Self-identification through brands

Master Thesis

Self-identification Through Brands

Rafaëla Wiertz

28-01-2016

Faculty: Economics and Business Master: Business Administration Direction: Marketing Version: Final Supervisor: dhr. drs. ing. A.C.J. (Antoon) Meulemans Student number: 6161308 Word count: 14.527

Abstract

The aim of this study is to investigate if the extent to which people use brands to identify themselves varies for different personality dimensions and their traits. In this study, the focus is on clothing brands. The following factors are thought to have a moderating effect on this relationship: inter-generational influence, peer pressure and the activity at social networking services. Additionally, it is of interest to know if there is a relationship between the different personality dimensions and their traits and the brand personalities that people choose. The results of this examination indicate that there is no significant relationship between personality and the degree to which people include brands into their self-concepts. Furthermore, none of the moderators was found to have a significant effect. Additionally, there is no significant relationship between human personality and brand personality. However, there is a significant relationship between inter-generational influences and the degree to which people incorporate brands into their self-concepts. A similar relationship was found between peer pressure and the extent to which people incorporate brands into their self-concepts. The effect of inter-generational influences appears to be twice as strong as that of peer pressure. Statement of originality

This document is written by student Rafaëla Wiertz 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 supervision of completion of the work, not for the contents.

Contents 1. Introduction to the thesis p.5 1.1. General introduction p.5 1.2. Purpose of this thesis p.6 1.2.1. Aim of the study p.6 1.2.2. Expectations p.7 1.3. Structure of the thesis p.7 2. Literature review p.9 2.1. Current literature p.9 2.2. Research gap p.10 2.3. Literature on variables p.12 2.3.1. Personality p.12 2.3.2. Self-identification through brands p.15 2.3.3. Inter-generational influences p.17 2.3.4. Peer pressure p.18 2.3.5. Social networking services p.19 2.3.6. Brand personality p.20 2.4. Conceptual Model P.22 3. Methodology p.24 3.1. Data gathering p.24 3.2. Sample p.25 3.3. Pre-test p.26 3.4. Main questionnaire p.26 3.5. Measurements p.27 4. Results p.36 4.1. Pre-test p.36 4.2. Descriptive statistics p.38 4.3. Factor analysis p.44 4.4. Creating groups p.49

4.5. Hypothesis testing p.52 4.5.1. Hypothesis 1 p.52 4.5.2. Hypothesis 2, 3 & 4 p.54 4.5.3. Hypothesis 5 p.56 5. Conclusion and Discussion p.57 5.1. Discussion p.57 5.1.1. General findings p.57 5.1.2. Answering the research question p.60 5.1.3. Link with literature p.60 5.1.4. Managerial implications p.62 5.1.5. Limitations p.63 5.2. Conclusion p.64 5.2.1. The survey p.64 5.2.2. Research contribution p.65 5.2.3. Further research p.66 6. References p.68 7. Appendix p.72 7.1. Appendix A: Pre-test 1 p.72 7.2. Appendix B: Big-5 personality scale p.74 7.3. Appendix C: BESC, Brand personality p.76 7.4. Appendix D: Main questionnaire p.77 7.5. Appendix E: Pre-test 2 p.84 7.6. Appendix F: Groups first approach p.86

1. Introduction to the thesis

1.1 General introduction “Be sure that whatever you are is you.”

These are the inspiring words of Theodore Roethke (1975, p. 171), an award winning American poet. This quote reflects the identity challenge of every individual. How do you know who the real you is? In other words, how can a person identify him or herself? There are many ways in which people attempt to identify themselves (Chernev, Hamilton, & Gal, 2011). One way is by means of brands (Marazza, and Saviolo, 2013, p. 15-31). Brands that are found representative are incorporated into the self-concept. This means that the way people see themselves is represented by certain brands; brands therefore represent who you are. This concept is called self-identification through brands, the main focus of this research.

Another way to identify oneself is by looking at a person’s personality. Personality is thought to consist of specific personality dimensions. These dimensions vary along a continuum between two extremes, representing two personality traits (Goldberg, 1981). These traits are thought to be relatively stable over time, differ among individuals, and influence behavior. The purpose of these personality dimensions and their traits is to help define people as a whole (J. Feist, & G. Feist, 2009, p. 3-14).

