The influence of electronic word of mouth on Instagram users: An emphasis on consumer socialization framework

Abstract The present study aimed at exploring the extent to which users model their behavior—as well as their brand attitudes, their perception of relationship quality, their use of Instagram, and the number of brands they follow—on other users. This distinction can help marketers and social media experts discover the most likely perspectives of brand management. The statistical population was composed of the users of the Instagram social network, out of which 384 individuals were sampled by convenience non-probable technique. The hypotheses were tested by the logistic regression analysis. The results show that peer communications, brand attitude, brand relationship quality, and Instagram usage influence the manifestation of eWOM in the Instagram social network. The paper is concluded with some recommendations to marketers in the context of the Instagram social network.

ABOUT THE AUTHORS Narges Delafrooz is an assistant professor at the Islamic Azad University, Rasht Branch, Faculty of Management and Accounting, Department of Business Management. She received her PhD degree in Consumer Behavior from University Putra Malaysia (UPM) in 2009. Her current research interests include consumer behavior, online shopping, and management marketing. She has publications in national and international journals of repute in the areas of consumer behavior and management.
Yalda Rahmati is an assistant professor at the Islamic Azad University. She is expert in the Business Management and has publications in national and international journals of repute in the areas of consumer behavior and management.
Mehrzad Abdi received BA degree and Master's Degree in Business Management from the University Azad, Iran.

PUBLIC INTEREST STATEMENT
Due to high user engagement rate, Instagram is also a valuable social media marketing tool. The most important consumer socialization variables are with an emphasis on peer communications, brandrelated factors, and the use of the social network for eWOM among the users of Instagram. Social media facilitate peer communications and show a new form of consumer socialization that affects eWOM. When consumers follow brands in a social network, the brands may create their desirable attitudes and build loyalty in consumers. Also, engagement through social media platforms can play a very important role in building brand relationship quality. By studying how users imitate the behavior of other users and their attitudes towards brands, they can distinguish the users' perception of relationship quality, their use of social networks, the number of brands that they follow, and the individuals who participate in brand-related eWOM. This distinction can help marketers discover the most likely perspective of brand management.

