Determinants of online brand communities ’ and millennials ’ characteristics: A social influence perspective

Online communities have evolved to allow larger numbers of individuals to interact with other users to form a collective virtual environment influenced by members within the community. Existing studies on online brand communities (OBCs) tied millennials ’ participation and interactions to a unidimensional view. Specifically, OBCs scholars generally aggregate individual millennials ’ participation and commitment, ignoring the variance among the demographic cohort. Our exploration challenges not only the existing ensemble interpretation within studies of OBC but also the characterisation of millennials ’ burgeoning participation in OBCs. Unlike other competing epistemologies, the authors developed a conceptual framework that links a holistic set of OBCs ’ characteristics (brand sentiment, identification with source, affirmative experience, conspicuous effect) to consumers ’ perceptions in the fashion sector. Drawing on social influence theory along with a constructivist perspective, we conducted fine ‐ grained in ‐ depth interviews to explore millennials ’ participation in online communities and brand perceptions in the fashion industry. The main findings reveal four categories of customer engagement in OBCs (bias situators, sugar ‐ coaters, rationalisers, judgmentalists). These key categories are explored to create a framework for future research in this area, and further contribute to the field of online brand engagement, particularly in the fashion industry.


provided that
• the authors, title and full bibliographic details of the item are cited clearly when any part of the work is referred to verbally or in the written form • a hyperlink/URL to the original Insight record of that item is included in any citations of the work • the content is not changed in any way • all files required for usage of the item are kept together with the main item file.

You may not
• sell any part of an item • refer to any part of an item without citation • amend any item or contextualise it in a way that will impugn the creator's reputation • remove or alter the copyright statement on an item.
The full policy can be found here.Alternatively contact the University of Cumbria Repository Editor by emailing insight@cumbria.ac.uk.
network of fashion or brand followers (Lin et al., 2012).With the high volume of social interactions revolving around fashion brands, social influence has, to an extent, a significant role in maintaining loyalty within OBCs.Several online communities are highly diversified: The behavioural characteristics of the individuals differ.Many factors, including shared outlooks, values and principles, result in individuals finding common ground with others causing them to identify with the community, indicating an effective application of social influence from the community itself (X.Cheng et al., 2019;Huang et al., 2017;Kara et al., 2018;Kelman, 1958;Venkatesh & Brown, 2001;Venkatesh & Davis, 2000).
The persuasive influence of emerging OBCs is anecdotally and empirically well documented in existing studies (Beck, 2011;Ibrahim et al., 2017).Extant studies have provided contrasting arguments as to which social networks have the most significant social influence over individuals.One stream of studies suggests that an individual's online social status is measured by their number of social networks (Kim & Dennis, 2019;Muller & Peres, 2018).A key issue of expectation and influence is the receiver's conflicting criteria used to measure the strength of the source, which has led to contradictory conclusions on how social influence is formed (Ismagilova et al., 2019).Other researchers have demonstrated that sources initiating the influence may not be persuasive to all (Cacioppo & Petty, 1982;Dholakia et al., 2004;Flache et al., 2017;Lin et al., 2012).
In their study on social media platforms, Ibrahim et al. (2017) contended that the emergence of OBCs has given opportunities to customers to engage and express their thoughts towards brands.They argued that OBCs can significantly affect consumers' perceptions of brand image and the way in which companies engage with social media and handle customers' participation and opinions.Ibrahim et al.'s (2017) research provided insights into the role that participation and engagement play in millennial consumers' and companies' perceptions of online brand strategies.This is important, given that customers' involvement and engagement vary depending on individuals' insights and participation in OBCs due to several contextual factors (Cayla & Eckhardt, 2008;Coles & West, 2016;Gruzd et al., 2011;W. Ozuem et al., 2018).Ibrahim et al.'s (2017) study remains a valuable tool for understanding patterns of engagement between companies and customers on social media platforms; however, it failed to capture the unavoidable complexity of customer-level involvement and the millennials' characteristics in the OBCs.Specifically, OBCs scholars generally aggregate individual millennials' participation and commitment, ignoring the variance among the demographic cohort (Helal et al., 2018).Though this scholarship on OBCs offers a compelling insight into treatment of customers' participation, it remains disconnected from the inherent complexity of millennials' participation and commitment in OBCs.We explicitly take into account that individuals in the millennial demographic cohort do not necessarily have the same level of participation and interactions in OBCs.The unrestrictive nature of online communities compared to corporate interactions has increased millennial consumers' confidence to share and post their comments based on the trust generated from using platforms (Kong et al., 2019).Consumers may experience intrinsic rewards as well as obtain functional value from engaging in online communities, which lead to their potential identification within the community that advances them to become sources of influence to others.
Drawing on arguments from social influence theory, we propose that levels of customer involvement and engagement can actively determine consumer influence in social media platforms.A central objective of this study is to determine whether consumers' levels of involvement and participation provide multiple characteristics and commitment in OBCs.The theoretical framework contributes to OBC literature by advancing knowledge about customers' levels of involvement and participation in social media platforms.We empirically demonstrate that customer engagement in social media platforms is not merely a stable individual construct but is a dynamic driven process based on individual levels of involvement.Further, our study extends OBC theory by incorporating the overlooked social influence perspective.Specifically, we demonstrate the importance of the level of individuals' involvement and participation that could potentially have an impact on companies' development of customer engagement strategies in the fashion industry.
The above theoretical framing motivates us to structure the paper as follows.We revisit extant literature on OBCs and identify how social influence theory may shed light on the new theoretical insights.Next, we discuss how empirical data were generated and analysed.Finally, we discuss our study's conclusions, implications for theory and offer directions for future research.

| Community and OBCs
'Community' has become an integral part of today's vocabulary.In the 21st century, scholars and practitioners in the field of OBCs and engagement have revived the word community and its popularity continues to grow.The word 'community' had been used sporadically with differing views and meanings due to the emerging computer-mediated marketing environments (CMMEs).Historically, the definition of community was thought to be based on geographically bounded populations.Within those boundaries are diverse groups of people with multiple differences including age, gender, religion, ethnicity, religion, wealth and even power (Cornwall & Jewkes, 1995;Navarro, 1984).Online community platforms, such as social media, may create an 'imagined community' among their targeted groups, exceeding the boundaries that limited physical communities (Anderson, 1983).An important foundation of the imagined community concept is an individual's conscious recognition that they are following similar events with others and that they share common effects with each other.Thus, when individuals find a community with a large population that has interests, values or hobbies that are similar to their interests and values, they are likely to feel part of such a community, even though they may not know the people within that community.
Studies revealed that consumers are placed between two forces: The individual and psychological influence of personal opinions and preferences, and the social weight of beliefs and attitudes (B.Carlson et al., 2008;Nowak et al., 1990).Researchers have recognised a need for a more relational social identity perspective in conceptualising the emerging CMMEs, especially in the fashion industry (Moon & Sprott, 2016).Consumer brand judgement is influenced by the value a brand is perceived to deliver, as customers are encouraged to voice their association with a brand that enhances self-presentation and builds social identity (Algesheimer et al., 2018;Dholakia et al., 2004;Mousavi et al., 2017;W. Ozuem et al., 2016).
Fashion is a powerful social symbol used to create individual and group identities (Ahuvia, 2005); it is also adapted according to users' norms, values and preferences; arguably, fashion trends are cocreated by a number of consumers who preserve and adapt them along the way (Wolny & Mueller, 2013).At a consumer level, if a trend is adopted by a significant number of people, the product's perceived value will be affected, either positively or negatively, by the reception of social influencers.Fashion brands are often described in terms of human personality traits that may possess an emotional component that evokes strong attitudes (C.J. Thompson & Haytko, 1997;Ozuem et al., 2020).Due to the perceived expense of fashion brands, purchasing intentions may be less emphasised in online discussions, yet loyalty towards such a brand will probably be reflected through a significant level of psychological mechanisms directed towards supporting the brand itself.
The fashion industry is known for establishing the concept of embracing individuality among consumers which is evidenced by various social media channels owned by different fashion brands.
However, despite individualistic behaviour, the fashion industry encourages the concept of community, which profoundly applies to groups of fashion fans both offline and online.Traditionally in online communities, followers of fashion brands or trends share information related to their stylistic choices with their peers, with the intention of obtaining feedback on their choices, although the topics of OBCs associated with fashion brands are not simply limited to clothing style discussions (Lin et al., 2012).Online communities allow consumers convenient ways of sharing information with other consumers and enable them to connect and act as social influence towards each other within OBCs (Azemi et al., 2020;Quach & Thaichon, 2017).

