Explaining fashion diffusion from multiple perspectives
A study examining the effect of network closure on susceptibility to interpersonal influence and brand tribalism
Nathalie Stolwijk (11815396)
Final Version
Supervisor: Marco Mossinkoff
MSc. in Business Administration – Marketing track Date: 21/06/2018
Statement of originality
This document is written by Nathalie Elisabeth Stolwijk, who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the
Table of contents
List of figures and tables 4
Abstract 5
1. Introduction 6
2. Theoretical framework 10
2.1 Definition of fashion 10
2.2 Models of fashion diffusion 11
2.3 Symbolic interactionist theory of fashion 12
2.4 The fashion transformation process 12
2.5 Research gap 13
2.6 Social psychological perspective: susceptibility to interpersonal influence 14
2.7 Sociological perspective: tribal marketing 14
2.8 Research question 15
2.9 Complexity approach 17
2.10 Network closure 19
2.11 Network closure and susceptibility to interpersonal influence 19
2.12 Network closure, conformity and susceptibility to interpersonal influence 20
2.13 Network closure and brand tribalism 21
2.14 Network closure and susceptibility to interpersonal influence and brand tribalism 23
combined 3. Method 24 3.1 Participants 24 3.2 Procedure 24 3.3 Measures 26
3.3.1 Development Network Closure Scale 26
3.3.2 Susceptibility to Interpersonal Influence Scale 28
3.3.3 Brand Tribalism Scale 29
3.3.4 Conformity Scale 29
3.4 Statistical method 29
4. Results 30
4.1 Correlational analysis 30
4.2 Main analyses 31
4.3 Mediating role of conformity 37
5. Discussion 39
5.1 Reflection on the results 39
5.2 Limitations and future directions 42
5.3 Implications 43
6. Conclusion 44
References 45
Appendices 51
Appendix 1. Back translation questionnaires (English to Dutch) 52
Appendix 2. Informed consent 56
Appendix 3. Scales 57
Appendix 4. Exploratory factor analysis 60
Appendix 5. Discriminant analysis tables 61
List of figures and tables
Figure 1: Conceptual model 17
Figure 2: Mediation model 30
Figure 3: Scatterplot canonical discriminant functions 35
Figure 4: Mean score low network closure condition 35
Figure 5: Mean score moderate network closure condition 35
Figure 6: Mean score high network closure condition 36
Figure 7: Plot means and standard deviations 37
Table 1: Last finished education (frequencies) 24
Table 2: Explorative factor analysis 27
Table 3: Pearson correlations, means and standard deviations 30
Table 4: Mediation analysis 38
Abstract
The global fashion industry covers a large range of different activities and is one of the biggest industries in the world (Aspers, 2010). Furthermore, fashion trends also affect architecture, vehicles and more (King & Ring, 1980). Fashion forecasters try to explain why certain trends emerge (Shaw & Koumbis, 2017). However the fashion industry frequently fails to accurately predict trends (Hauge, Malmberg & Power, 2009). The present study does not try to explain why fashion diffuses, but how. The aim of the present research is to examine the effect of network closure on susceptibility to interpersonal influence (1) and brand tribalism (2), and both concepts combined (3). A quasi-experimental research was conducted to test the hypothesis (N = 162). It is quite unique in this field to take a quantitative approach. MANOVA analysis showed a significant effect of network closure on susceptibility to interpersonal influence and brand tribalism combined. Post-hoc analysis revealed that a higher level of network closure had an effect on the level of susceptibility to interpersonal influence. However, only the low and high network closure conditions differed significantly. Besides, the found effect was not mediated by conformity. Furthermore, there was no significant effect of network closure on brand tribalism individually. The present study was among the first empirical studies to combine social psychological and sociological perspectives to explain fashion diffusion. Future research should continue examining the factors that help discover how fashion diffuses.
Keywords: fashion diffusion, network closure, susceptibility to interpersonal influence, brand
1. Introduction
Certain trends seem to spread in an almost epidemic way. As an example, a blue and white coat from the Spanish fashion label Zara got spotted on the streets so many times that it even got honored with it’s own Instagram account called ‘that coat’. Here everyone who owns or encounters the blue and white coat can share pictures of the coat with the hash tag #ThatCoat (Linch, 2016). In contrast, at the end of every fashion season millions of items end up in sale or even remain entirely unsold. In 2015, 21.5 million pieces of clothing remained unsold after being discounted in the Netherlands (Wijnia, 2016). This raises the question, how does one fashion item become a massive trend, like the blue and white Zara coat, while millions of other items remain unsold. Fashion does not only apply to clothes. Fashion trends are also apparent in other material products such as furniture, architecture and vehicles (King & Ring, 1980). Nowadays, the factors that are critical in purchasing fashion products are becoming more important. A growing amount of consumers is making purchase decisions based on the aesthetic and symbolic value of products. Even in other sectors than the fashion sector, such as the technology sector (were functionality used to be the main criteria), these factors are becoming increasingly important (Verganti & Dell’era, 2014).
In contrast, companies are still loosing large amount of money because of redundant fashion stock (Desai, 2012). Changing fashion trends have made these fashion items out-of-date (Jain, Bruniaux, Zeng & Bruniaux, 2017). Companies are using sales as a tool to sell unsold fashion items. But the underlying problem continues. Companies appear to be having great difficulty forecasting fashion preferences of consumers. The fashion industry frequently fails to accurately predict trends. The type of fashion items that consumers demand are constantly changing in style and aesthetics (Hauge, et al., 2009; Jain et al., 2017; Wijnia, 2016). Besides leading to negative financial consequences for companies (Shaw & Koumbis, 2017) the production of fashion items is also polluting the environment and causing negative work
conditions for employees (Claudio, 2007). Compared to the food industry, trend forecasting has to be done within large ranges in fashion (Shaw & Koumbis, 2017). This leads to high levels stocks, markdowns and unsold goods. Previously, scholars have developed theories on fashion diffusion (see Cholachatpinyo, Fletcher, Padgett, & Crocker, 2002; Kaiser, Nagasawa, & Hutton, 1995; Simmel, 1957). These theories provide insight in the emergence of trends. Still, the empirical question how fashion diffuses remains unsolved. Research could investigate trends in fashion more thoroughly, in order to find out how one fashion trend diffuses quickly, while so many others fashion items are not even sold. When this process is revealed, fashion trends may be predicted more accurately. This will help decrease the redundancy in fashion items. Moreover, insights into the process of fashion trends could be applied even further. It could help explain fashion adoption beyond the fashion industry, explaining fashion adoption in all sectors in which trends have an influence, such as in the previously mentioned technology industry. Insights in fashion can help understand the contemporary consumer, because fashion now plays a part in all principles of life (Atik & Firat, 2005). Therefore research on the fashion phenomenon is highly relevant in order to gain more knowledge on consumer behavior.