These two approaches of self-identification are very interesting. They fulfill a similar role, however they provide different ways of looking at self-identification. Taking these two approaches into consideration, it would be interesting to know if a relation ship between the two exists.

Besides the possibility of identifying people, brands also claim to have their own identity. This concept is called “brand personality” (Aaker, 1997). Would there then naturally be a relationship between the personality dimensions and traits of people and

that of brands? For example, do people with an extravert personality trait choose the same type of brand personality to identify themselves?

Taking into consideration the topics discussed, the main research question of the present study is: does the extent to which people use brands to identify themselves vary for different personality dimensions and their traits? Moreover, if this is the case, is there a relationship between the different personality dimensions and their traits and the brand personalities people choose?

1.2 Purpose of this thesis 1.2.1 Aim of the study

To contribute to current literature, this study intends to provide an insight into the relationship between personality and the degree to which consumers use brands to identify themselves. The dimensions Extraversion and Neuroticism will represent personality. In the literature review chapter, this choice will be explained. Subsequently, research will be done to discover if certain factors have a moderating effect on this hypothesized relationship. These so-called moderators are inter-generational influences, peer pressure and the extent to which participants are active on social media such as Facebook, LinkedIn, Instagram or Twitter.

Furthermore, the aim of this research is to clarify if there is a relationship between personality and the type of brand personalities that people chose to identify themselves with. The personality traits within the personality dimensions and all brand personality traits are taken into account. In subsequent sections of this report, the above-discussed topics will be studied into greater extent.

1.2.2 Expectations

When looking at the aim of this study, it is expected that the personality dimensions extraversion and neuroticism differ with respect to the degree to which they include brands into their self-identification. Furthermore, it is expected that within the Extravert personality dimensions, the degree of self-identification through brands is higher for the personality trait extraversion than for the personality trait introversion. Within the personality dimension neuroticism, it is expected that the degree of self-identification through brands is higher for the neurotic than for the emotionally stable personality traits.

Furthermore, it is expected that inter-generational influences, peer pressure and social media activity will increase the relationship between the personality dimensions and its traits and the degree of self-identification through brands. Finally, people with the same personality characteristics are expected to choose the same brand personality category, whether or not this is the brand personality category that is most similar to their personality trait. 1.3 Structure of thesis

The outline of this research is as follows. Firstly, a literature review chapter is presented. This chapter will elaborate upon current literature and the research gap. Followed by the main topics and hypotheses of this research. The main topics that will be discussed are: personality, degree of self-identification through brands, brand personality and possible moderation effects.

Secondly, the methodology will be described, starting with an explanation of the method of data gathering that has been applied. Secondly, information will be provided regarding the sample. The data will be collected by means of an online questionnaire. In short, the sample will consist of all ages, levels of education and nationalities. Next, the conveyed pre-test and main questionnaire will be discussed into great depth, followed by an extensive clarification of the different measurements. Within this chapter, the section regarding the independent variables will be subdivided into the following paragraphs: personality dimensions and their traits, used scale to obtain the information, the division into groups (based upon two different approaches), factor analysis, and statistical implications. The methodology chapter will be concluded with a section on the analysis.

In chapter 4, the results of this study will be presented. This will be done in the following order: pre-test, descriptive statistics regarding the general sample and the studied variables, creating groups based on personality dimensions and traits, factor analysis and hypotheses testing.

The final chapter of this research will consist of a discussion and conclusion section. The discussion section will elaborate upon the general findings, followed by answering the research question. Next, a comparison will be made with current literature. Additionally, managerial implications and possible limitations will be discussed. The conclusion will summarize the conveyed research and the results obtained. This section will clarify the contribution of this study to current literature and will give indications for possible further research.

