whether the product type, for the purposes of this study have an effect, for example, to explore different product types, such as cars, mobile phones, luxury, etc.
Limitations of data methods: due to the low internal and external validity of the experimental method in this study, future research can be combined with the questionnaire survey method to increase the application of the conclusion.
Limitations of explanation mechanism: in exploring the influence of review publishers on the marketing effect of positive online reviews, this study focuses on exploring consumers' brand identification. Considering the multi-dimension of cognition among social groups, follow-up studies can further explore how consumers' different stereotyping of evading groups, such as low perceived warmth or low perceived ability, affect the marketing effect of positive online reviews. Further research can also explore the moderating effect of self-construction on positive online plane theory. In the study of social comparison theory, this study is only used as an explanatory mechanism, and has not been further verified to see whether upward comparison will generate jealousy and other emotions, and whether downward comparison will generate disgust and other emotional mechanisms. This study focuses more on the influence conditions of avoidance groups and does not delve into the influence of desire groups.
Limitation of language style: meanwhile, in the future, we can further explore the influence of different language styles on positive online comments, such as metaphor, exaggeration, statement, etc. In addition, when positive online comments have negative effects that threaten product sales due to avoiding groups, the countermeasures of enterprises should be taken. The exploration of this issue is conducive to enterprises to better deal with the negative effects of positive online comments, and to further extend the research on the marketing effects of online comments.