Based on this theoretical and practical relevance, the purpose of the research is to investigate the fashion process more thoroughly, to get a better understanding of how fashion spreads. Most theories that attempt to explain fashion change do so from different perspectives (Cholachatpinyo et al., 2002). The present study is combining different perspectives in order to investigate fashion diffusion. Surprisingly little quantitative research has been conducted in an attempt to explain this phenomenon. In practice, trend watchers often tend to look at why fashion spreads. They look for the motivation a consumer has to buy particular fashion items (Shaw & Koumbis, 2017). In contrast, the present study assumes that the emergence of fashion trends is not just based on the aggregate of individual preferences. The whole context should be taken into account. Therefore the present study takes a more holistic view when looking at the diffusion process, so focusing on how fashion spreads. Complexity theory could help explain the fashion
phenomenon, since this theory states that the whole is more then the sum of its parts. In the present study I empirically test one factor deducted from complexity theory, network closure. In closed networks everyone is connected, internal cohesion is emphasized and behavior is very noticeable (Burt, 1992; Lin, Cook & Burt, 2001). The type of network might influence how fashion trends spread. More specifically, the research question is: to what extent does network closure affect susceptibility to interpersonal influence and brand tribalism? Individuals in closed network are more closely connected, more homogeneous and information spreads more quickly (Baxter & Woodside, 2011). It is expected that the level of closure in a network has an effect on the amount of influence that others have on an individuals purchase behavior. The is called the susceptibility to interpersonal influence (Bearden, Netemeyer & Teel, 1989). It is also expected that network closure has an effect on brand tribalism. Brand tribalism is an individual’s tendency to adopt particular brands to display membership to a certain social group (Khare, Mishra, & Parveen, 2012). The aim of the current study is to examine the effect of network closure on susceptibility to interpersonal influence (1) and brand tribalism (2), and both variables combined (3). Building on the exploratory and descriptive research conducted previously, the present study takes a confirmatory approach. A quasi-experimental study is conducted in order to empirically test the hypotheses. In this way the present study tries to make an attempt to confirm some of the qualitative statements of the previously constructed theories on fashion.
The present study aims to contribute to the literature in five ways:
− A first important contribution of the present study is the integration of social psychological perspectives (susceptibility to interpersonal influence) and sociological perspectives (brand tribalism) in order to investigate the fashion phenomenon. Hence, the study attempts to integrate earlier theories on fashion, combining micro-level and macro-level theories. The micro-level refers to the study of the interaction between individuals. The macro-level refers to structures in society. These structures are sustained by social control and can provide
opportunities and constraints on the interaction and behavior of the individuals (Münch & Smelser, 1987).
− A second contribution to the literature is the usage of the concepts susceptibility to interpersonal influence and brand tribalism to operationalize fashion adoption. In order to be able to explain how fashion spreads both concepts are used. The present study assumes that under the influence of others and by wanting to display membership individuals adopt a certain fashion trend. It is unique to use both concepts to operationalize fashion diffusion.
− In the third place, the present study adds to the theory of tribal marketing. It investigates some of the qualitative statements made by Cova and Cova (2002) on the use of similar brands in tribes. Some hypotheses are based on the assumptions of brand tribalism and these are statistically tested in this quantitative study. Moreover, the effect of network closure on brand tribalism is investigated and could provide new insights into tribal marketing.
− A fourth contribution is the usage insights derived from complexity theory to explain fashion diffusion. By taking the assumptions from complexity theory as a foundation to explain the fashion process, this study is unique in this field. The present study tests whether one factor deducted from complexity theory, the closure of the network, influences fashion diffusion. Because the present study is based on the principles of complexity theory, it also provides insights in how to complexity theory can be applied in order to study fashion diffusion. It is expected that taking this approach can help a great deal in explaining how fashion diffuses. It could provide new insights for predicting fashion trends, which could not have been found when only looking at the individual preferences.
− A last contribution is the development and validation of a scale to measure the network closure, the Network Closure Scale. Often social network studies use
equations on the interconnectedness in a network (see for an example: Allcott, Karlan, Möbius, Rosenblat, & Szeidl, 2007) to measure network closure. The present study is based on self-report. It looks at the perception an individual has on its own network. In order to test the premises made in the present study a network closure scale was developed and used. To my best knowledge, only a few scales have been developed to measure network closure in a survey.
To practitioners, this study aims to provide more insights into the process of how fashion diffuses. As mentioned earlier, the prediction of trends remains a difficult task for fashion producers and retailers. This study provides valuable knowledge, since there is little insight in how fashion trends spread. This study investigates one factor (network closure) that companies could keep into mind, when they bring new fashion items to the market.
Firstly, this research paper provides a theoretical framework, discussing previous research on fashion change and relevant theories. Secondly, the research method will be discussed. Afterwards, the results are presented. In the last place, the results are discussed and implications and suggestions for future research and a conclusion are given.
2. Theoretical framework
In this section a theoretical framework is presented. First fashion is defined, afterwards relevant theories on fashion diffusion, susceptibility to interpersonal influence, brand tribalism and networks are discussed. Next, the hypotheses are presented.
2.1 Definition of fashion
Fashion as a phenomenon reflects imitation (Simmel, 1957) and is a broader phenomenon then just clothing, applying to all types of fashion items (King & Ring, 1980). In the present study fashion is defined as: “an unplanned process of recurrent change against a backdrop of order in the public realm” (Aspers & Godart, 2013, pp. 185). From a social perspective fashion
does not only apply to clothes. The process of fashion change applies to almost all human activities (Aspers & Godart, 2013), it reflects changes in economics, politics, social life, culture and aesthetics. The taste and lifestyle of individuals and society is mirrored in fashion (Cholachatpinyo et al., 2002). Fashion is one of the most visible mediums of change. Since fashion means change, it can help in the investigation of the complex processes of change (Kaiser et al., 1995). Therefore my research results may be generalized further then apparel, explaining general consumer behavior when adopting fashion.