2. Literature review

2.1 Current literature

It’s interesting to understand the role of brands in people’s self-identification, not only from a sociology point of view, but also from a marketing perspective. A better understanding of the extent to which people include brands into their self-identification is crucial for the marketing branch, as this has a significant influence on the success of companies and their brands (Matzler, Bidmon, and Grabner-Kräuter, 2006). This can be explained as follows. It is known that consumers have a need to express themselves by means of brands (Chernev, Hamilton, and Gal, 2011). According to the research of Chernev, Hamilton, and Gal (2011), brands are currently building their marketing strategies in such a way that they target consumers that incorporate the brands into their identities. Companies do so because they state that consumers that integrate a specific brand into their self-identification are very valuable for a brand due to fact that they are found to be more loyal then customers that are inclined less to incorporating brands into their self-concepts. This is believed to be due to the fact that a person’s need for self-expression is finite, meaning that when a consumer has found sufficient brands to identify him or herself, other brands are no longer of interest since the need has already been fulfilled (Chernev, Hamilton, and Gal, 2011). Additionally, such customers are not only found to be more loyal, they also provide more positive word of mouth (Tuskej, Golob, and Podnar, 2011) This means they talk positively about the brand to their surroundings, creating positive marketing for the brand. This all leads to the situation that a brand’s marketing plan is created precisely to attract such customers that indeed have high degrees of incorporating brands into their own identity. The above explains why this research is also critical from a marketing perspective.

This is found to be an important means of communicating with- and attracting customers (Levy, 1959). Currently, this thought is a widely excepted principal for both marketing specialists (Gardner and Levy, 1955) and advertising practitioners (Plummer, 1985). This concept is called “brand personality” (Aaker, 1997). Literature explains that there is a relationship between consumers and brands based on brand personality (Fournier, 1998). Nevertheless, they don’t specify the role of these consumers’ personalities. Therefore, this research will include the study of a possible relationship between human- and brand personality into its scope. However, even though there might be a relationship between the two, it doesn’t mean that they are the same. Research by Vittoria, Barbaranelli, and Guido (2001) showed that the five factors of human personality are insufficient to successfully describe brand personalities. Besides this, the meaning of the descriptors varied per brand, making them unstable predictors.

2.2 Research gap

The main research questions presented in the introduction are of extra interest since it is seemingly not possible to answer these questions by means of current literature. Previous research confirms the thought that people use brands to identify themselves. In addition, it is known that customers that do so are of great interest to brands and have great effect on their marketing plans. Nevertheless, it seems that not much research has been done to see if there is a relationship between personality and the degree to which people include brands into their self-concept. In this respect, this research is complimentary to existing literature. If a relationship is indeed found between personality and the extent to which people include brands into their self-identification, this would tell companies which type of consumers they should target

factor personality into consideration when creating their marketing plan to target their customers.

Since brands prefer customers that include brands into their self-concept, and such customers have a specific personality, brands will like to know how to attract such customers. If then a relationship between human and brand personality is found, a brand can implement a complementary brand personality to attract exactly these types of consumers. Currently, a lot of research has been done on the question if brand personality is the same as human personality trying to link each human personality dimension to the most similar brand personality category. This research applies a different approach. It studies if people with the same personality characteristics choose the same brand personality category. Neglecting if the chosen brand personality category is the one most similar to his/her own personality. In other words, the aim is to see if people with the same personality characteristics chose the same brand personality category in general.

Currently some research has been done on this topic. A research by Long-Yi (2010) found that there was a significant relationship between the personality dimension extraversion and the brand personality category excitement. Additionally, they found a relationship between the personality dimension agreeableness and the brand personality categories excitement, sincerity and competence. The authors did not specify the details of the traits within the personality dimensions. Verani (2013) also investigated the relationship between human personality and brand personality. This study showed a significant relationship between the personality dimension conscientiousness and the brand personality category excitement. This author also found a correlation between extraversion, neuroticism and openness with again the brand personality category excitement. In this study, however, only 2 out of the three