Introduction
Despite the fact that information technology (IT) is gaining an increasing power, IT users and managers are still struggling with a long-lasting problem-final users are mostly unwilling to use information systems (Jalilian, Ebrahimi, & Mahmoudian, 2013). Electronic Word of Mouth (eWOM) can be broadly defined as"any positive or negative statement made by potential, actual, or former customers about a product or company, which is made available to a multitude of people and institutions via the Internet (Chu & Sung, 2015). eWOM is a relatively new research domain that has been evolved in the last decade. Some researchers have integrated the theories that had been put forth to explain eWOM communications (Gao & Bai, 2014). This has provided the researchers with valuable knowledge to identify the underlying mechanism of eWOM communication happening through social networks.
With the emergence of social technologies available on the Internet and to the users of smartphones, social media like Facebook, Twitter, and Instagram have enabled consumers to get involved in brand-related eWOM and appear as an advertisement tool in electronic marketing and commerce (Chu & Sung, 2015). It has been documented that social media facilitate eWOM relationships among the consumers (Jeff, Jennifer, Catherine, & Elke, 2014) and enable marketers to interact with their consumers and establish long-term relationships (Chu & Sung, 2015).
Today, Instagram is one of the faster-growing social apps for brand buildings in marketing. Instagram is an important part of our life, especially for online business. Individuals actively use Instagram on a daily basis, showing an approximate growth rate of 2% between 2016 and 2017 (Chaffey, 2017). A survey research program conducted by Social Media & User-Generated Content found that 73% of responding internet users aged 13 to 24 years used the social photo-sharing app Instagram. A total of 76% of respondents currently use Facebook, followed by Instagram use with a 73% share (https://www.statista.com). In fact, modern-day consumers are becoming highly dependent on social media, increasingly utilizing the various platforms on this ecosystem (Hanna, Rohm, & Crittenden, 2011), which enable them to gather information about brands and subsequently, shape their purchase intentions (Sharma & Sheth, 2010).
Instagram has had the fastest growth rate among all social media so that over 400 million users use it and 80 million photos are shared every day (The statistics Portal). Also, 71% of the globally known brands use this social network. The capability of sharing videos and sending photos through direct messages was added to this app in 2013 (Guidry, Messner, Jin, & Medina-Messner, 2015). Guidry et al. (2015) analyzed the Instagram images for the 10 largest fast-food companies in the world and stated that Instagram has had a very fast growth rate among all social networks. According to their results, public relations professionals should place Instagram after Facebook and Twitter as a challenge in the agenda of their social media strategy. These professionals perceive Instagram to be an effective way to create brand personality and loyalty (Guidry et al., 2015). According to the general concept of socialization, the theoretical framework of consumer socialization assumes that individuals gain their attitude and behavior pattern partially from their interaction of socialization parameters and their learning, such as parents and peers. The social learning view emphasizes the external sources of socialization, such as peers and parents (Chu & Sung, 2015).
Instagram is regarded as an important marketing tool in social networks because of its advantages and features. Most globally known brands use this network in their advertisement programs. Due to the apps visual nature and high user engagement rate, Instagram is also a valuable social media marketing tool (https://www.statista.com). The potential impact of eWOM on Instagram expands exponentially. With the growing popularity of social media, interests have been provoked for customer socialization through the websites of social media in recent years. Social media facilitate peer communications and show a new form of consumer socialization that affects eWOM (Wang, Yu, & Wei, 2012). However, no research has yet examined the factors that influence Instagram users' decisions to engage or not to engage in eWOM on the site. While a few studies have analyzed brand-focused eWOM patterns on Instagram (Ferrara, Interdonato, & Tagarelli, 2014;Park, Ciampaglia, & Ferrara, 2016), it is unclear what makes consumers willing or reluctant to participate in brand conversations on Instagram. To fill this gap, the objective of the current research is to demonstrate that Instagram is a socialization agent that facilitates eWOM by examining the factors that discriminate Instagram brand followers' decisions to engage in eWOM behaviors on the site. Therefore, the concept of eWOM is extremely important for a variety of online consumption communities, especially for Instagram as brands seek new ways to develop long-term relationships with their consumers and enhance customer equity (Kim & Ko, 2012).
Accordingly, the present research aims to identify the most important consumer socialization variables with an emphasis on peer communications, brand-related factors, and the use of social networks for eWOM among the users of Instagram.

Review of literature and conceptual model
The main objective of this study was to demonstrate that Instagram is a socialization agent that facilitate eWOM. To achieve this goal a conceptual framework that delineate the relationships among the key variables was developed to identify factors that discriminate users' decisions to engage in brand-related e-WOM on Instagram. Figure 1 presents the conceptual framework.

Electronic word of mouth (eWOM)
eWOM is broadly defined as "any positive or negative statement made by potential, actual, or former customers about a product or company (Dinh & Mai, 2016), which is made available to a multitude of people or institutions via the Internet" (Chu & Sung, 2015). It refers to all informal communications of consumers through the Internet-based technology with respect to the applications or features of certain commodities or services and their retailers (Schmäh, Wilke, & Rossmann, 2017). The growing acceptance of online social networking services by customers and businesses calls for new knowledge to understand their effect on consumer behavior, and more importantly, how eWOM occurring via these sites influence consumer decisions (Kudeshia, Sikdar, & Mittal, 2016). eWOM can be regarded as the counterpart of the traditional inter-personal communications within the new generation of the cyberspace (Jalilian et al., 2013). The advantage of eWOM is that it attracts many customers with the least cost and high effectiveness (Sheu & Chu, 2017). Marchand, Hennig-Thurau, and Wiertz (2017) investigated the variables of social networks, digital games, websites, and WOM and concluded that WOM was the most effective in social networks because of its high number of audience and extensive applications. Usman, Fatimee, and Sajjad (2014) conducted an empirical analysis of the factors underpinning the adoption of e-commerce with a focus on consumers' intentions to use e-commerce. Erkan and Evans (2016) concluded that quality, reliability, and information requirements were the key factors affecting the purchase intention via website advertisements. Lueg and Finney (2007) report that online purchase behaviors largely depend on online peer communications. Ntale and Ngoma (2019) mentioned that word of mouth had a mediator role in the relationship between relationship marketing and customer loyalty. Ozdemir, Tozlu, Şen, and Ateşoğlu (2016) found the effect of eWOM factors and components on consumers' purchase intentions and reported a significant relationship between WOM factors and purchase intentions. Hudson, Roth, Madden, and Hudson (2015) focused on the variables of WOM, interactions in social networks, and emotional engagement and concluded that advertisement in social networks was very effective in engaging customers emotionally so that it made them interested in WOM.