Fashion is a selection process for which the objects have to be considered as new, in order to be able to become fashionable. Furthermore, a fashion change starts with a small number of people. After this the fashion object gets adopted by a large group of people. Subsequently, after people have adopted the fashion object, it goes out of fashion again (Aspers & Godart, 2013). 2.2 Models of fashion diffusion
One of the earliest theories on fashion was developed by Veblen (1984). His theory stated that the dress gives an indication of the status and wealth of a person. Later, Simmel (1957) developed the trickle-down model of fashion diffusion, aiming to find an explanation for how fashion trends are adopted. In this sociological model he stated that the process starts when the elite selects and dictates a certain fashion style, in a top-down manner. Subsequently, the majority imitates this style. Because the fashion style then becomes too popular, the style looses its appeal for the elite. The elite then tries to differentiate again by introducing a new fashion style. Processes that are underlying this model are differentiation, imitation and social contagion (Crane, 1999; Simmel, 1957). In contrast, in the trickle-up model subcultures introduce new fashion styles, which later on get adopted in the rest of society (Crane, 1999). This bottom-up model of fashion diffusion states that new fashion styles are introduced by lower class groups and then adopted by higher social classes (Ma, Shi, Chen & Luo, 2012). In addition to these earlier models, the trickle-across model is developed. In this model fashion styles are adopted
simultaneously in all levels of society (Ma, Shi, Chen & Luo, 2012). For trickle-across fashion change it is more difficult to determine what factors caused individuals to adopt fashion.
2.3 Symbolic interactionist theory of fashion
Through the symbolic interactionist (SI) theory of fashion Kaiser and colleagues (1995) examined the factors that lead to acceptance of new fashion styles. In this attempt the researchers tried to bridge appearance processes at the micro-level and cultural processes at the macro-level. Kaiser and colleagues (1995) argue that ambivalence is important in the fashion adoption process. This ambivalence occurs when the individual is deciding between individual freedom while seeking new stimuli and conforming to the social norm (See also Simmel, 1906). This ambivalence motivates the individual to search for a resolution through fashion. When individuals decide to change their way of dressing, this can cause fashion change. This process occurs continuously. One reason for this is that entering different social contexts causes changes in identity construction of the individual. In addition, companies can profit from offering new appearance styles. Therefore companies will continue introducing new fashion objects that consumers can consider, causing more ambivalence.
2.4 The fashion transformation process
In a reaction to the SI theory of fashion, Hamilton (in Cholachatpinyo et al., 2002) argues (among other things) that the theory cannot fully answer the question how macro-level systems influence the individual’s fashion decision process. Cholachatpinyo and colleagues (2002) presented the fashion transformation process model as addition to the SI theory of fashion, in order to fill in research gaps. The theory states that individual psychological factors are important in explaining the emergence of new fashion. Bridging micro-level and macro-level, the collective selection theory can be used. At the macro-level Cholachatpinyo and colleagues (2002) propose that there are different lifestyles, with their own attitudes, behavior and practices. These are formed by the different sub-cultures in society. This theory states that fashion formerly served a socializing function and as a social standard. From an individual consumer behavior
perspective the decision to adopt a fashion trend could be caused by a specific need to either differentiate or socialize (Cholachatpinyo et al., 2002). Needs underlying the differentiating process are for instance the need to escape boredom and to be different. Needs underlying the socializing process are for instance the need for social affiliation and adjusting to the changing society (Sproles & Burns, 1994 in Cholachatpinyo et al., 2002). Put differently, the inner self is negotiating both the need to be different and the need to conform to the social norms. The balance between these two opposite psychological processes causes the individual to adopt to fashion (Cholachatpinyo et al., 2002).
2.5 Research gap
Despite these attempts to explain fashion diffusion, Christopher (2004) states that it is very difficult to predict fashion trends. In addition, Brun Petersen, Mackinney-Valentin and Melchior (2016) call for more research on trend mechanisms based on contemporary understandings. Therefore a more in-depth understanding of how fashion diffuses is valuable in order to improve fashion trend predictions. Besides trying to look for evidence for ‘why’ the ongoing processes of fashion change occur, research should also look at ‘how’ these changes in fashion preferences emerge. Cross-disciplinary approaches could help explore changes in fashion, because it is a multisided phenomenon (Kaiser, Nagasawa & Hutton, 1995). Different theories from varying perspectives, as the ones mentioned above, have been constructed to explain the fashion phenomena (Cholachatpinyo et al., 2002). In addition, Janssen and Jager (2003) argue that the integration of theories on diffusion, consumer decision-making and social networks is valuable (Janssen & Jager, 2003). The purpose of the present study is therefore to contribute to the theory of fashion by investigating how this process of fashion diffusion occurs, taking insights from these different theories and integrating them with theories on networks. This knowledge is valuable for theories of consumer behavior. First I will give a social psychological view on fashion. Next a sociological view is given. Later on I will combine both perspectives in an attempt to intergrade the micro-level and macro-level and introduce network theory.
2.6 Social psychological perspective: susceptibility to interpersonal influence
The fashion process can be described as the process that looks at how a potential fashion trend goes from its emergence to acceptance by the public (King & Ring, 1980). Based on the previously described self-negotiation, certain individuals will adopt more quickly to an emerging fashion trend then other individuals. Based on Roger’s adoption model (Rogers, 1983) at the society level fashion innovators are the first ones to adopt to a trend. This fashion trend will be more visible for the followers, and the chances of the trend being adopted by the majority of people increase. Once the majority has adopted, the laggards will adopt the fashion trend. Through this cycle a new fashion trend can emerge (Cholachatpinyo et al., 2002; King & Ring, 1980; Law et al., 2004).
Fashion can help consumers construct their self-image and appearance (Law, Zhang, Sun Leung, 2004). In a qualitative study on fashion diffusion based on chaos theory, Law and colleagues (2004) found that the importance of being fashionable was the most important factor influencing the reaction towards new fashion trends. This reaction is based on the opinion of others. It seems that individuals are being influenced by others when making purchase decisions. The current research takes into account how individuals are being influenced by others when making purchase decisions, this is encompassed in concept of the susceptibility to interpersonal
influence (Bearden et al., 1998). This construct consists of a normative dimension, which is the
inclination to conform to other’s expectations. The second dimension is an informational dimension, which is the inclination to see the information from others as evidence about what is real (Bearden et al., 1998). Therefore, it seems that the influence of others has an influence on fashion adopting.
2.7 Sociological perspective: tribal marketing
In addition to the above-mentioned social psychological processes that explain adoption of fashion, consumers adopt brands to show that they belong to certain social groups (Khare et al., 2012). This behavior can be explained by taking a tribal marketing perspective. According to
the tribal marketing perspective not only the individual activity, but also how the individual is embedded in the social context should be taken into account by researchers. In this view proximal social groups, tribes, influence consumers behavior more then modern institutions or formal cultural authorities. Membership to tribes is based on a shared passion or emotion. Therefore tribes are not fixed and individuals can belong to several tribes at the same time. Tribes are therefore more like micro-cultures than subcultures (Maffesoli, 1996 in Hardy, Bennett, & Robards, 2018).