brand personality categories were measured, possibly leading to different results. In this research all five of the brand personality categories will be taken into consideration. Additionally the traits within the human personality dimensions will also be taken into account when studying the relationship with brand personality. 2.3 Literature on variables 2.3.1 Personality As explained above, personality describes who a person is and helps identify a person’s specific characteristic. There are numerous different approaches to explain the principal of personality. In this study, the Big five approach to personality is chosen. This approach investigates personality by analyzing the five basic dimensions assigned to personality: Extraversion, Neuroticism, Openness to Experience (or Intellect), Agreeableness, and Conscientiousness (Goldberg, 1981). Subsequently, within each dimension, either a high or low score can be given. This leads to a continuum within each dimension, with both extremes accounting for a certain personality trait. This can be illustrated by the example of extraversion. A person can either score high or low on the extraversion dimension. A high score on this dimension indicates that the participant is relatively extravert. On the other hand, a low score on this dimension indicates that this participant is rather introverted. So within the extraversion dimension personality traits, extraversion and introversion oppose each other on the continuum. Similarly to the extraversion dimension, neuroticism and emotional stability are the personality traits within the neuroticism dimension. The score that is obtained on the scale that quantifies these dimensions determines which trait will be assigned.

When studying people’s personalities using the Big five scale, people have a specific “score” on each dimension (Gosling, Rentfrow, and Swann Jr., 2003). This makes it challenging to study if there is indeed a relationship between certain personality dimensions and the extent to which someone uses brands to identify oneself. This issue is complicated by the fact that personality dimensions are not isolated but intertwined. For example, a person can be both contentious and extravert.

In the Handbook of competence and motivation by Elliot and Dweck (2005, p. 122-140), Dweck and Molden used a different approach to personality ignoring this interweaving of dimensions. They state that personality is ever changing and influenced by our self-theories (our beliefs about ourselves). According to this theory, the type of behavior that people show can classify people. There are two types of behavior: adaptive and maladaptive (Dweck and Leggett, 1988). The authors state that there are underlying psychological processes that cause this behavior. These psychological processes account for the idea that individual’s implicit theories orient them towards a specific goal, which in return leads them to a specific behavioral pattern (adaptive vs. maladaptive). They show how each feature (cognitive, affective, behavioral) of the adaptive or maladaptive pattern can be caused directly by different goals of an individual. This theory makes it easier to distinguish personality types, since people will be classified based on either adaptive or maladaptive behavior. However, since this study mainly focuses on specific personality traits, it is preferred to use the reputable Big five approach. Nevertheless, it would be ideal to create a clear distinction in personality types, like Dweck and Leggett have done, to be able to study the differences between personality dimensions. Taking the work of Dweck and Leggett as an example, this study will implement their approach with regards to personality. This implies that participants will either be

assigned to the extravert or the neurotic personality dimension. This approach will allow firm statements on the differences between the personality dimensions, since they are independent instead of intertwined. One may wonder: why study only two personality dimensions? This is mainly due to the practical concern that when the above explained approach of appointing the participants to only one trait is applied, there might not be sufficient participants per dimension if all dimensions are included. This would make it impossible to draw any conclusions with regards to the differences between the dimensions. If more dimensions were to be added, the amount of participants per dimension would be smaller (since it would be beyond the scope of this research to include more participants). How the participants will be assigned to one of the two dimensions will be further discussed in the methodology chapter also providing a possibly necessary second approach to calculate the groups in case the two dimensions already watered-down the quantity of participants in each group beyond an acceptable amount.

The trait groups within the dimensions will be studied in the same way. This will lead to clear-cut groups within each dimension, meaning that within a dimension, a person is clearly appointed to one of the two traits; one person can’t have both traits. By doing so, a profound analysis can be performed to study if there is a difference between the two traits within a personality dimension. Furthermore, the degree to which participants implement brands into their self-identification will be investigated.

As stated before, the two dimensions that have been chosen are extraversion and neuroticism. These dimensions are the independent variables of this study. They will be measured by means of the big five scale based upon “The Big-Five trait taxonomy:

Their scale is derived from their chapter in the Handbook of personality: Theory and research by Pervin and John (1999, p. 102-138). The reason that these two dimensions are chosen is primarily because these dimensions most likely have an effect on the extent to which people include brands into their self-identification. Extraversion can be described as enthusiastic, sociable, assertive, talkative and active. On the contrary, introversion can be defined as reserved or shy (Gosling, Rentfrow, and Swann Jr., 2003). The second dimension, Neuroticism vs. Emotional stability, can be seen as the extent to which someone is emotionally stable, meaning calm, relaxed, self-confident, and not anxious, moody, easily upset or stressed (Gosling, Rentfrow, and Swann Jr., 2003).