| OBC as a dynamic phenomenon: Situating social influence theory
Social influence theory provides a context that outlines an individual's social behaviour through communicated identities (T.Becker et al., 1995;Kelman, 1961).It considers how the influence of social networks enforces individuals to imitate community behaviours (Lucero-Romero & Arias-Bolzmann, 2019;Venkatesh & Brown, 2001;Venkatesh & Davis, 2000).A study by Kelman (1958) identified three levels of influence that impact an individual's attitudes and behaviours: Compliance, identification and internalisation.Compliance involves adapting behaviour to gain rewards or avoid negative consequences, such as community disapproval (Bagozzi & Lee, 2002).
Social influence theory identifies that individuals transform their behaviour to deal with emerging changes in their social environment.Kelman (1958) argued that social influence is determined through levels of compliance, identification and internalisation.Individuals comply through accepting rewards and identifying with groups that comply.Identification happens when individuals accept sources of influence to maintain a desired relationship (Kelman, 1958;Warshaw, 1980).Internalisation takes place through an individual's adoption and acceptance of new behaviours and values within a community with the recognition that these are rewarding (Kelman, 1958).At the internalisation stage, an individual's integration of community norms into their own norms strengthens their connection with the community.
Membership is often associated with the development of the relationship between consumers and brands (Algesheimer et al., 2005) and the development of a harmonious community of members with collective motives, confidence and group attachment (Ellemers et al., 1999;Luhtanen & Crocker, 1992;C. Luo et al., 2015;Ren et al., 2007).This study will inform the literature by recognising that some consumers perceive the value of an online community through its members' community and its 'we' culture with which they may identify as an individual, thus motivating them to become part of that community (Bagozzi & Dholakia, 2006;Bergami & Bagozzi, 2000;Fournier, 1998).This concept has also been applied to consumers' brand love: Consumers participate in online communities that are specifically linked to their brand preferences (Coelho et al., 2019).
From an individual's perspective, belonging to a brand provides a uniqueness to their identity as it implies emotional involvement with a group that shares their values and preferences (Dholakia et al., 2004).Dholakia et al.'s (2004) study showed that higher levels of perceived value lead to stronger community identification.From an individual's point of view, belonging to a brand is of unique significance as it implies an emotional group involvement resulting in attachment or commitment to the community and the brand.Dholakia et al.'s (2004) study suggested that a sense of belonging to a community can increase the value of basic online activities, such as information searching, online task completion and virtual community participation, as consumers develop a sense of personal achievement.A sense of belonging also derives from consumers' longing for recognition from community members, either for self-enhancement of their social status (Baumeister, 1998;Hars & Ou, 2002;Tonteri et al., 2011) or to obtain companionship from those the individual perceives to be like-minded people (McKenna & Bargh, 1999).This prompts the process of consumers converting their broad and general group-oriented goals towards specific areas of social interaction.Thus, members will identify with a specific group or groups of individuals rather than the online channel itself.
Though it is unlikely that individuals will personally know specific members, they still identify with the whole group community (Algesheimer et al., 2018;Dholakia et al., 2004).Every consumer will have different behavioural traits that influence their decision to practice loyalty intentions within a community, so they cannot all be expected to generate the same level of engagement in the same social interactions among members.Furthermore, effort and habitual components are relevant factors because many community members will have a previous history with communities; the histories of existing members would have been developed through routines causing future behaviour to be automatic.Algesheimer et al.'s (2005) study noted that community influence is stronger for knowledgeable members compared to novice consumers.Consumers with an established interest in a brand and previous experience are more likely to engage with the community compared to novice consumers because novice consumers are still in the process of learning about the community and forming a connection with it.It is expected, therefore, that as consumers learn more and engage more with a brand, they will eventually form identification with the brand community.Coelho et al.'s (2019) study focused on identifying how brand communities may contribute to establishing long-lasting relationships with customers based on the mediating effects of brand love.
The study focused on the relational reasoning and the outcomes for members of being part of a brand's community, indicating that commitment to a brand is more important than commitment to the group.Alvarado-Karste and Guzmán (2020) examined how brand identity-cognitive style fit with the three levels of social influence (Kelman, 1958).The authors concluded that the identification and internalisation forms of social influence have a significant positive effect on the perceived value of the brand.The study suggests that regardless of whether the individuals apply rational or emotional reasoning for brand associations through brand identity, it is the individuals' identification with the brand that generates the value of brand equity.In other words, the appeal a particular brand has for a consumer is the supporting source of influence that enhances community identification.This is an undisputed fact as the luxury fashion industry revolves around various psychological mechanisms including consumers' brand personality (M.Pham et al., 2018;Ranfagni et al., 2016;Wolny & Mueller, 2013) and social identity (B.Carlson et al., 2008;Helal et al., 2018;Nowak et al., 1990).Although social identity may be associated more with individualistic behaviour, it captures the main aspects that influence an individual's identification within a group and how they view themselves as a community member (Dholakia et al., 2004).A wide network linked to such a brand may often motivate consumers to join the community to improve their online image and for relational reasons.This can motivate individuals to actively participate as a member of the brand community and maintain relations with other members (Bergami & Bagozzi, 2000;Bhattacharya & Sen, 2003;Eastman et al., 2018).
Other studies have focused on the connections between consumer-members, identifying a 'we' culture in which there is a shared feeling of belonging with other users that separates them from users of other brands (Bagozzi & Dholakia, 2006;Bergami & Bagozzi, 2000;Fournier, 1998;Patel, 2016).According to these studies, the connections between consumer-members were stronger than the relationship between firms and consumers.These authors addressed mechanisms that enable consumers to feel a sense of belonging within a community, yet the instruments that lead to identification directly with the community require further understanding.It is clear that financial motives are not the only concerns for consumers in OBCs.Studies on diverse online habits have helped to identify and extend the different values consumers develop from participating within online communities, including emotional value, relational value and entitativity value, as well as functional value (J.Carlson et al., 2018).These values can be linked to consumers with various community motivations, whether it is for their individual benefit or for group-orientated goals.Through a constructivist perspective this study will provide an in-depth understanding of social influence theory within the context of OBCs and through this methodological position provide an in-depth comprehension of emerging communities and social identity formation.

| Paradigm of enquiry
This paper applied a constructivist research paradigm and abductive research strategy.In contrast to the postpositivist idea of a single reality, constructivist ontology considers that multiple realities exist in relation to subjective conceptualisations of epistemological interactions (W.Ozuem et al., 2017).Its key characteristic is the assumption that individuals subjectively form realities based on preexisting ideas related to social constructions (Guba et al., 1994).The abductive research strategy recognised the difficulty in identifying purely inductive and deductive positions when undertaking research in terms of a clean slate and purely deduced notions/ideas.In an inductive approach, researchers begin with raw data and allow theory to emerge (Strauss & Corbin, 1998, p. 12), whereas deductive analysis considers that a priori assumptions identify theories and hypotheses (Thomas, 2006).The abductive approach is a synthesis of the inductive and deductive approaches, which enables researchers to identify frequent and significant themes that emerge from raw data as well as recognise that preexisting ideas and theories impact upon the research process.
In this way, the key factor linked to OBCs is social influence, and how this attracts participants to the online domains within the fashion industry and what motivates them to remain.Individuals develop understanding following experience of situations that vary among individuals.Habermas (2007) challenged positivist notions by replacing the process of controlled observation with the participatory relationship between the understanding subject and the subject being confronted: 'The paradigm is no longer the observation but the dialogue-thus, a communication in which the understanding subject must invest part of his subjectivity ' (pp. 10-11).
The postpositivist ontology that underpins quantitative approaches is critically realist and considers that one external reality exists and probable truths can be discovered through value-free analysis.That is, there is an external reality that can be discovered and falsified through statistical and deductive techniques, which enable reliability and explanation of a phenomenon.However, counter to this, the ontological position of constructivism considers various accounts of social realities are constructed and rejects value-free axiology; this approach has enabled this study to address real-life processes regarding millennials' constructed loyalty intentions in OBCs.This approach enables the valid exploration and understanding (rather than explanation) of the dynamic marketplace in OBCs.The approach is suitable for research that explores the emerging processes of behaviour in real-life cases; it enables a detailed investigation of how these participants form an attitude to maintain loyalty within luxury fashion brand online communities (Hartley, 2004).To do justice to the explanation of emerging hypotheses through a postpositivist quantitative analysis another study and paper would be required.

| Data collection methods
The constructivist and inductive nature of this study supported open-ended questions which allowed respondents to interpret questions through their own preconceptions and answer through their own narrative and discourse (Geer, 1988) without limiting the length of responses of the individual towards the subject (Kelley, 1983).Even though the aim was to allow respondents more control over their responses, the researchers applied a semistructured approach.According to Crittenden and Hill (1971), levels of intercoder reliability with open-ended coding are low.Social constructivist studies are expected to generate different social realities as specific research questions require investigators to find specific answers to enable logical coding formation.Therefore, researchers must enhance their ability to locate relevant information within a large population (Montgomery & Crittenden, 1977).It is important to locate participants who can respond to the research questions addressed to them.When being addressed with questions, participants may be unable to respond due to a lack of relational experience, which could have an impact on the extent of elaborated responses they can provide (Geer, 1988).To address this issue, the authors ensured selection of participants whose experiences and knowledge could be closely linked to the topic.Indeed, this study draws on theoretical and purposeful sampling to guide data collection.No incentive was given to the participants other than the outcome of the study would provide a much richer understanding of OBCs.Recruitment criteria for the sample were initially individuals from the millennial generation of an age ranging between 18 and 39 years and those with the highest social media usage, which evolved towards individuals who were active users of social media and eventually those who had been influenced by social media on brands linked to the fashion industry.
The initial participants were university students from the United Kingdom and exchange students in United Kingdom from abroad.
The first set of interviews had an emphasis on questions relating to purchasing.However, these interviews identified a need for industry input, questions relating to brand community perspectives and more in-depth knowledge of the online sector in an experiential context.
The next set of interviews led to a greater understanding and the inclusion of theoretical perspectives.Indeed, the questions evolved within the interviews in relation to the extended scope of the respondents and the data gathered provided more precise comprehension and in-depth understanding.
A total of 40 semistructured interviews were conducted; several responses were discounted from the analysis as these were not relevant to the study, and responses that appeared similar or repetitive were discounted.