Since 2000 there has been a change in what consumers value when consuming products. Where marketing usually characterizes the western society as an individualistic society, there is a shift to consumers searching more and more for social links. In the tribal marketing view, it is hypothesized that products and services that permit social interaction within the tribe will be valued more by consumers than the use-value. In this perspective consumers are less interested in the object itself, but more interested in the social links and identities that arise when consuming certain objects. The link becomes more important than the object. This is called the linking-value of the object (Cova & Cova, 2002). Consumers consume certain brands (partially) to belong to a certain group (Roane, 2015). In the process of adopting a certain fashion style, consumers will value the specific attributes of the clothes less than the linking-value that the clothes provide in a close social group (Cova & Cova, 2002). Meaning is constructed in the tribal culture and then interpreted by individuals in that subculture (Cova & Cova, 2002). Research by Ruane and Wallace (2015) shows that individuals use fashion to demonstrate the values of the tribe they belong to. Brand tribalism therefore can explain adoption of fashion. This will be incorporated in the current research to complement the social psychological perspective to adopt a style.
2.8 Research question
perspectives to further understand the factors that are related to the fashion diffusion process. For instance, individual and social or cultural influences are interacting in fashion change (Cholachatpinyo et al., 2002; Kaiser et al.,1995). In this research, the focus lies on explanatory factors that help clarify how fashion styles diffuse. The assumption is that it might not be the (aesthetic or practical) characteristics of the fashion style, such as the colors or textures, which are driving the fashion diffusion (Esposito, 2011). It is not about what color is in style, but for instance how this particular color, that is in style, spreads. Previous research has not been able to fully integrate the different fashion theories, and empirically test the assumptions of the constructed theories. By analyzing two concepts, susceptibility to interpersonal influence and brand tribalism simultaneously, a social psychological perspective and sociological perspective (Zayasenko, 2012) are combined to explain fashion diffusion. For both of these processes the shape of the network is essential. How fashion diffuses could be investigated by looking at differences in the type of networks (Janssen & Jager, 2003). The current research takes one factor, the closure of the network, to explain how fashion diffuses. Figure 1 shows the expected relations between the concepts. Based on the literature review and research gap the research question is:
To what extent does network closure affect susceptibility to interpersonal influence and brand tribalism?
Figure 1. Expected effect of network closure on susceptibility to interpersonal influence,
conformity and brand tribalism. 2.9 Complexity approach
Regarding the research question and when trying to find an explanation for how fashion change diffuses, taking into view that systems can be viewed as complex, self-organized structures could help. The complexity approach studies the complicated and surprising things that can arise when individual components interact. In this view seemingly unrelated connections are studied by combining systems from different disciplines. This can be used as bridge building between the individual and the higher order macro-levels (Johnson, 2011). Complex adaptive systems are made up out of a big amount of interacting components, which cause patterns of behavior to appear at higher levels. At the macro-level, new complex properties are exhibited by the system, which do not exist at the local level of the different components. This view could help explain the previous observation that it is hard to predict fashion by only looking at individual fashion preferences. It seems that when explaining how fashion emerges, research
should look at the whole system, and not just the aggregate of individual behavior. This is incorporated in the complexity, since in this approach applies: the whole is more then the sum of its parts (Rupert, Rattrout & Hassas, 2007).
One characteristic of complex systems is that they are self-organizing and adapting to changes in the environment, without interference of central control or rules. A complex system is characterized by having many interacting individual components. These are all interconnected. The rules of behavior in the system are constructed at the lower level, when agents are adapting to their environment (Rupert et al., 2007). In fashion individual agents are deciding whether to adopt a trend and which fashion items to buy. Through interacting with the other components in the system, the behavior of the whole system is shaped. At the macro-level order then emerges (Rupert et al., 2007). This can also apply to fashion change. Based on the meaning developed when interacting with others, individuals adopt new fashion styles (Kaiser et al., 1995; Kim, Rhee, &Yee, 2008). This then again influences the fashion process at the level of the society. Order can be considered as the dominance of one particular style in the system.
The complexity approach could help bridging the micro-level and macro-level processes that are involved in fashion change diffusion (Johnson, 2011). By understanding the external conditions that bias the systems, we can more accurately anticipate the systems future behavior. Certain arrangements in the system will therefore happen more often then others based on external influences. The manner in which agents interact, affects the arrangements in the system, how long they remain and when they change in arrangement. The output is a certain amount that is created by the objects collectively (Johnson, 2011). In the case of how fashion trends diffuse, this number might be the amount of agents that adopts a fashion trend.
Complex systems collectively respond and adapt to external stimuli coming from the boundary of the system (Chandler, Rycroft‐Malone, Hawkes, & Noyes, 2016). Complex systems can be open. This means that the environment surrounding the system can influence it. In
systems are rare (Johnson, 2011). It may be assumed that the external stimuli providing new energy to the complex system are the organizations that are introducing new fashion styles. In this part marketing plays a role in the diffusion of new fashion trends, providing new energy to the complex systems.
2.10 Network closure
In complexity theory networks and connections are essential. The relationships between the components, not the components by themselves, form the fundament for understanding the world (Kleindorfer & Wind, 2009). The shape of the network has a significant influence on consumer decision-making. It influences the number of products that are dominating the market (Janssen & Jager, 2003), and could therefore have an influence on fashion diffusion. Individuals in networks with a closed structure are closely connected to each other. Communication goes direct and efficient and information spreads more efficiently. In general, closed networks are more homogeneous and the individuals in the network mostly have access to the same
information. Norms restrain the behavior of people in the closed network and the opinion of the majority determines what is the right behavior. In a network with an open structure people are more loosely connected to each other (Baxter & Woodside, 2011; Granovetter, 1973). The fashion adoption process operates both within a social environment and between different social environments (King & Ring, 1980). In more closed networks, individuals circles of
acquaintances overlap for the most part. A person’s acquaintances probably know each other as well.
2.11 Network closure and susceptibility to interpersonal influence
It is expected that in closed networks individuals are more susceptible to interpersonal influence. Bearden and colleagues (1998) define the susceptibility to interpersonal influence in three components. The first component is that consumers are buying and using certain brands in order to enhance their image or identify with certain significant individuals. The second one is
that consumers are willing to conform to the expectations that others have on their buying decisions. The last component is that consumers have the tendency to observe others purchase behavior or request information about certain products or services.