Consumer socialization
Consumer socialization is defined as the process of learning skills, knowledge, and attitudes about consumers (Ward, 1974). Family, school, and peers are considered the key factors in socializing an individual as per their priorities in the individual's life. The study of how consumers are developed through social interaction. Consumer socialization is defined as a lifelong process through which individuals acquire skills, knowledge, habits, attitudes, and values that affect their "present" and "eventual" behavior as consumers in the marketplace (Baumrind, 1978;Senthil Kumar & Ramachandran, 2011). The consumer socialization approach was first introduced as a tool into the study of consumer behavior by Bindah and Othman (2011). Therefore, consumer socialization refers to the social process "by which young people acquire skills, knowledge, and attitudes relevant to their functioning as consumers in the marketplace" (Ward, 1974;Bindah & Othman, 2011). Mousavijad and Payvandi (2017) studied the effect of socialization factors on secondary and high-school student consumers' decision-making methods. The results showed that the teenage consumers who used utilitarianism decision-making were affected only by their parents and peers, whilst the teenagers who used sociable decision-making were influenced by both peers and media, and those who used unsuitable decision-making method were influenced by media and school.

Peer communications
Given the essential characteristics of human being, interactions with individuals and their relationships with their peers occur to satisfy their sociological and psychological needs (Ward, 1974;Bindah & Othman, 2011). Peer communications are defined as explicit interactions with peers with respect to commodities and services. This is regarded as an important factor of socialization within the framework of consumer socialization. Peer communications have two aspects: spoken or reinforcement (users are reinforced by their peers to follow a certain brand) and unspoken or modeling (users imitate the behaviors of their peers as to a certain brand) (Lueg & Finney, 2007). Wang et al. (2012) state that the socialization of consumers via peer communications concerning products existing in social media is positively related to their attitudes towards the products. Wang et al. (2012) studied peer communications via social media websites and their effect on attitude towards product and purchase decisions. Elaheebocus, Weal, Morrison, and Yardley (2018) identified three main areas with regard to the effects of social media features on users, namely, usage, participants' perception, and behavioral outcome. The results showed that consumers' purchase intentions were influenced by their peers directly and by the strengthening of their participation in product purchase indirectly. In addition, De-Gregorio and Sung (2010) argue that peer communication is a vigorous factor in predicting adult consumer attitudes and their behavioral responses to product placement.
Social media facilitate peer communications with one another and manifest a new form of consumers' socialization that influences users' eWOM (Chu & Sung, 2015). Accordingly, it is hypothesized that H1. peer communications influence eWOM among the users of the Instagram social network, H1a. peer communications influence brand following activity on the Instagram social network, and H1b. peer communications influence brand tagging activity on the Instagram social network.

Brand attitude
Attitude may be desirable or undesirable and/or positive or negative. Most authors have consensus that attitude is acquired (Maheri & Hosseini, 2015). Brand attitude can be defined as "a consumer's overall evaluation of a brand" (Mitchell & Olson, 1981). It is an assessment of favorable or unfavorable responses to brand-related stimuli or conviction (Murphy & Zajonc, 1993). Schivinski and Dąbrowski (2014) found that effect brand attitude of product, whereas user-generated content significantly influences the purchase intention of the reviewed product. When consumers follow brands in a social network, the brands may create their desirable attitudes and build loyalty in consumers. Therefore, it is important to study the attitudes of brand advocates about brands and the quality of the brand relationship (Chu & Sung, 2015). Marketers have been obliged to undertake market research because of the need to explore consumers' attitude, how they think about their product, and how they respond given their attitude, positive or negative (Ranjbaran, Jamshidian, & Dehghan, 2007). However, only a few studies have examined how brand attitude affects the quality of brand relationship (e.g. Chang & Chieng, 2006;Lee & Kang, 2012;Morgan-Thomas & Veloutsou, 2013). Accordingly, it is hypothesized that H2. brand attitude influences eWOM among the users of the Instagram social network, H2a. brand attitude influences brand following activity on the Instagram social network, and H2b. brand attitude influences brand tagging activity on the Instagram social network.