| Sampling technique
Strauss and Corbin (1998) considered that to sample theoretically, the research should progress in an evolutionary fashion rather than through a predetermined programme.Theoretical sampling is 'based on concepts that emerged from analysis and that appear to have relevance to the evolving theory' (Strauss & Corbin, 1998, p. 202) and maximises opportunities to compare incidents, events and occurrences.It provides opportunities for each sample to build on previous data and analysis.Flexibility and consistency are required: Consistency in terms of the comparison being systematically related to emerging categories to ensure full development; flexibility in relation to serendipity while out in the field (Strauss & Corbin, 1998, p. 203).Charmaz (2006) argued that theoretical sampling identifies a different logic to traditional research design.The purpose of theoretical sampling is to obtain data to help explicate categories.When categories are full, they reflect quality of respondent experiences and provide useful analytic handles for understanding them (p.100).
In the context of theoretical purposive sampling, to collect the required qualitative data, it is important that researchers recruit participants with past, preferably lived, experiences related to the subject of the study (Roulston, 2010).In theoretical sampling, the selection of participants acts as a representation of a population that delivers relevant information and emergent theoretical perspectives.
The aim of qualitative collection is to prompt a direction and a description of specific real-time events and situations without generating interpretive generalisations from the participants (Adams & van Manen, 2008, p. 618).
This study focuses on the millennial age group.A growing body of multidisciplinary research has provided differing views on millennials (de Kerviler & Rodriguez, 2019; J. Luo et al., 2018;Ng et al., 2010).Not only is there much information, but there is also a lack of consistency among scholars and practitioners in the way they define and characterise millennials.J. Luo et al. (2018) (Danias & Kavoura, 2013;Kavoura et al., 2014).This generation is frequently involved in online activities, such as online purchasing, information sharing and social interactions (Bilgihan, 2016;Mangold & Smith, 2012), which puts them at the centre of empowerment within online communities and brand information sharing (Hur et al., 2017).A theoretical and emergent criteria-based sampling procedure was applied to selected participants who possessed the experience and knowledge required to contribute towards the study.However, as further theoretical and empirical considerations emerged, more individuals were asked to participate and, if necessary, questions were reformulated (Table 1).

| Phase 1: Categorisation of codes
Data were analysed using thematic analysis following Braun and Clarke's (2006) six-step process of thematic analysis for psychology.
Thematic analysis is a useful method to summarise key findings in large data sets as it compels the researcher to adopt a wellstructured approach to handling qualitative data to produce a clear result (King, 2004).As the method does not compel the research to follow a pre-existing theoretical framework, the approach is flexible and can be modified when necessary.All interviews were recorded electronically and data were transcribed into a written form consisting of 58 pages using the exact wording of the participants.As the analysis was qualitative in nature, the researchers openly questioned   1996).The codes and themes were developed from previous literature and defined based on the responses of the participants and applied to develop a theoretical framework.The researchers were able to group words into codes reducing data to develop the analysis (Braun & Clarke, 2006).Using quoted statements from interviewed participants and theoretical literature, codes were allocated themes based on meanings revealed, supported with participants' own words and other theoretical literature (see Table 2).