It is expected that in more closed networks, there will be more susceptibility to interpersonal influence. In these closed networks individuals will adopt certain fashion in order to enhance their image. In closed networks norms are more salient, and because of that there is more pressure to conform. There are clear expectations on what (purchase) behavior is appropriate. Also, information about products and brands spreads easily in closed networks, therefore it is easier for individuals to request information and observe others buying behavior. (Baxter & Woodside, 2011; Bearden et al., 1998; Granovetter, 1973). In contrast, norms will be less salient, the pressure to conform is lower, and information spreads less easily in open networks. Therefore it is expected that individuals are less susceptible to the influence of others in more open networks (Baxter & Woodside, 2011; Bearden et al., 1998; Gravetter, 1973). In the current research individuals will be divided in three different network conditions, individuals in a network with low, moderate and high network closure. Based on the reasoning above the following hypotheses are formulated:
Hypothesis 1: The three conditions (low, moderate and high closure of the network)
differ in their level of susceptibility to interpersonal influence.
Hypothesis 1a: In the conditions with high and moderate network closure there is a
higher level of susceptibility to interpersonal influence then in the condition with low network closure.
Hypothesis 1b: In the condition with high network closure there is a higher level of
susceptibility to interpersonal influence then in the condition with low and moderate closure of their network.
The mechanism underlying the relation between network closure and susceptibility to interpersonal influence might be caused by the conformity concept alone. Conformity to the expectations of other persons when making purchase decisions is one of the characteristics of susceptibility to interpersonal influence. Since norms constrain the behavior in a more closed network, it is expected that individuals will be more prone to conform to these norms. Norms seem to exert a big influence, especially when the sources of influence are similar to us (among other things) (Cialdini & Trost, 1998). This homogeneity is present in networks with more closure. Therefore it might be the conformity to the norms on which products or brands to buy, that mediates the relation of network closure and susceptibility to interpersonal influence. The fashion transformation model (Cholachatpinyo et al., 2002) also mentions the need to conform to social norms as one of the processes that can influence fashion adoption. Taking this into regard, it may be expected that in networks with higher closure, the norms regarding certain styles cause more conformatity to adopt certain styles. In contrast, in more open networks these norms are less salient, members are less similar, and individuals are less likely to conform to the norms and customs in the network (Adler & Kwon, 2002; Cialdini & Trost, 1998). Therefore it might be expected that the individual is less likely to conform to certain styles. This leads to the following hypothesis:
Hypothesis 2: The positive relation between closure of the network and susceptibility to
interpersonal influence is mediated by conformity.
2.13 Network closure and brand tribalism
Taking a sociological perspective, the assumption is made that tribal members adopt fashion to display membership to particular tribes. Through self-expressive brand consumption in these tribes a social link is created, which helps tribal members achieve group acceptance (Ruane & Wallace, 2015). Consumers therefore adopt to certain brands to display that they are conforming to the social norms of the network they want to belong to (Khare, et al., 2012). When
the network has more closure, these norms might be more salient and the consumption of tribal brands will be more socially visible (Ruane & Wallace, 2015). Also, when a network has more closure, members have more relationships within the same network, and less outside of the network (Häuberer, 2011). A network is closed when all individuals in the network are connected. So person A, B and C all know each other. In these types of networks individuals are more likely to comply with the norms and customs within that network (Adler & Kwon, 2002). In their social network analysis study Reingen, Foster, Brown & Seidman (1984) found that when there are stronger links between the people, they buy more similar brands and products. More closure in a social network seems to stimulate similar consumption. Therefore it is expected that in more closed networks trends will be adopted quicker by tribes.
In contrast, when the network is more open, some members in a network have relations to members of other networks (Häuberer, 2011). This means that the contexts of these individuals are more heterogeneous. In more open networks, person A knows person B and C, however person B does not know C. This is called a structural hole. In these networks there is more diversity (Flynn, Reagans, & Guillory, 2010). If norms get violated in open networks, this goes without notice and remains unpunished (Adler & Kwon, 2002). Therefore it is assumed that in open networks brand tribalism, and the use of fashion adoption to display membership, will be less salient. Brand tribalism is probably lower in these open networks. Based the reasoning above, the following hypotheses are formulated:
Hypothesis 3: The three conditions (low, moderate and high closure of the network)
differ in their level of brand tribalism.
Hypothesis 3a: In the conditions with high and moderate network closure there is a
higher level of brand tribalism then in the condition with low network closure.
tribalism then in the condition with low and moderate closure of their network.
2.14 Network closure and susceptibility to interpersonal influence and brand tribalism combined
When individuals select products, there are two social processes occurring. In the first process when deciding which products to buy is based on the observation of others. Because the
influence of others seems to have an important influence on the purchase behavior, it is expected that fashion will diffuse when individuals adopt a fashion trend under influence of others.
(Bearden et al., 1998; Ruane & Wallace, 2015; Janssen & Jager, 2003). In the second process products are selected based on the need to belong to a group and to express status and identity. In this way a particular product becomes more valuable, when a particular group of people is using it already (Janssen & Jager, 2003). This process occurs in brand tribalism when individuals adopt certain brands to display that they belong to the tribe they want to belong to (Khare et al., 2012). Since it is expected that the networks with different extent of closure will differ in their
combined level of susceptibility to interpersonal influence and brand tribalism (in order to explain fashion diffusion), the following hypothesis are formulated:
Hypothesis 4: The three conditions (low, moderate and high closure of the
network) differ in their level of susceptibility to interpersonal influence and brand tribalism combined.
Hypothesis 4a: In the conditions with high and moderate network closure there is higher
level of susceptibility to interpersonal influence and brand tribalism combined then in the condition with low network closure.
Hypothesis 4b: In the condition with high network closure there is a higher level of
susceptibility to interpersonal influence and brand tribalism combined then in the condition with low and moderate closure of their network.
3. Method 3.1 Participants
To determine the minimum sample size a power analysis was conducted with the program G*Power 3.1 (Faul, Erdfalder, Lang, & Buchner, 2007). To achieve a power of .95, α = .05, with a medium effect size of f ² = .15 (Cohen, 1977) a sample size of 107 was calculated. However, to conduct the exploratory factor analyses the rule of thumb is 10 participants per item (Velicer & Fava, 1998). Therefore 160 participants were the minimum requirement to find factors for the developed network closure scale used in the present study.
During a period of two weeks data for the research was collected. The participants were gathered in personal network of the researcher. 182 people were invited to participate in the research. The data of those who completed the research were analyzed. The sample consisted of consumers who buy fashion items. The questionnaires were completed by 89% of the participants (N = 162). The sample consisted of 30.2% male participants and 69.8% female participants. The age varied from 14 to 85 years old (M age = 30.1, SD age = 13.9). Table 1 shows the latest finished education of the participating sample. It indicates that the sample was high educated, with 59.9% having obtained a University Degree. In order to include Dutch participants Dutch scales were created using back translation (See appendix 1).