Brand relationship quality
The concept of brand relationship has been derived from the theory of relationship market research whose final goal is to strengthen a vigorous relationship and turn uninterested customers into loyal customers. High relationship quality means that the customer can rely on the future performance of the service provider at a level similar to the already delivered satisfactory service (Al-Tit, 2015). Erkan and Evans (2016) introduced information quality, information credibility, information needs, and attitude towards information as four factors underpinning eWOM advertisement. Therefore, the mental probability formed in users about the usefulness of a system depends on the information provided by it so that when a system provides users with more precise and up-to-date information, it is more likely to be used and this will positively influence the experience of the users (Usman et al., 2014). WOM advertisement refers to the inter-personal relationship among the consumers about their personal assessments and experiences of a firm or a product (Zhang, Craciun, & Shin, 2010). Tsao and Hsieh (2012) studied the brand relationship quality and its impact on eWOM. They showed that brand commitment might build positive eWOM whilst brand satisfaction and trust did not have this positive impact. Another theoretical perspective that plays a critical role in the process of consumer socialization and may be able to explain the engagement of brand advocates in eWOM is brandconsumer relationships. Research shows that engagement through social media platforms can play a very important role in building brand relationship quality (Kudeshia et al., 2016). Earlier studies have used the concept of relationship in the consumer-brand domain and have presented the brand as a partner in the relationship (Chu & Sung, 2015). Accordingly, it is hypothesized that H3. brand relationship quality influences eWOM among the users of the Instagram social network, H3a. brand relationship quality influences brand following activity on the Instagram social network, and H3b. brand relationship quality influences brand tagging activity on the Instagram social network.

Instagram usage
The profile pages of a brand are the tools of eWOM for the brand to provide brand information, answer the questions about the product, and fight with negative eWOM so as to build awareness, interest, brand image, and profit for brand owners (Parry, Kawakami, & Kishiya, 2012).
By investigating users' behavior in following other users, their attitudes towards brands, their perception of quality, their use of the Instagram social network, and the number of brands followed by them can shed light on the individuals who participate in brand-relative eWOM in the Instagram social network and distinguish them from non-participants. This understanding may help marketers and social media experts discover the most likely brand management perspective. Not only do these likely users participate in brands and involve their behaviors in those brands via marketers, but they are also customers that are likely to influence others. As Kim and Hanssens (2017) recommend, it is vital to identify the likely individuals who are effective in eWOM in social media in order to motivate them to extend their brand-related positive eWOM.
Given the potential of social networks as a factor of socialization and an instrument of eWOM about the brands, people who frequently use social networks follow more brands and are involved in eWOM to a greater extent. The number of interested brands may affect the likelihood of participation in eWOM; that is, the more the number of the brands a user uses, the more the chances to converse on them and receive the marketing messages that are transferred (Chu & Sung, 2015). Instagram was founded in 2010 and was purchased by Facebook in 2012. Instagram has had the fastest growth rate among all social media so that over 400 million users use it and 80 million photos are shared every day (Official Website of Instagram, 2015). Also, 71% of globally known brands use this social network. The capability of sharing videos and sending photos through direct messages was incorporated into this medium in 2013 (Geurin-Eagleman & Burch, 2015). Accordingly, it is hypothesized that H4. Instagram usage influences eWOM among the users of the Instagram social network,

H4a. Instagram usage influences brand following activity on the Instagram social network, and
H4b. Instagram usage influences brand tagging activity on the Instagram social network.