| Brand sentiment
Brand sentiment is the perceived feelings and emotions individuals have for a brand that is expressed through user-generated content within OBCs. Brand sentiment is a popular measuring technique in evaluating the success of social media activity (Hoffman & Fodor, 2010;A. N. Smith et al., 2012).Sentiments are differentiated simply as being positive, negative, neutral or unclear based on the words in written messages.When measuring brand sentiment, an examination of the number of comments and likes is extended to include a measure of the linguistic meaning of the comments: The measure is based on categorising comments using preclassified words that describe positive or negative emotions (Humphreys & Wang, 2017).
Comments are divided based on the number of negative and positive words (Kübler et al., 2019).The number of positive comments and the number of negative comments are considered separate variables (You et al., 2015).Brands are highly concerned about user-generated comments published in social media as they reflect public sentiment towards the brand, and comments have positive and negative influences on consumers observing the online activity (Ibrahim et al., 2017).
According to A. Chang et al. (2013), electronic word of mouth (e-WOM) affects brand sentiment thus affecting community members'  (Ibrahim et al., 2017), although not all consumers will automatically comply with the influence of negative brand sentiment.Individual perceived control over actions is often associated with the measurement of individualism; a desire to remain independent from online social influence indicates a conscious awareness of how inappropriate a socially dictated online community can be (Algesheimer et al., 2018), as indicated by a 28-year-old female Human Resource Assistant: Luxury products are likely to have more positive followers; of course, trolls are expected, but if we opposed brands after some negative comments we might as well let others do the thinking for us.
Furthermore, a 23-year-old male Accounting university student, stated: I feel that allowing other people's perspectives to change your perspective about a brand or product makes you a follower or a reliant on others rather than a leader or an individual.
It is clear that individuals may desire to maintain their freedom and maintain control over influence (Brehm, 1966;Hsiao et al., 2016;Mueller & Thomas, 2001) and resist influence from others with whom they may not agree (Clee & Wicklund, 1980, p. 390) This indicates an element of self-efficacy where individuals will execute their ability to deliver their own course of action (Bandura, 1980) despite the attempted influence of others.Interestingly, identification influence does not solely emerge from other consumers posting comments but also from the brand itself; previous studies have supported the view that brand identification has a positive influence on maintaining customer loyalty (Algesheimer et al., 2005; Alvarado-Karste & Guzmán, 2020).Customers' strong identification with a brand is beneficial to the brand when negative sentiment arises against the brand because these customers will defend the brand against negative online word of mouth (A.E. Wilson et al., 2017).
Customers' strong identification with a brand and their positive internalisation of a brand's messages within OBCs are likely to reduce the negative effect of other individuals' negative sentiment towards the brand.However, individuals with low experience with a brand are likely to rely on the influence of others to support their judgement, as indicated by a 32-year-old male Receptionist: Reviews don't affect my loyalty to a fashion brand that I already like, but they will influence my loyalty intention towards a brand that I have no prior experience with.
A 31-year-old female Careers Consultant indicated that although brands have no direct control over her decisions, she requires support from online community comments when she is unable to decide on a brand; she stated: There are many fashion substitutes so the brand itself has no power to hold me hostage.But how do I decide when there are so many, so reading comments helps me out to narrow down the choices.
In psychology studies, individuals with an external locus of control perceive their lives or decisions to be controlled by factors beyond their control (Ye & Lin, 2015), whereas individuals with an internal locus of control feel they are self-reliable in their actions (Asante & Affum-Osei, 2019;Zigarmi et al., 2018), which is useful in maintaining brand sentiment despite negative comments.Given the degree of control over behaviours, individuals are expected to be able to carry out their intentions without feeling controlled by others.It is important to note that, although individuals may exhibit behaviour of having an external locus of control, it should not be confused with whether individuals allow themselves to be selfsufficient or controlled.According to Ajzen (2002), researchers have aligned self-efficacy beliefs with internal factors and aligned belief in the controllability of behaviour with external factors (Armitage & Conner, 1999;Manstead & van Eekelen, 1998;Terry & O'Leary, 1995).In online communities some consumers may be influenced by In contrast, although consumers may examine perspectives through submitted comments in the online community that does not mean that every mental process of the individual is controlled, because they may have their own brand sentiment.Compliance influence means that individuals accept others' actions and messages even if they do not agree with them (Kelman, 1958).However, a large volume of information is published within OBCs by individuals, making the compliance category too complex to apply because individuals are connected to many other individuals with contrasting perspectives.In this environment, internalisation influence emerges as the key category impacting an individual's processing of online comments.Individuals will attempt to internalise the beliefs of a few individuals with their own beliefs, but with a large volume of comments, this can be overwhelming for consumers (Bright et al., 2015;Lang, 2000).This is an issue identified by a 22-year-old male Finance university student: A lot of times, the ratings, whether they are good or bad, are initially supposed to indicate the overall value of a product.However, loads of people could like or dislike something, so how do you judge if it's good or not?
In most cases, individuals will attempt to filter out the high volume of information to narrow down the search and apply their individual judgement on the overall positive or negative brand sentiment, such is the case with a 21-year-old male Human Resource Management university student: Often, I will not read the individual reviews, there are too many, a summary of the total count of reviews from one star to five stars is already made available, to me that's enough, I feel I can work out the rest after that.
However, even with the desire to make their own decisions, the high volume of comments is a factor that may make it challenging for individuals (Hill & Moran, 2011) Consumers can have a brand sentiment that influences their thoughts and still feel empowered in their ability to choose whether to remain with the brand despite the influence of online communities.A 33-year-old male Procurement Officer supports this, he stated: If someone had a negative experience with the product, I do not want to have the same experience and there is no shortage of substitutes.But that's the product; it does not mean I stop my loyalty with the brand itself.
And went on to state: Online comments don't necessarily influence my loyalty to a fashion brand.These comments could sway my decision to initially buy from the brand, but I decide my loyalty to a brand.
The perception of online comments can be used to manipulate an individual away from one brand to another as indicated by a 21-year-old female Marketing and Fashion university student: The brand Fabletics uses marketing strategies to deter people buying from their competitor, Lululemon; but I know the brand and love it, so I don't pay attention.
A 22-year-old female Finance university student added to this argument stating: Every brand has had a product, a poor delivery of customer service or a bad PR stunt that made its way to Facebook.But we cannot base our judgement to not use brands on these incidents that happen to all brands.H&M and Nike have faced serious social issues addressed by consumers, yet they are still used today.
Through the sentiments of the respondents, it is clear that consumers' brand sentiment is not influenced solely by access to different sources of information; the volume of information and consumers' experience with a brand impacts their sentiments towards a brand.(Brown, 1969;Hall & Schneider, 1972; S. M. Lee, 1971;Sheldon, 1971) to an organisation, community or other individuals with identification and involvement.Buchanan (1974, p. 533)  individual's predictions on perceived similarity with others (H. S. Becker, 1960;Etzioni, 1961;Gould, 1979;Hall et al., 1970;Kidron, 1978;Meyer & Allen, 1984).Regardless of the type of commitment, individuals are motivated to deliver.An important mechanism linked to the development of an individual's psychological attachment to an organisation is the process of identification (Bowlby, 1982;Sanford, 1955;Stoke, 1950;Tolman, 1943).If an individual is not attached or able to identify with a specific source of influence, or agree with the information it delivers, they are more likely to resist the attempted influence (Clee & Wicklund, 1980, p. 390).The same effect can occur if the individual believes that actual influencers with genuine human characteristics are absent from the digital environment (Longoni et al., 2019).This is supported by a 22-year-old male Engineering university student: Knowing if the commenter is not a robot and genuine makes all the difference.
The same respondent went on to comment that online information from an actual individual or group can impact the acceptance of the information; he stated: Online messages seem more believable because they come from real people.When someone takes the time out of their day to write a comment or even offer their critique on a brand, that's better.If all comments came from the brand, they aren't going to openly admit flaws, making it one-sided.
In addition to identifying the importance of involving people to enhance social identification thus motivating individuals to accept online information, this statement refers to how much an online comment represents the overall perspective of a brand; for some individuals, observing too many positive comments is questionable and it results in speculation, whereas others use that to direct their decision to remain loyal to a brand.This is indicated by a 23-year-old male Marketing university student: When you don't know something, you rely on others' views to help your understanding.If 50 to a 100 people comment similar views, you know it's a shared perspective, it's easier to follow a majority as the numbers give clarification.
However, the perceived majority or minority view, although important, is not the only element related to individuals' identification with sources.As mentioned earlier, respondents mentioned the importance of knowing whether the sources are real and genuine in their statements and, most importantly, relatable.This is implied by a 25-year-old male American university exchange student who stated: Relatable everyday person creating the comment creates the tone in their message.Once I read and pick up on the tone of the message, I can tell whether it is real or not.
The source makes all the difference, in this case, it is either the company or everyday people.
Another respondent, a 23-year-old male Business Management university student, indicated that his experience in judging a conversation on a brand in online communities requires someone who he perceives as real and, if possible, known; he stated: I can't read and trust something published online from someone I don't know the identity of….I feel I have a better experience of an actual person I can see than just comments on a computer screen without meaning or knowing where they came from.
Linked to this statement is the debate regarding the perception of receiving and sharing information from sources that individuals have no particular knowledge of.One perspective is that information shared and exchanged among unknown people is perceived as unreliable and untrustworthy; therefore, it will not affect decision making (Mathwick et al., 2007).This links to the perspective that online individuals with high status are more likely to influence others (Hanaki et al., 2007;Hinz et al., 2011).Influencers from organisations and influencers associated with an industry are likely to signal greater expertise compared to personal networks (P.Wilson, 1983, p. 15).Individuals' perspectives of information can be impacted by 'who' is stating it.This introduces the concept of which individuals and how individuals identify with an online source.
Identification in Kelman's (1958) terms occurs when an individual accepts influence to establish a relationship that enables them to feel satisfied, respecting its values without adopting them as their own.When an attachment to an individual, object, group or organisation is made, the receiver of the influence may identify with the attitudes, values or goals of the source of influence and incorporate them into their own responses (Kagan, 1958); however, that does not mean the receiver fully changes their own attitude and behaviours to match others, a level of perceived similarity is required in order for identification with a source to take effect.For fashion brands, online influencers who associate or identify with luxury fashion brands can impact the perception of other individuals who identify with them.A 24-year-old female American university exchange student supported this stating: On Twitter, Instagram and YouTube, if I see comments about brands and their products, especially by well-known YouTube beauty gurus, I will be more inclined to use them.
A respondent from earlier, a 23-year-old female Marketing and Fashion university student, similarly stated: If a comment comes from a fashion blogger or celebrity I like, I'm very likely to take their side.I usually side with people who I believe have the same tastes and likings as I do.
However, a source with significant perceived influence and credibility does not guarantee stronger persuasion than those perceived to have low credibility (Cacioppo & Petty, 1982;Tormala et al., 2006), particularly on the basis of their expertise or their overall online status (Galeotti & Goyal, 2009;Trusov et al., 2010;Watts & Dodds, 2007).In contrast to the perspective of Mathwick et al. (2007), it can be argued that communication between individuals with no knowledge of each other or of any characteristics that would initially credit the online comments is more credible because it excludes corporate information and commercial motives (Bickart & Schindler, 2001).This is supported by a statement from a 28-year-old female Human Resource Assistant: I am not a person who needs to hear about fashion from 'professionals', and often professionals are paid to do so and don't often align with my look.
A 22-year-old female Business Management university student implied a similar view but argued that sources with low perceived credibility can still be useful, stating: It does not always have to be a person who does reviews for a living.If I read a comment from an everyday person, and I feel like their review is genuine, I will trust them as much as I trust an expert guide.
As mentioned before, for social influence to take effect, based on the information published in an online community, individuals must be able, to an extent, to identify with or find similarities with other individuals.Bandura (1977) found that when people encounter others similar to themselves, they are more likely to endorse their statements or actions.A 23-year-old female Marketing university student, stated: My acceptance of a source depends on whether they align with my style type.People have different expectations on fashion brands as opposed to me, so I have to be selective as to who I listen to.
Additionally, a 22-year-old female American university exchange student stated: A lot of reviews for fashion will have authors of different ages, height and styles which I think is helpful because then I can identify with people similar to me, so that way I get a more accurate comparison to how the brand might be for me.
Therefore, for some individuals, it is not the perceived expertise of the influencer themselves that impacts their acceptance of, or identification with, a source, but additional characteristics, such as fashion preference and, arguably, demographic profile.There is no dispute that the credibility of the author of online messages impacts the perceived authenticity of the online messages (Petty et al., 1983;Shan, 2016;Sussman & Siegal, 2003) and the extent to which they will be shared in online communities (Cheung et al., 2012).However, the degree to which an individual will identify with a source will vary as will the reasons for their psychological attachment to a brand; therefore, groups of consumers will judge the source based on their own criteria, including whether the source is relevant or relatable to them.Therefore, the identification category of social influence is a significant factor for this theme as it affects the acceptance of information, making it a key predecessor for the internalisation category of social influence.each other (Frat & Dholakia, 2006;Tikkanen et al., 2009).According to Gruen et al. (2006), an exchange of information and experiences between consumers has a positive effect on perceived product value and increases the likelihood of recommendations.A 23-year-old male Business Management university student supported this view, stating:

| Affirmative experience
If there are a majority that have actually experienced products of the brand and share that, I am more motivated to consider the brand.
Additionally, a 22-year-old female Business Management university student stated: Outside of my current knowledge of a brand, I will frequently rely on the use of reviews and comments by verified purchasers.These are crucial to persuade a sale to happen as these can outline the quality of the item.
Consumers can adjust their beliefs when exposed to new types of information or ideas, this has been referred to as 'openness-toexperience' (John, 1990), indicating that they are open to new ideas (McCrae & Costa, 1987); thus, consumers consider a variety of individuals' experiences.It is possible to assume that individuals may be motivated to obtain a good experience, as enjoyed by others, due to the volume of information-seeking activity within online communities (Currie et al., 2008), and to avoid potential risks.However, an important element to consider is how online consumers distinguish individuals who have actual experience with the brand; that is, the reader's perception of authenticity.A 30-year-old female Accountant stated that: Comments without anything to support or justify them become purely what they are, writings on a wall.
Additionally, a 27-year-old male IT Technician stated: The person must have had some experience with the brand to rate it.I do not trust reviews from a person who judges a brand purely by their own assumptions.
These statements indicate that despite the availability of comments that can guide consumers' perceptions of a brand, individuals may speculate about the perceived authenticity of online comments, more so when they do not know who the commenters are.In this case, readers are likely to consider whether commenters used the brand or are company-paid influencers, or whether individuals' comments are based on whether they benefited from it (Sashi, 2012).
For some consumers, evidence of other individuals' experience is a supportive indicator that assures them of comment authenticity.A 28-year-old female Human Resource Assistant indicated that fashion enthusiasts who have affirmative experience using products can impact consumers' perception of the brand and its products stating: Fashion or beauty product YouTubers have actual experiences to share.Experience helps judgements, so I trust them more.
Another respondent, a 29-year-old female Blogger, specifically focused on how the quality of published content can have an impact on her acceptance of the comments, stating: Pictures help validate comments because they indicate the consumer is legitimate about the review and they have had a personal encounter with it.
In studies of online posts in social media, it is accepted that content requires a satisfactory level of vividness and richness to deliver visual or in-depth message content (Fortin & Dholakia, 2005;Kaplan & Haenlein, 2010;Kisielius & Sternthal, 1984).Consumers can identify that others have had actual brand experience through their use of specific details, which can impact the quality of online comments as indicated by the following two respondents: I tend to look at posts with in-depth detail compared to a company's short comment post (25-year-old male American university exchange student) Online comments are more believable when they are detailed and tell a legitimate story, rather than someone who just gives it a star rating (24-year-old female American university exchange student) This indicates a clear relationship between active conversations and the functional benefits consumers receive when provided with details of real experiences related to a brand (Parra- López et al., 2011).
| 805 Actual involvement with a brand enhances the detail that individuals can publish in online communities adding value to the product or service and moderating the perceptions of other community members (Andersen, 2005); in addition, depending on the consumer's experience, satisfaction with the brand could extend to delight (Oliver et al., 1997), which could expand comments into further detail about the brand itself rather than just about the quality and functionality of its products and services.The perceived authenticity and trustworthiness of the source leads to more positive perceptions of online messages (Chaiken, 1980;H. H. Chang & Wu, 2014;Filieri et al., 2018;Filieri, 2015;López & Sicilia, 2014;C. Luo et al., 2015;Teng et al., 2017).However, the major issue of social influence that is generated from different sources is the diverse and conflicting perceptions of consumers; in different industries, consumers are likely to have various criteria of what makes sources credible sources of influence (Ismagilova et al., 2019).This suggests that consumers will probably search for details in comments that indicate differentiation in individuals' experiences, which could impact the perceived authenticity, especially when the online environment consists of individuals with different mindsets that arise from experiences.This is supported by a 32-year-old female Social Media Coordinator: When someone provides a response, or states different or specific love/hate experiences, it's more likely to be legit and honest because not everyone, who is different, should say the exact same thing, and I don't expect people to have the same perspective.
As mentioned earlier, evidence to indicate online individuals' experience or usage of a brand may be used as an indicator to measure perceived authenticity of the comments they publish in online communities; people who have actual experiences appear to have a bigger effect compared to passive brand fans (Henderson et al., 2010).Currie et al.'s (2008) study found that close friends or networks do not always impact consumers' choices, but the comments of individuals who demonstrate in-depth knowledge and can provide tailored suggestions are more trusted and perceived to have greater authenticity; these expert individuals are from the company that owns the brand and are more likely to possess knowledge of products and services and buying behaviours.However, consumers have become more critical when information is delivered from the company through social media and they think that the information might be misleading or dishonest (Kaplan & Haenlein, 2010).However, the same principle can apply to external commenters as indicated by a 20-year-old female Marketing university student: It's easy for a lot of people to gloat or troll on a fashion brand in one short comment but did they even have anything to do with the brand?
Though earlier respondents indicated that comments from verified buyers impact their decision, a 23-year-old male Marketing university student argued that sometimes it may not be appropriate to consider online comments as a representation of the nature of a brand, stating: Reviews can be helpful, but they sometimes stain the image of a brand based on a poor experience of that person, which may be irrelevant to the experience of others.
A 30-year-old male IT Consultant supported this, stating: If a comment comes from an angry customer, I try to look at what happened in the situation: Who was in the right or wrong?
These statements identify the importance of relevant information as well as details of quality aligned with actual experience.Experience is a significant factor that impacts consumers' brand choice, but those with limited experience or low emotional attachment to brands rely more on individuals who communicate their own personal scenarios in relation to the brand.However, even when that information is shared and exchanged, consumers can be sceptical of online information if the conversation as a whole does not seem believable to them.Therefore, it is important to understand what they look for in conversations found in online forums that lead them to feel they can trust the experience of others to add value to their own.

| Conspicuous effect
Conspicuous effect occurs when individuals are attracted to content within OBCs that contains highly noticeable qualities, such as the volume of comments, the level of differentiation in commenters' information and the source of the comments.Individuals will have different expectations of these qualities and their belief in the information will vary based on the noticeable qualities of the content and its source.It can be argued that the social influence category internalisation can be linked to conspicuous effect which influences individuals to accept others beliefs which there agreeable and authentic to them (Kelman, 1958).
Social media users are able to share information of their choice (Baumann, 2006;Beer & Burrows, 2007) compared to company websites that are selective about the information they publish.With the increased use of social media platforms, companies are perceived to have limited control over the publication of online content (Baumann, 2006) creating an open collective community (Khor & Marsh, 2006) which is visible to the public.Individuals meet through online conversations based on similar interests and ideas; however, for the collaboration of information exchange to work, trust must exist (Parra-López et al., 2011).The nature of online communities means that individuals will encounter people other than just their close networks and friends.Therefore, rather than measuring trust based on interactions between specific individuals, consumers will look at the community's conversations as a whole (Wang & Fesenmaier, 2004).A common studied characteristic relevant to community conversations that impacts individuals' perceptions of brand is perceived critical mass (F.Cheng et al., 2018;Hsu & Lu, 2004;Lou et al., 2000), including its effect on social network site adoption (Sledgianowski & Kulviwat, 2009) and switching intentions (Wu et al., 2017).In regard to perceived critical mass, some consumers' decisions are based on the number of positive feedbacks from others as indicated by a 21-year-old female Digital Media student: If the positive comments outweigh the negative comments, they are contributing factors to my decision to remain with a specific brand.
Similarly, a 23-year-old male Economics university student stated: If a product has a majority of their reviews in the five-star category, I will then skim over the first couple of reviews of the product to see what consumers liked or noticed most about the brand's product.
Having a perceived majority of positive comments may be used to judge other comments that appear to have a more negative perception of the brand.For example, a 23-year-old male American exchange student identified that the different experience encountered by a minority may not be as strong compared to the majority, stating: If there are 20 five-star reviews of a product and only one or two one-star reviews, it is hard for me to believe that a lone consumer had a completely different experience compared to the majority of consumers of the same brand.
Similarly, a 23-year-old male International Management university student supported this perspective stating: If a brand has three reviews and they are all negative, I might take time to look more closely and see who wrote the reviews, because three is not a high representative.
This respondent went on to state: If the negative review is from one upset consumer out of a majority of satisfied consumers, I am unlikely to consider the legitimacy of the lone consumer's complaint, especially when the majority of consumers are completely satisfied.
Other consumers may measure perceived critical mass based on the number of negative comments to judge the perceived authenticity of positive comments as indicated by a 34-year-old female Digital Marketing Consultant: If it's about practical quality or integrity, negative comments are needed to judge the authenticity of positive comments, negative ones carry more weight than positive ones.
Similarly, a 33-year-old male Procurement Officer stated: 'I first look at the good ratings such as the five-star rating and see how well the material was, the durability of what I am buying and if the price is worth the buy.Although I do look at the lower ratings a lot as well, I take all with a grain of salt because I know people could be biased'.
Although a minority may be less significant compared to a larger population to represent a universally agreed perspective, the inclusion of an element of minorities can potentially be useful to some individuals.Although majority and minority influence differs, the nature of their influence process remains similar (Nemeth & Wachtler, 1983).In contrast to the previous two comments, a 35-year-old male Customer Service Specialist stated: Usually, if I only see one negative review surrounded by a bunch of positive reviews, I will question whether the person who wrote the negative review is just one person who had a bad experience.
It can be argued that voicing a minority perspective reflects a view of reality, that individuals will not think alike and will stay loyal to their perspectives (Moscovici & Personnaz, 1980), reflecting a consumer's positive acceptance of published comments that may not support a brand in online communities whilst remaining true to their own beliefs.Although it is natural for brands to not desire negative comments, if brands fully controlled what consumers can read online and did not publish negative comments, consumers' compliance with social influence would be reduced, thus reducing the development of internalisation, which can prevent consumers from developing negative perceptions of the brand.A 25-year-old male American university exchange student addressed this issue stating: Even top fashion brands will have some negative comments, such as on price or style; if they were blocked out it would show the company's full control over the forum.
Gilmore and Pine's (2007) study identified key drivers behind the demand for authenticity, including the historical concern regarding the control companies exerted to depict the brand's image based on what they wanted consumers to see.This prompted concerns regarding brands enabling consumers to observe or practice authentic interactions (Henderson et al., 2010).In a practical sense, the activity of balancing positive and negative arguments before making decisions has been found to be useful for individuals and it increases positive psychological outcomes (Collen & Janis, 1982;Hoyt & Janis, 1975;Mann, 1972).Internalisation influence involves individuals accepting others' beliefs and behaviour, which they process as agreeable to them and authentic, both privately and publicly due to rewarding outcomes (Kelman, 1958).Facilitating consumers' ability to find and consider all perspectives not only enables consumers to develop judgements for themselves but also potentially enhances their view of the brand's online community.
This is supported by a 24-year-old male Finance university student; he stated: Anything that sounds vague in either a praising or criticising comment can be hard to believe.Something like 'this is great' or 'this sucks' on every comment, I can't really trust or consider.
Another respondent, a 29-year-old male Sales Representative similarly stated: Sometimes comments appear too good or extreme to be true; with a little critical thinking, it is easy to tell exaggerated from down-to-earth feedback.
The same respondent stated: Companies have been known to create fake comments.
You can easily spot the fake comments because they sound too made up.I can tell when a company has just made up a name and posted too many good comments.
This respondent indicated disbelief in conversations that appear to be too supportive towards a brand.Although it may be a subject that requires a debate, this view reflects the serious issue regarding online individuals' fear of being subject to control by an online community (J.Chen et al., 2016) as consumers may desire to maintain their freedom within an online community.However, an important group to consider are the consumers who have a genuine love for a brand (Beverland et al., 2008).Many luxury fashion brands have been active for years with an established offline reputation before social media had an effect on people's social lives.This emerged from the comment of a 21-year-old female Marketing university student: When it comes to a luxury fashion brand, I expect it to be the best.So, I barely consider negative comments, I'm usually surprised if I see one.How can you base your preference on Chanel or Hugo Boss just because someone else may comment their dislike of them?
Additionally, a 29-year-old female Fashion Blogger supported this stating: Luxury fashion is not available to everyone, so surely very few people will have something negative say, I can't even think of a bad thing about any particular one other than they won't match everyone's taste.
Responses on the relevance of negative comments in the online communities of luxury fashion brands do indicate that brands' reputation and image can outperform the influence of negative comments directed at them.Their image of being a high-end product category generates an expectancy of high quality; therefore, online community comments are expected to reflect the brand's performance.Judging from respondents' comments, it is possible that consumers judge brands if they feel that comments seem too commercialised to appear aligned with an individual's value traits or personality.Additionally, some consumers may be concerned about the source of the information; some are likely to question the authenticity of information if it comes from a single source or from a source that consumers cannot identify or relate to.