Table 1
The frequencies and the percentages of the latest finished education of the participants
Primary education MAVO, LBO, VMBO HAVO, MBO VWO HBO degree University Degree Total Frequency 3 5 18 11 28 97 162 Percentage 1.9 3.1 11.1 6.8 17.3 59.9 100.0 3.2 Procedure
Participant received a message with the invitation to participate in the research. Participants chose the location where they wanted to fill in the questionnaires themselves. Completion of the questionnaires took about 5 to 10 minutes. The research started with an informed consent, asking participants to participate in a study on personal experiences and brands. As was mentioned in the informed consent (see appendix 2), participation was completely voluntary and the participants could end participation prematurely. Subsequently, the participants could fill in the questionnaires, which can be found in appendix 3. The survey ended with a demographics questionnaire. The questionnaires were completed in the program Qualtrics. 3.3 Measures
3.3.1 Development Network Closure Scale. A scale was constructed to measure the network closure of the network of the participant. Five items were used from a scale developed by Huijsman (2018). These items were: ‘Most of my friends know each other’, ‘My good friends do also know my family’, ‘At work I meet very different people compared to the people I meet during my leisure time’ (reversed scored), ‘My neighbors visit my birthday parties’ and ‘My colleagues visit my birthday parties’. However, these items did not form a reliable scale, therefore a new, more elaborate scale was developed to measure network closure.
Based on the literature on closed networks (Baxter & Woodside, 2011; Granovetter, 1973), which states that closed networks are characterized by being more homogeneous, the items ‘Most people in my close environment are quite similar to each other’, ‘In my close environment most people look like each other’, ‘In my close environment most people have similar interests’ and ‘My friends look like me’ were added. For these items a higher level of agreement with the statement indicated a higher level of network closure. Openness of the network is characterized by more heterogeneous people in the network and is measured by the items ‘I am friends with very different people’ and I have a lot of loose connections’. For these two items a higher level of agreement indicated a lower level of closure of the network, therefore
these items were reversed scored.
Based on the literature on structural holes in networks, which create openness in network (Burt, 1992) the following item was constructed: ‘A large part of my friends don’t know each other’. Since structural holes are defined by a network with a high level of diversity in attitudes and perspectives the following items were constructed: ‘The opinions of the people in my close environment are very different’ and ‘In my close environment people have very different political stands’ (Burt, 1992; Flynn et al., 2010). For these items more agreement on the statements indicated more openness of the network, therefore these items were reversed scored.
Furthermore, individuals in networks with a higher level of structural holes have access to different resources. This is the access to information/ resources characteristic (Burt, 1992; Flynn et al., 2010). Based on this the following items were constructed ‘The people in my close environment have different incomes’ and ‘In my close environment I’m in contact with people from different ages’. Here a higher level of agreement with the statement also indicated more openness of the network. Therefore these items were reversed scored.
Two items, ‘My neighbors visit my birthday parties’ and ‘My colleagues visit my birthday parties’, correlated very low with the total scale (r < .3). In addition, the reliability increased to the sufficient α = .70 when these items were removed (Nunnally, 1998 in Peterson, 1994). Also, these items showed skewness to the disagree side. Based on literature the removal of these items could be justified, therefore it was decided to remove the items from the scale.
Exploratory factor analysis was conducted to find the underlying structure and dimensions of the scale. Since some correlation is expected between the potential discernable dimensions, orthogonal rotation using Promax rotation was used (Costello & Osborne, 2005). The item ‘In my close environment I’m in contact with people from different ages’ did not load on any of the dimensions, and the reliability remained unaffected when this item was removed.
Based on the literature it was decided to remove this item. As is shown in table 2, there were four factors discerned. Except for the item ‘At work I meet very different people compared to the people I meet during my leisure time’, the dimensions were found as was expected based on the literature. The factor loadings can be found in appendix 4.
Table 2
Explorative factor analysis: the eigenvalues of the first 7 factors of the Network Closure Scale
Factor Eigenvalue Explained variance per factor
1 3.08 23.71 2 1.88 14.45 3 1.41 10.87 4 1.28 9.81 5 .92 7.56 6 .87 6.70 … … …
All items on the Network Closure Scale were rated on a 7-point Likert-scale ranging from strongly disagree (1) to strongly agree (7). A 7-points was chosen for the most optimal reliability and validity. Larger scale points then 7 would probably not improve the reliability and validity and it should not be lower scale points then 5 (Dawes, 2008). The scale had a sufficient reliability of α = .70 (Nunnally, 1998 in Peterson, 1994), and can be found in appendix 3. The reliability coefficient indicated that this scale was reliable and suitable for conducting analysis (Nunnally, 1998 in Peterson, 1994).
3.3.2 Susceptibility to Interpersonal Influence Scale. The Susceptibility to
Interpersonal Influence (SII) Scale developed by Bearden and colleagues (1989) was used to
measure the level of susceptibility to interpersonal influence. The questionnaire comprised of twelve items. The reliability was α = .88. This questionnaire consisted of the normative
dimension (eight items, α = .89, e.g. ‘It is important that others like the products and brands I buy’) and the informational dimension (four items, α = .72, e.g. ‘If I have little experience with a product, I often ask my friends about the product’)1. All items were rated on a 7-point Likert-scale ranging from strongly disagree (1) to strongly agree (7). A higher mean score indicated a higher level of susceptibility to interpersonal influence. The items can be found in appendix 3.
3.3.3 Brand Tribalism Scale. Brand tribalism was assessed using the Brand Tribalism
Scale developed by Veloutsou and Moutinho (2009). The questionnaire consisted of 16 items
and comprised five dimensions. Degree of fit with lifestyle (four items, α = .78), passion in life (two items, α = .72), reference group acceptance (five items, α = .81), social visibility of brand (α = .72), collective memory (two items, α = .84)2. In order to fit the current research, the questionnaire was adjusted. The items asked the participants about brands that the person likes in general. In the original scale the participant is asked to keep one single brand in mind. All items were rated on a 7-point Likert-scale ranging from strongly disagree (1) to strongly agree (7). A higher score indicated a higher level of brand tribalism. A sample item is ‘I know of many people who own/use the same brands as I like’. The total scale had a reliability of α = .87. The items can be found in appendix 3.