Methodology
In the present study, the method was used to develop a theoretical framework and review the literature, and a questionnaire was employed in field method (online) for data collection from the statistical population. So, the standard questionnaire was derived from Chu and Sung (2015) in which it has been validated. The data were analyzed by the logistic regression test in the SPSS software package. Since the dependent variable was a nominal dichotomous one (1 = following activity and 2 = tagging activity), we employed the dichotomous nominal logistic regression to determine the effect of independent variables on the dependent variables. The reliability of the research instrument was, also, estimated by Cronbach's alpha. As is evident, Cronbach's alpha is greater than 0.7 for all research constructs, implying their reliability. The statistical population was composed of Instagram users. Finally, 384 individuals were sampled by convenience non-probable technique. Among all respondents, 198 (52%) were female and 186 (48%) were male. Their age ranged from 18 to 45.

Output of block (0)
The outputs of the logistic regression model include two blocks (0 and 1). Block (0) merely displays input data, but block (1) is the main block of logistic regression and presents the results of the regression model. Table 1 shows the results of block (0) in the logistic regression. In this block, no step was run for inputting the data in the model. In this block, using the stepwise regression analysis, the independent variables are included in the model in case they are significant.  Table 1 shows that the variable of the following can account for 88% of the variance of the dependent variable (following activity) and the variable of tagging can account for 80.7% of its variance. Now, we deal with the significance status of the independent variables of the regression equation in Table 2.
The results in Table 2 show that all independent variables (peer communications, brand attitude, brand relationship quality, and Instagram usage) can predict the variance of the dependent variable at the p < 0.05 level. In other words, the variables that were included in the regression analysis can capture the variance of the dependent variable (tagging activity).

Output of block (1)
The main output of the logistic regression analysis is the output of block (1), which is used to interpret the results of the logistic regression. This output presents the results of the logistic regression for each stage. We used pseudo-coefficient of determination to account for the goodness-of-fit of the logistic regression model. This coefficient is composed of two coefficients of determination. These are the approximations of R 2 in linear regression and vary in the range of 0 and 1. Table 3 presents the results for these coefficients.
The results for pseudo-coefficient of determination imply that at the first stage that all significant independent variables were included in the regression model, Cox-Snell and Nagelkerke coefficients were 0.267 and 0.513 for following activity and 0.339 and 0.543 for tagging activity, respectively. This means that the four independent variables of the study could capture 26-51% of the variance of the dependent variable of the following and 33-54% of the variance of the dependent variable of tagging. Table 4 summarizes the role of each variable in the model and show which variables remained in the model after running the logistic regression. This table interprets the results of significance and indicates the impact of each independent variable on the dependent variable. Table 4 is interpreted using the B and Wald statistics and the significance level. The B-statistic is the un-standardized coefficient of regression impact or the estimated coefficient with standard error. The Wald statistic was the most important statistic for testing the significance of the impact of each independent variable. This is equivalent to t-statistic in the linear regression model. The odds ratio statistic refers to the change in the adoption of the dependent variable as per one unit increase in the independent variable. This ratio is equivalent to standardized regression coefficients in linear regression. When the odds ratio is smaller than one, the increase in the value of the independent variable results in the decrease in the likelihood of the dependent variable occurrence (negative impact) and vice versa.