| Phase 2: Community conversation and involvement matrix
A review of codes led to further categorisation of themes, regarding the types of online influences consumers identified, from the emerging keywords delivered by interviewed millennial participants and their brand sentiments.As the analysis progressed, it triggered some interlocking features which replicated and compounded some of the primary themes but were too complex and differentiated to be categorised and broadly placed in the primary categories.
These patterns identified a necessity to extend the scope and categorisation of individuals' brand perceptions.Individuals observing activity contributed by other individuals within OBCs perceived the published information to be either based on speculative or justified thinking.Speculative thinking and justified thinking are processes that an individual can have in relation to a brand.Speculative thinking is contemplation of an issue or event, whereas cognitive thinking requires an individual to remember specifics of a situation or event, that is, cognitive thinking is based on past practical experience (Kant, 1998, p. 585), which this study refers to as justified thinking.Thus, speculative thinking refers to a thinking process that is not based on specific actual experience or on evidence of experience that would influence the thought process about an object or situation.In contrast, justified thinking is a mental process that can refer to actual experience.Customers who have invested time to participate in purchasing and online socialisation activity linked to a brand, are likely to develop emotions towards the brand (Fournier, 1998; M. T. Pham et al., 2018) resulting in the brand securing a salient position in the customers' mindset (Stokburger-Sauer et al., 2012).However, lacking actual experience with a brand does not mean customers will not have sub- comments that appear too positive or negative regarding the brand as a topic.Thus, each individual will have different reasons that enable them to identify with sources and their information (e.g., Allsop et al., 2007;Watts & Dodds, 2007).Individuals from both the process thinking categories illustrate the overall conspicuous effect of online content within OBCs.Both speculative and justified thinking processes will influence individuals' judgements on user-generated content based on criteria, such as source of information, and the context and how it portrays the brand, along with other characteristics such as brand loyalty or preference, and social and personal goals.However, some individuals will not allow the conspicuous effect of online content to determine their decision to remain with a brand or choose another brand (e.g., Chae et al., 2017;Monga & John, 2008;Umashankar et al., 2017).Thus, conspicuous effects will have varying effects on individuals' behaviour.Additionally, individuals will have varying levels of involvement with the community to either support or oppose a specific brand, but those who are indifferent towards a brand or other online users' comments will be less involved within the community (Dick & Basu, 1994).
From the four themes of brand sentiment, identification with source, affirmative experience and conspicuous effect (BIAC) and the speculative and justified thinking processes, the following categories emerged based on the varying behaviour of online users: Judgmentalists, bias situators, rationalisers and sugar-coaters.The types of conversations online users encounter, their perceptions of those conversations and how they link the online conversations to their own final decisions, identify these distinct types of influencers who shape different brand sentiments, which stem from the thematic categories identified in Table 2. Figure 1 presents the category matrix diagram.
Judgmentalist and sugar-coater influencers are more aligned with the speculative thinking category than the justified thinking category; these individuals are often associated with a more passive experience with a brand as they often publish information that is perceived too vague to be justifiable.Judgmentalists may appear to be too critical of the brand, whereas sugar-coaters may highly praise a brand, making these two groups contradictory.Regardless of positive or negative active experience, these types of millennial influencers follow their own 'feelings' and identify with others who appear similar.Rationalisers and bias situators are associated with active experience as they often identify a real-time event or outcome related to the brand, which makes them appear more justified in their information and thinking.Bias situators, with either a positive or negative experience, apply their experience in their justification when identifying with a source and brand sentiment.Rationalisers with low experience are open to discussions linked to brand sentiment to rationalise their decision.Regardless of whether these groups individually are involved in the community, conversations would not emerge if one existed without the other (Table 3).

| Phase 3
Following identification of themes, each was allocated meaning as part of understanding the relationships between brand sentiment and social influence identified in OBCs in the fashion industry.Using thematic analysis, meanings were allocated to the generated themes and the Community Conversation and Involvement Matrix separately.The flexibility of thematic analysis enabled the researchers to generate different themes (Braun & Clarke, 2006) that were linked and developed from respondents' comments.Data analysis led to the development of the BIAC model (see Figure 2).The researchers' own thoughts were significant in the formation of the model as it is important to develop an understanding of new insights that go beyond the words of respondents (W.F. Ozuem, 2004).| 809 opinions through online venues (Howe & Strauss, 2009;Mangold & Smith, 2012) resulting in the publishing of highly critical information.
Judgmentalist speculators do not appear to identify with the brand and are expected to be highly critical of information published online in communities; information that is perceived to be 'overly' supportive of a brand causes them to question the content and author (Baker et al., 2016;Ma et al., 2015;Villarroel-Ordenes et al., 2018).
Consumers with no loyalty to the brand are expected to have made no patronage and have no relative attitude towards the brand (Dick & Basu, 1994).Although it is likely that judgmentalist speculators have no active purchase experience with a brand, it is possible to measure their level of passive loyalty through their responses to company-generated content.Judgmentalists are attracted to authors they perceive to be 'everyday people' as their source of influence rather than professionals; however, they remain highly critical of the brand, reflecting more on the brand's equity (Herhausen et al., 2019) than on product quality due to the absence of active purchasing experience.Judgmentalists are more likely to act if a brand's online actions arouse negative emotions, which accommodate a negative conversation thereby affecting their overall brand sentiment.Some consumers are likely to perceive judgmentalists as 'online trolls' who disregard the positive experiences of others as a justifiable source of influence, whereas others may find them useful to ensure they do not invest in a brand that will generate negative consequences; furthermore, judgmentalists generate conversations when they come into contact with brand loyalists.