3.3.4 Conformity Scale. Conformity was measured by adjusting items of the items developed by Cholachatpinyo and colleagues (2002) to create statements on which the participants indicate their level of agreement. The items were scored on a 7-point scale ranging from strongly disagree (1) to strongly agree (7). This answering scale was chosen for the most optimal reliability and validity (Dawes, 2008). Items 4 t/m 9 (appendix 4) are adjusted from questions into statements. A sample item is ‘My friends usually influence me in the type of clothes I buy’. All items were rated on a 7-point Likert-scale ranging from strongly disagree (1) to strongly agree (7). A higher mean score indicated a higher level of conformity. The scale had
1 The factor analysis shows that the item ‘To make sure that I buy the right product or brand, I often observe what 2 The item ‘I often discuss with friends about brands I like’ loaded higher on the factor of the social visibility dimension (.413) than on the reference group acceptance dimension (.136). However since the total scale was used and not the separate dimensions this is not a major concern.
a reliability of Cronbach’s alpha α = .68, which is slightly below the sufficient reliability coefficient of α = .70. This could be caused by a low number of items or because the construct is quite the heterogeneous (Tavakol & Dennick, 2011). The items can be found in appendix 3. 3.4 Statistical method
In the present study a quasi-experimental between-subjects design was used to test the proposed hypotheses. Prior to this analysis, assumptions were checked. The independent variable, network closure, was split into three different conditions (condition: low, moderate and high network closure). The dependent variables were susceptibility to interpersonal influence and brand tribalism. Through the program SPSS statistics (version 24) a one-way multivariate analysis of variance (MANOVA) was conducted. MANOVA was the most suitable method to check if the groups differed on a combination of dependent variables and to take the relationship between the dependent variables into account (Field, 2012). Post-hoc tests using separate ANOVA’s were conducted in order to test what conditions differ on the dependent variables individually. Discriminant analysis was conducted to check the effect of the different conditions on both variables combined. In order to test the mediation-effect of conformity, PROCESS macro using bootstrapping was used (Hayes, 2013). The tested model is shown in figure 2. Besides, factor analyses are conducted to analyze the constructed scales.
Figure 2. Mediation model of the total effect (c), indirect effect (a b) and direct effect (c’).
4. Results 4.1 Correlational analysis
In table 3 the Pearson correlations, means and standard deviations of the dependent, independent and control variables are shown. As expected the dependent variables, susceptibility
to interpersonal influence and brand tribalism, correlate highly (r = .70). One of the dimensions
of brand tribalism (reference group acceptance) is related to susceptibility to interpersonal influence. The correlation between the dependent variables does not exceed r = .90, therefore the assumption of multicollinearity was not violated (Field, 2012). Besides, it is noteworthy that some main variables correlate negatively with age, indicating a higher age is related to lower levels of network closure, susceptibility to interpersonal influence and brand tribalism.
Table 3
Pearson correlations and means (M), standard deviations (SD) and reliability coefficients of the dependent and independent variables
M SD 1. 2. 3. 4. 1. Network Closure 3.65 .66 (.70) 2. Susceptibility to Interpersonal Influence 3.44 1.01 .26** (.88) 3. Brand Tribalism 4.00 .75 .18* .71** (.84) 4. Conformity 3.28 .86 .13 .58** .43** (.68) 5. Age 30.07 13.90 –.22** –.32** –.41** –.13 * p < .05; ** p < .01 (2-tailed) 4.2 Main analyses
In order to analyze multiple dependent variables at the same time and take into account the relationship between these variables, while reducing the chances of type 1 error, multivariate analysis of variance (MANOVA) was conducted (Field, 2012). Prior to the main analysis, assumptions were checked. Assumptions regarding the independence, cell size (> 30) and equal cell sizes were met3. There were three different cells with equal sizes (low, medium and high network closure). This means that the analysis was robust if normality was violated and for the equality of variance assumption. Normality was tested looking at the Shapiro-Wilk test for non-significance. For the variable conformity this test was significant, indicating a violation of normality. Since the group sizes are equal the F-statistic was probably robust against violations of this normality (Field, 2012). The assumption of multivariate outliers4 and multicollinearity were analyzed and were met5.
A multivariate analysis of variance MANOVA (N = 162) was conducted to examine whether the network closure had an effect on the susceptibility to interpersonal influence and brand tribalism. Using Wilks’s lambda, findings showed that network closure (low, moderate, high) had a significant effect on the combined dependent variables, Λ = .954, F (2, 159) = 3.82,
p = .024, partial η2 = .046.
3 The participants could only fill in the survey once and participate one time.
4 Two cases exceeded the Maximum Mahalanobis Distance χ2 = 13.816 (df = 2 at α = .001), however, after inspection and taking into account that the Cook’s distances which were <1, it was decided to keep the data. 5 The VIF scores did not exceed 10, and tolerance was not smaller then 0.1 (Field, 2012).
The dependent variables were analyzed individually. There was no effect found on brand
tribalism. In contrast, the susceptibility to interpersonal influence was statistically significant at a
Bonferroni adjusted alpha level of .025, F (1,161) = 7.03, p = .009, partial η2 = .042. To
investigate which conditions differ significantly on susceptibility to interpersonal influence, a one-way ANOVA was conducted. The post-hoc test Tukey indicated that the low network closure condition (M = 3.20, SD = 1.08) and the high network closure condition (M = 3.70, SD = 1.00) differed significantly in their level of susceptibility to interpersonal influence p = .024. This indicated that individuals with more closure in their network also have a higher level of susceptibility to interpersonal influence.
Regarding the hypotheses it can be concluded that the three network closure conditions differed significantly on susceptibility to interpersonal influence, therefore the null hypotheses of
H1 can be rejected. However post-hoc analysis revealed that the three network closure
conditions did not differ significantly on susceptibility to interpersonal influence in the way that was expected. The effect was not large enough to cause significant differences for the moderate network closure condition. Therefore the null hypotheses of H1a and H1b cannot be rejected. The found effect is in the hypothesized direction, namely individuals with a higher closure in their network experience more susceptibility to interpersonal influence. Since it is relatively rare to do empirical quantitative research in this field, a rule of thumb is used to determine the effect size (Cohen, 1977). The found effect size (η2 = .042) indicated a low to medium effect size (Cohen, 1977). Therefore network closure played a part in affecting the individual in adopting a fashion trend by being influenced by others.
In addition, the results from the ANOVA for brand tribalism revealed that network closure did not influence a person in adopting certain fashion trends to display membership to certain tribes. The null hypotheses for H3, H3a and H3b cannot be rejected based on these results.