Discussion and conclusion
With respect to Hypothesis 1, no significant relationship was found between peer communications and the following activity, but this variable was significantly related to the tagging activity. The current investigation extends the existing communication literature by examining the role of peer communication in social media, and specifically with regard to eWOM. These results support recent propositions that media user draws a measurable value with eWOM (Chu & Sung, 2015;Elaheebocus et al., 2018;Gvili & Levy, 2018;Wang et al., 2012). Thus, peer communication is an important theoretical construct that helps explain brand-related information sharing in Instagram. In the case of Hypothesis 2, it can be observed that positive or negative attitude of individuals has a significant positive relationship between the following activities. On the other hand, there is no relationship between positive attitude and tagging activity whereas a significant positive relationship was observed between negative attitude and tagging activity. Our findings provide empirical support that Instagram users with a more favorable attitude toward the brands they follow. Past research emphasizes the importance of brand attitude and brand relationship quality in the context of consumerbrand relationship (Chu & Sung, 2015;Guidry et al., 2015;Maheri & Hosseini, 2015). Chu and Sung (2015) found a significant positive relationship between brand attitude and eWOM. Guidry et al. (2015) reported that Instagram was an effective method to build brand personality and loyalty. Maheri and Hosseini (2015) stated that individuals shared their positive attitude more than negative attitudes. Thus, it can be concluded that the results of earlier studies are somewhat consistent with our findings.
The results about Hypothesis 3 reveal that brand relationship quality is related to both following activity and tagging activity significantly and positively. Higher perceived relationship quality with those brands is more likely to motivate participants to create and spread brand-related information on Instagram. That is, engaging in eWOM about brands that they like and they have a relationship with can be considered as a means of connecting the self and the brand. Thus, when a brand is highly associated with the self (high relationship quality), individual consumers' willingness to engage in eWOM is greater (Chu & Sung, 2015;Kim, Sung, & Kang, 2014;Tsao & Hsieh, 2012). This is in agreement with a study on Twitter in the US in which Chu and Sung (2015) found a significant positive relationship between brand relationship and tweeting/not-tweeting behavior. Also, in their study on the relationship between brand relationship quality and eWOM, Tsao and Hsieh (2012) found that brand commitment may have a positive relationship with WOM.
The results of logistic regression equations for Hypothesis 4 show a significant positive relationship between Instagram usage and following activity, while this variable has no relationship with tagging activity. The same pattern of responses also predicted posting behavior indicating that brand engagement occurs both inward toward the brands (tweeting brand information). Our findings offer additional support for a number of recent studies (e.g. Chu & Sung, 2015;Kim et al., 2014) suggesting that Since the respondents of this study were mainly from younger age groups, i.e. up to age 35, the results suggest that managers should engage with the young respondents on Instagram, which are popular among the age group.

Managerial implications
Our findings provide marketers and social network activists with useful managerial information. By studying how users imitate the behavior of other users and their attitudes towards brands, they can distinguish the users' perception of relationship quality, their use of social networks, the number of brands that they follow, and the individuals who participate in brand-related eWOM. This distinction can help marketers and social media experts discover the most likely perspective of brand management. Given the substantial role of social networks in brand marketing, the following recommendations can be drawn from the hypotheses that were tested: • It is recommended to social media marketers to identify the possible effective people in eWOM and encourage them to expand their eWOM behaviors concerning a certain brand. It should be noted that to be successful in a brand advertisement in social media, individuals who follow us are of crucial importance. Thus, it is imperative to find effective people in social networks. In fact, we aim at finding individuals with the highest impact on product advertisement.
• It is recommended to the marketers to provide users, especially opinion leaders, with content that is important enough to them. This is a very effective approach to encourage users to exhibit eWOM behavior across social networks, such as Instagram.
• Instagram is a software tool for image sharing. The software has audiences that look for attractive photos. Hence, marketers are recommended to share attractive and comprehensive images and designs to provide better information about their products. This will encourage users for eWOM. Examples are eWOM messages that show customers' high perceived quality of the products or comparative eWOM that compel customers to purchase certain products and prefer them over competitive products.
• In publishing informing ads about a brand, it should be tried to use high-quality advertisements with unique designs, and unnecessary savings or low-quality advertisements should be avoided as they can be anti-advertising and impair the brand value.
It is important to be noted that people who are posting frequently might be considered as spammers and will actually not be trusted. Future research should take the content of posting into consideration and investigate consumers' perception of those retweets. However, if consumers trust those who post a lot, identifying those most engaged with brands is the first and most important step in developing a successful Instagram campaign. Angelis, Bonezzi, Peluso, Rucker, and Costabile (2012) found that self-enhancement is a unique motive that explains when positive versus negative WOM is more likely to occur. Future research could examine the effect of selfenhancement on eWOM behaviors in Instagram and explore whether this motive shapes Instagram brand followers' posting/no-posting patterns.