| Sugar-coater speculators
Although millennials are perceived to desire control in online environments, they are the least homogenous demographic group.
The online environment consists of millennial subgroups with different online behaviours (Kilian et al., 2012, p. 115).In contrast to judgmentalists, sugar-coaters may be perceived to be genuine loyalists towards the brand.However, their language may not appear to reflect active experience with the brand, which makes other online users sceptical towards this group.It is not easy to measure their precise loyalty due to vague published comments; however, the nature of brand fans can provide an understanding of their psychological mechanisms.Sugar-coaters probably align with the need to feel a shared feeling of belonging with others to separate them from users of other brands (Bagozzi & Dholakia, 2006;Bergami & Bagozzi, 2000;Fournier, 1998) and when they have a strong identification within a community they are likely to conform to the norms (Algesheimer et al., 2018;Bhattacharya et al., 1995;Wellman et al., 1996).Though other individuals may be sceptical of the community if they feel pressured to follow the influence of the community (McMillan & Chavis, 1986), sugar-coaters are less likely to perceive pressure from the community due to community identification.
Sugar-coaters are more likely to remain loyal through positive e-WOM despite any negative e-WOM or other experience.Similar to judgmentalists, they will align more with 'people like them' (Allsop et al., 2007) who confirm preconceived assumptions (Kim & Dennis, 2019); this supports the view that inductive reasoning is the source of influence with which they identify.Although they can be classified as 'fake loyals' who are more passive than active (W.Ozuem et al., 2016), they shape positive brand sentiment in the community.

| Bias situator justifiers
When consumers have a positive purchase experience with a brand, they are often expected to have repurchase intentions.When a F I G U R E 2 BIAC (brand sentiment, identification with source, affirmative experience and conspicuous effect) model negative experience occurs, it is likely to create an 'emotional scar' on the individual (Pullman & Gross, 2004).Bias situators do not heavily rely on others for bias confirmation compared to other users but align more with their own active experience with the brand.
These individuals can be perceived to be self-sufficient or self-reliant and are less likely to commit personally to a community and they remain objective, (Tang et al., 2008;Zhou et al., 2009) thinking that people are responsible for the outcomes of their own decisions (van Laer et al., 2013).E. J. Lee's (2012) study addressed how usergenerated comments can bias interpretation of media coverage or company-generated content.However, some consumer groups feel that media can also be biased against their perceptions (Gunther et al., 2009;Schmitt et al., 2004;Vallone et al., 1985).Regardless of the source of the information, bias situators measure credibility based on their own experience or influencers' experience if it is active experience.Therefore, bias situators focus on sources with expertise, such as company experts or other consumers who can relay, for example, purchasing history.For brands individually, this group can be considered true loyalists justifying their decision through experience.Although they are aware of others' experience, they remain biased due to believing their own experience to be fact and others' experience to be speculation, which prevents them from considering alternatives or choosing brands even with perceived critical masses.

| Rationaliser justifiers
This group can be viewed as: (i) Consumers who provide comments worded to help rationalise arguments in conversations and (ii) consumers who observe conversations to rationalise future decisions.
The first are perceived to be more experienced and are likely to engage more as they are already connected to a community (Algesheimer et al., 2005) and have low susceptibility regarding their judgements (J.Chen et al., 2016).The latter typically have low experience or certainty, indicating their likelihood to follow others' leads until they establish their own perspective.Research showed that consumers' choices, when faced with uncertainty preference, are best understood when based on reasons that are for or against alternatives (Shafir et al., 1993;Simonson & Nowlis, 2000).Consumers' need for justification influences, strengthens the trade-off effects, especially when consumers expect others to evaluate their decisions (Simonson & Nowlis, 2000;Simonson, 1989).These individuals may reflect the characteristics of 'indifferent loyals' who are unconcerned about a specific brand (W.Ozuem et al., 2016).
However, it is possible to group individuals with uncertainty under the category as they are yet to decide their brand preference.When these individuals lack actual experience, they will search for comments from different sources.The perceived credibility of sources with which to identify will vary more for this group due to different values and perceptions (Galeotti & Goyal, 2009;Hanaki et al., 2007;Trusov et al., 2010;Watts & Dodds, 2007) as well as limited experience.Therefore, other characteristics, such as perceived critical mass on perceptions (F.Cheng et al., 2018;Wu et al., 2017), combined with sources will help rationalisers summarise brand sentiment, which they will link with their own pending sentiment.

| THEORETICAL AND PRACTICAL IMPLICATIONS
Our study provides a set of practical insights that brand managers and companies can employ to take maximum advantage of consumers' levels of participation and interactions in OBCs and it conveys an important set of four key customer types: Judgmentalists, sugar-coaters, bias situators and rationalisers.The task of managing social influence involves: Identifying the amount of community involvement individuals will contribute or accept in their decision making and determining the level of conspicuous conversation effect in terms of how online conversations impact consumers' perceptions and involvement in online communities (Shim & Koh, 1997).It is expected that customers' negative sentiments can be changed to positive sentiments if they observe a high frequency of company replies directed to customers that emphasise willingness to help customers (Ibrahim et al., 2017).Though this may work for rationalisers, the perspective presents a challenge for all four studied groups who either reflect high or low self-focused behaviour (H.H. Chang & Hung, 2018).Individuals who are self-focused are more likely to be resistant to persuasion and standard norms and behave according to their central values.A high self-focus is likely to be associated with judgmentalists and sugar-coaters who are expected to implement a feeling-based strategy, which is more automatic (Zajonc, 1980) and more interpersonal and intrapersonal between members (L. Lee et al., 2009; M. T. Pham et al., 2001) and less focused on numerical quantities.
Judgmentalists and sugar-coaters have already established a perspective of a brand and will contribute to conversations based on their feelings in contrast to a reason-based strategy, which is more aligned with bias situators and rationalisers.Judgmentalists and sugar-coaters are not easily influenced by other members, even if the other members appear to be reasonable and justified.The reasoning behind the low conspicuous conversation effect is that sugar-coaters remain significantly involved in the community.Schwartz et al.
(2011) studied why consumers are less likely to seek second opinions and retain a specific relationship and found that the followers are more concerned with preventing damage to a valued relationship.
Thus sugar-coaters align with a feeling-based strategy that prevents them from considering alternatives as they have an established feeling of belonging with the community (Bagozzi & Dholakia, 2006); therefore, they are more likely to have high community involvement, as do judgmentalists, but there is a low likelihood that conspicuous conversations will have an effect unless they favour the brand.
In contrast, judgmentalists and bias situators are more likely to feel conformity pressure within the community but will be impacted by the conversations and community involvement effect differently.
Judgmentalists' online behaviour is more likely to be feeling-based OZUEM ET AL.
| 811 and they are critical towards sugar-coaters who praise brands within the community.However, a condition that may affect some judgmentalists is a lack of a positive social connection with the brand that impacts their attitude reflected within a community (Pozharliev et al., 2015).Company-generated content or overly positive endorsement of content are not typically well received if individuals cannot relate to them; thus, selected consumers are recruited as 'seed agents' to target their peers who are more sceptical of companies' content (Dost et al., 2019); so, if judgmentalists feel they cannot identify with a social network or the brand, they will not participate positively in conversations.Conspicuous conversations will have a significant effect on judgmentalists if the community has limited representation of a group of individuals.This links to Grewal et al.'s (2019) study which highlighted that a community that has a particular identity (Reed et al., 2012) will not be attractive if a brand signals the same identity through posts and conversations.Therefore, although judgmentalists may appear highly critical towards a brand if they are unable to relate to a brand through the community, they will search for individuals to whom they feel closer.
Bias situators are not devoid of feeling, but they are the least likely to be moved by online conversations or to incorporate the community environment into their decision making, but their reasonbased approach makes them rationalisers for the adoption of the brand's products.In contrast, rationalisers, who are more likely to have limited experience with the community, are more reliant on others (J.Chen et al., 2016) and will have a high conspicuous conversation effect, but will have a low active involvement with the community and have opportunities to evolve into one of the three other types of customers.
We identified four types of customer levels in OBCs which have not been previously articulated.This study demonstrates major differences among the four mentioned customer groups.Although they respond differently within communities, they act as paradoxes that can contribute to formulating brand sentiment within the community.For each group, it is important to understand the characteristics that impact their brand sentiment, which is reflected through conversations they contribute to or observe.
The study presents a model illustrating the process of the development or maintenance of brand sentiment, one of the four thematic categories in this study; the other three thematic categories are consumers' identification with source, affirmative experience and conspicuous effect.For online conversations to commence, opposing groups maintain on-going conversations depicting contrasting arguments, leaving observing consumers to see visible arguments and formulate their brand sentiment.If a conversation is perceived to have too many negative comments, groups that support that brand may attempt to outweigh the negatives, and vice versa for brand opposers regarding too many perceived positive comments.It is important to note that consumers will follow either a feeling-based strategy or a reason-based strategy (H.H. Chang & Hung, 2018) so their responses to the conversations will vary; for example, sugarcoaters and judgmentalists will probably align with how they identify with the brand, however, judgmentalists will be critical of the involved sources of information, whereas bias situators will act independently and align with their personal active experience.
The existence of both judgmentalists and sugar-coaters ensures continuing active conversations about the brand, giving fans the opportunity to openly support the brand when encountering opposers or consumers still considering whether to connect with a brand.
Bias situators may offer a practical contribution to the community concerning product endorsement but less about social interaction.
Thus, the three types of customers are significantly important con-