The MANOVA was also followed up with discriminant analysis. This post-hoc procedure can also help identify the group differences, taking all dependent variables into account (Yu & Chick, 2010). Yu and Chick (2010) recommend using both types of post-hoc analysis, F-test and discriminant analysis. Discriminant analysis revealed two discriminant functions. The first one explained 80.8 % of the variance, canonical R2 = .05, whereas the second one explained 19.2% of the variance, canonical R2 = .01. In combination these discriminant functions significantly differentiated the network closure conditions, Λ = .94, χ2(4) = 9.51, p = .05, but removing the first function indicated that the second function did not significantly differentiate the closure conditions, Λ =.99, χ2(1) = 1.84, p = 0.17. The correlations between the outcomes and the discriminant functions revealed that susceptibility to interpersonal influence loaded more highly on the first function (r = .93) then on the second function (r = .38); brand tribalism loaded more highly on the second function (r = .93) then on the first function (r = .36). The discriminant function plot showed that the first function differentiates the high from the low closure condition, the second function discriminates the moderate closure condition from the low and high closure condition.
Even though in both variate functions higher susceptibility to interpersonal influence was combined with higher brand tribalism, in variate function 1 susceptibility to interpersonal
influence was more influential then brand tribalism, and the opposite was found for variate
function 2. The high and low closure condition could be separated by a variate that had more effect on susceptibility to interpersonal influence then on brand tribalism. Figure 3 shows the canonical discriminant functions. This fits with the results of the ANOVA, which showed that low and high network closure condition differed on susceptibility to interpersonal influence. The moderate closure condition could be separated from the low and high closure condition by a variate that had more effect on brand tribalism then on susceptibility to interpersonal influence.
The MANOVA revealed that network closure had a significant effect on susceptibility to interpersonal influence and brand tribalism combined. The follow up ANOVA analyses showed
that this difference was found for susceptibility to interpersonal influence. However, to find the linear combinations of the dependent variables that best separate the network closure conditions, discriminant analysis was conducted (Field, 2012). This analysis revealed that the effects found in the MANOVA could be explained in terms of two underlying dimensions in combination. It may be assumed in the context of the present study that this underlying dimension was fashion adoption, which was assumed to be made up out of both dependent variables.
The expectation that the low, moderate and high network closure conditions differed on susceptibility to interpersonal influence and brand tribalism combined (H4) was supported by the results. The significant results in the MANOVA provided support to reject the null hypothesis of
H4. The found effect size (η2 = .046) was low to medium (Cohen, 1977).
Regarding H4a it is more likely (based on the slopes in figures 4, 5 and 6) that high network closure condition (1.25) and moderate network closure condition (1.25) had higher levels of susceptibility to interpersonal influence and brand tribalism combined then the low network closure condition (0.883).
Regarding H4b (based on the slopes in figures 4, 5 and 6) it is less likely that the high network closure condition (1.25) had higher levels of susceptibility to interpersonal influence and brand tribalism combined, then the conditions with low (0.883) and moderate network closure (1.25). Figure 7 shows the means and standard deviations of the scores on both dependent variables separated by condition. In appendix 5 the discriminant analysis tables can be found.
Figure 3. Scatterplot showing the canonical discriminant functions for the low, moderate and
high network closure conditions.
Figure 4. Mean score on susceptibility to interpersonal influence and brand tribalism in the high
Figure 5. Mean score on susceptibility to interpersonal influence and brand tribalism in the
moderate network closure condition.
Figure 6. Mean score on susceptibility to interpersonal influence and brand tribalism in the low
Figure 7. Plot showing the means and standard deviations of the low, moderate and high network
closure conditions on susceptibility to interpersonal influence and brand tribalism.
4.3 Mediating role of conformity
Hypothesis 2 proposed that conformity would mediate the relationship between network closure and susceptibility to interpersonal influence. In order to investigate whether the found effect of network closure on susceptibility to interpersonal influence was mediated by conformity, PROCESS macro using bootstrapping was used (Hayes, 2013). As can be seen in table 4 participants in a network with relatively higher closure are estimated not to differ in their level of conformity (a) (b = .14, p = .152) and participants in a network with relatively higher closure are estimated to be higher in susceptibility to interpersonal influence (b) (b = .66, p = .002). A bootstrap confidence interval for the indirect effect (ab) (b = .09) based on 5,000 bootstrap samples contained zero [–.034, .222], indicating that higher network closure did not lead to more conformity and in turn into more susceptibility to interpersonal influence. There was evidence that network closure influenced susceptibility to interpersonal influence
independent of its effect on conformity (c’= .26, p = .006). Based on these results H2 was not confirmed. Table 4 summarizes the results from the mediation analysis and table 5 gives an overview of the results per hypothesis.
Table 4
Mediation of conformity in the relationship between network closure and susceptibility to interpersonal influence
b SE 95 % CI
Network closure – conformity (a) .14 .10 [–.0525, .335]
Conformity – susceptibility to interpersonal influence (b)
.66** .08 [.508, .805]
Total effect (c) .36** .11 [.133, .581]
Direct effect (c’) .26** .09 [.078, 450]
Indirect effect (ab) .09 .07 [–.034, .222]
Note. N = 162. * p < .05; ** p < .01; R2 (a) =.01; R2(bc’) =.36
Table 5
Hypotheses, p-values and rejection of the null hypothesis
Hypothesis p-value Rejection null hypothesis
H1 p = .009 Rejected
H1a ns Not rejected
H1b ns Not rejected
H2 ns Not rejected
H3 ns Not rejected
H3a ns Not rejected
H3b ns Not rejected
H4 p = .024 Rejected
H4a More likely based on the results
H4b Less likely based on the results
5. Discussion
5.1 Reflection on the results
The aim of the present study is to examine the effect of network closure on susceptibility to interpersonal influence (1) and brand tribalism (2), and both variables combined (3). It is assumed that the influence of others on an individuals purchase behavior (Bearden et al., 1998) and desire to buy certain products to display membership to a tribe (Ruane & Wallace, 2015) reflect fashion adoption. By analyzing these concepts simultaneously, social psychological and sociological perspectives (Zayasenko, 2012) are combined in a unique way in order to explain fashion diffusion. In addition, the present study looks at whether the relation between network closure and susceptibility to interpersonal influence is mediated by conformity. This research took insights from complexity theory and is the first to take one factor, network closure, and examine its effects on the fashion diffusion process. The research question is: to what extent does
network closure affect susceptibility to interpersonal influence and brand tribalism? The results
from the present study could help explain how fashion spreads beyond apparel, explaining fashion trends in general (King & Ring, 1980).
With regard to the first question it was expected that in networks with low closure, moderate closure and high closure there are different levels of susceptibility to interpersonal influence. It is expected that in more closed networks the urge to conform is higher, norms are more salient and information spreads quickly (Baxter & Woodside, 2011; Gravetter, 1973). For more open networks the opposite was expected. Results show that the closure of the network has a small to medium effect on susceptibility to interpersonal influence. However, because the low