| LIMITATIONS AND SUGGESTIONS FOR FUTURE RESEARCH
The BIAC model is not a tool that determines the type of engagement online community users will participate in as the complex The study expands understanding of how characteristics of online communities shape brand sentiment within online communities, focusing on how conversations published in a forum are received by different types of customers and how members who are active in online conversations influence other consumers' perception of the brand.However, there is a need for greater understanding of increasing customer loyalty (Knox & Walker, 2003), particularly of how loyalty is affected by the influence of online community members.
Our findings identify the value of social interactions in sustaining long-term community engagement, but they do not provide a specific link with loyalty intentions.Future research could investigate what types of participation there are in online communities and the extent to which they contribute to customers' loyalty intentions to brands.
Although this study was based on a range of experiences, the researchers acknowledge that results are taken from a limited number of interviews and focus on the surface of conspicuous effect, source identification and affirmative experience.Further research could involve in-depth investigations as to how one group of consumers, such as the rationalisers, can be shifted to another group of customers based on conversations or experiences that impact their brand sentiment.Furthermore, this study focuses on one industry, so it is recommended that the BIAC model be tested in further research applied to other industries, and that other social influence groups be identified and the impact of their interactions on brand sentiment or consumer decisions and responses be investigated and how that impacts loyalty directed to brands.The current study focuses on one group, millennials, so the findings could be different for other demographic groups.Future researchers exploring these issues may draw from different groups, using different methodological orientations.A longitudinal survey strategy may offer more breadth and insights in terms of triangulating both quantitative and qualitative data to examine the customer types identified in the current study.
regarded millennials as individuals born between 1979 and 2002, with increased motivation to use social media as a medium for social interaction.Azemi et al. (2020) suggested that millennials are dominant users of online platforms and they have an elevated inclination to participate and engage in social interaction.In defining and characterising millennials, this current study builds on Helal et al.'s (2018) conceptualisation of three distinct sociocultural dimensions of millennials: Tech-savvy, socially conscious and active social media users.In today's youth-oriented society, the millennial group is the most tech-fluent generation; millennials have adopted social media into regular everyday communication, including social interactions and reflected on their perceptions of the emergent data.It was challenging for the researchers to balance their perspectives with the participants' expressed opinions.van de Ven (2007) argued that consensus among coders increases the consistency of interpretations of the decision rules used to identify themes.Following this line of thought, two authors were tasked with the development of the themes and the keywords, and the third author read through the original data to ensure that the themes reflected the data.With an overarching sense of purpose, the researchers converged to resolve their differences around the emergent data, concepts, themes and the relevant literature.In the next step, the researchers read and analysed transcripts from the 40 participants from the millennial generation who were users of social media to identify specific patterns that emerged from the participants' responses.FollowingSeidel and Kelle's (1995) suggested approach to reducing data and coding, relevant phenomena (repetitive mention of specific words or sentences) were highlighted in the transcripts and were analysed to determine similarities, differences and patterns among the participants(Coffey & Atkinson, . The participants' comments arguably indicate that the compliance category of social influence has less of an effect on individuals if they have limited or no identification with the source of information.If individuals do not agree with the comments of other individuals, their sentiment regarding the brand is unlikely to change, which supports the view that a lack of identification and internalisation influence has a significantly stronger effect on an individual than compliance influence.However, it is important to consider an individual's experience with brands and how that affects their perceived control over the attempted influence addressed to them.Experienced individuals are likely to follow their own initiative in a community(Algesheimer et al., 2005) whereas less experienced members may mostly follow others because they seek security (J.Chen et al., 2016).A 35-year-old male Accountant identified that personal experience determines his loyalty actions:If I am already a loyal customer to the brand and I, myself, have been happy with the quality and style of their products, then I will not let other people's perspectives change mine.An example of this is Nike shoes breaking and tearing on people, however, I have never had this problem, so I see no reason to stop choosing them.Similarly, a 30-year-old female Data Risk Analyst stated:My loyalty to a brand remains regardless of online communities' reviews.If I ever decided to change luxury brands it should be whether I had a negative experience, not what others over the internet say.My experience is fact to me, online comments, from my perspective, are speculations until proven otherwise.
OZUEM ET AL.| 801 online comments published by others when they do not feel assured because of a desire to receive direction to generate a positive outcome as indicated by a 25-year-old female Accounting university student:Even if it's a luxury brand I like, I will always check comments.Specific products from brands are usually hit or miss and I cannot guarantee consistency in quality for them all, even if I have an established customer relationship with them.
, which can have an impact on their full practical control, thus affecting their outcomes.A 21-year-old female American exchange student indicated responsibility in searching for information in a high volume of information with limited ability and resources: It's quite hard to figure out through 100s of comments what's the best decision.I don't have time to read through such a large number, I only look at the top 10 or so but then I risk missing information that could have helped me more.
Consumers' experience in obtaining information from influencers enables them to consider options or alternatives, whether they act upon the influence will depend on their perception of the arguments as well as their own developed sentiment.From the responses, identification and internalisation seem to exert more social influence on the participants than compliance.The majority indicated they make the final decision to remain with a brand despite online influencers, and they use the availability of comments as a guide and not as something that determines final decisions; this finding indicates the significance of internalisation, which emphasises the acceptance of influence based on whether an individual agrees with those delivering the influence, in contrast to the compliance category of social influence, which requires acceptance of influence regardless of disagreement.These findings clearly indicate that consumers are capable of making decisions, but uncertain situations can cause them to turn to the support of influencers.4.1.2| Identification with sourceIdentification with sources refers to an individual's ability to relate to the person providing the information, based on some perceived similarity or shared ideas.Identification is often determined by the perceived similarity between parties which prompts a desire to maintain a relationship that helps reach goals.Strong identification can have a major influence on the acceptance of information (S.A.Thompson et al., 2017), which can prompt individuals to internalise the perspectives of other individuals as their own, thus influencing them to remain involved and attached to a community.Several researchers have associated individuals' psychological attachment viewed commitment as an individual's biased and emotionally based attachment to an organisation's goals and values that aligned with their own goals and values; the individual is committed to the organisation itself and not solely to its practical worth.Other researchers studied two different types of attachment: One which is based on calculative involvement for exchange of specific extrinsic rewards and the other based on a moral attachment that is based on an Affirmative experience emerges when individuals convince others they have supporting evidence of having actual experience of participating in an activity.In order for individuals to accept the influence of others delivering information related to a brand, they require assurance that the individual has actively consumed the brand's products and services.This can be considered the starting point of the internalisation of social influence in which individuals perceive others' experience to be highly authentic and this can result in their acceptance of the information.Experience has been traditionally associated with individuals who have influenced an event or have been involved in an event resulting in the learning of new skills or the development of psychological mechanisms.Offline environments, social clubs or networking events enable individuals to exchange information and interact to form networks with the intention of increasing their knowledge or developing relationships with people that would contribute to their development.In online communities, the concept of accessing people with experience or obtaining experience is very similar.Many individuals visit online communities and social networking sites to seek information; thus, if consumers are motivated to use a brand, they are likely to search for content in online forums (de De Vries et al., 2012).Several authors have explored how individual consumers apply their own practical experience in using brands' online forums to make decisions(Bright et al., 2015;Dunn et al., 2010;Mollen & Wilson, 2010), which can include the experience of observing other users' comments or actively exchanging real-time information with jective thoughts and knowledge of a brand.Though some individuals attain knowledge through experience first, others may develop it through impressions prompted by their thoughts (N.K.Smith, 2011,   p. 1).Speculative thinkers' perceptions of a brand are based on their own ideas and reasoning formed without evidence from actual experiences.Whereas individuals with a justified thinking process use their brand experience to justify their perception of brands.Within OBCs, individuals who comment about a brand without reference to actual experience apply a speculative thinking process, whereas individuals influenced by brand experience who comment on a brand apply a justified thinking process.Specifically, regarding identification with sources, individuals with speculative and justified thinking processes will exhibit similar behaviours, such as aligning with like-minded individuals and opposing those who are not like-minded.As a result, they may question other individuals'

5F
| DISCUSSION OF MILLENNIALS' SUBGROUPS 5.1 | Judgmentalist speculators Millennials are the dominating customer segment and are associated with a need to control their environment, particularly in expressing tributors towards online conversations that shape the perspectives of other individuals including rationalisers.Even if one group does not have affirmative experience or another does not identify with certain sources, the information they generate or observe from others impacts overall brand sentiment, and even though individual brand sentiment can be categorised as speculative or justified, both types contribute to the overall community brand sentiment.The BIAC model can support the tracing of consumers' mental processing behind their brand sentiment, which links to the information they generate or conversations they have participated in, supporting marketers to identify what encourages consumers to follow the sentiment of others or to contribute towards the brand sentiment.
nature of users consists of a wide range of values and purposes of engagement.However, the BIAC model can be used to evaluate the nature of participants' involvement in online conversations and the characteristics in OBCs that define types of customers within a community.The model provides guidance on the level of social influence members will accept from fellow members and whether their influence is a matter of what they feel or see.However, though the majority of interviewed participants provided insights into how community conversation content affects their brand sentiment, the study had few participants that would actively contribute to online conversations; these types of customers were identified mostly based on conversations the majority of participants stated they had encountered.A greater number of active contributors to online conversations would have provided further insight into what motivates active contributors to become involved and would have expanded the types of participation in online communities.