2.2.2 Self-Identification through brands

A study by Belk (1988) and Fournier (1998) showed that people identify themselves with brands for the sole purpose of signaling their self-identity to themselves rather than to others. This type of consumers is very interesting for brands since they are found to be more loyal than consumers that do not integrate the brand into their self-concepts (Belk, 1988). This loyalty also becomes apparent in an investigation by Marazza, and Saviolo (2013, p. 15-31). They found that consumers buy certain brands for the main goal of identifying themselves. This leads to the use of brands not only for its practical purpose, but also as a means to express oneself. They also found that the form of loyalty that these customers have goes beyond the normal form of loyalty. These consumers can become brand ambassadors. When this is the case, such brands are called “life style brands”. These brands become a very important part of the person’s self, making people essential customers due to their loyalty and their communication of

Lastly, research by Berger, and Heath (2007) additionally affirms that consumers indeed use brands to create a “unique” self.

The word “brand” is a very divers concept, as there are many different brand categories in the world, in endless contexts. For this present study, the definition of “brands” should be narrowed down. Therefore, this study chose to focus on the category “clothing brands”. In Schroeder and Salzer-Mörling’s book regarding Brand Culture (2006), Elliott and Davies explained in their chapter (p. 155-167) that clothing brands are thought to be an important category for individuals to derive an identity from. For this reason this category was primarily chosen. Besides this, it is a category of brands that everyone is exposed to and everyone relates to in one way or another.

Currently, it is known that people use brands to identify themselves, additionally such types of customers are interesting for brands since they are found to be exceptionally loyal and to some extent even create marketing for the brand. Nevertheless, the link between personality and the degree to which people use clothing brands to identify themselves has not been studied before. For this reason, no previous research can be discussed with regards to this specific topic. If it is known which personality dimensions or traits are linked to brands, the companies can target potential customers more specifically. Subsequently, the fact that no current research has been done on this subject makes it increasingly interesting to study this in more detail. The first hypotheses, taking personality and self-identification into consideration, are as follows:

Hypothesis 1a: Personality dimensions differ in the degree to which they include clothing brands into their self-identification

Hypothesis 1b: The degree of self-identification through clothing brands is higher for people with extravert personality traits than for people with introvert personality traits

Hypothesis 1c: The degree of self-identification through clothing brands is higher for people with neurotic personality traits than for people with emotionally stable personality traits

2.2.3 Inter-generational influences

It is hypothesized that the relationship between personality traits and the extent to which people use brands to identify themselves may be influenced by other, independent factors. Such factors could for example be inter-generational influences (from here on referred to as IGI) and peer pressure. A survey by Childers, & Rao (1992) confirms that IGI and peer pressure influence others beliefs, attitudes and behaviors towards brands. Additional proof for this idea was found in an investigation conveyed by Dotson, & Hyatt (2005). They found that children’s brand preference is influenced by five factors, including familial influence. Additionally, this study showed that the choice children made was influenced by other social influences, for example their peers. A study by Mandrik, Fern, & Bao (2005) confirms this finding. They showed that there is a similarity between mothers and daughters regarding brand preferences, price–quality schema, convenience orientation, value consciousness and brand name–quality schema. The influence of peers was also proven with regard to prestige sensitivity.

To conclude, it is known currently that IGI has an effect on brand preference. Whether IGI influences the relationship between personality and the extent to which

people include brands into their self-identification has not been studied previously. This could be a reasonable expectation, since clearly IGI does have an influence on brand preference, so why not on this type of brand involvement? This thought led to the following hypotheses, taking IGI into consideration as a moderator:

Hypothesis 2a: Inter-generational Influences increase the relationship between the personality dimension extraversion and the degree of self-identification through clothing brands

Hypothesis 2b: Inter-generational Influences increase the relationship between the personality dimension neuroticism and the degree of self-identification through clothing brands

2.2.4. Peer pressure

A study by Venkatesan (1966) similarly indicates the influence of peer pressure in relation to brand choice. This investigation shows that consumers accept information provided by their peer groups on quality or other properties of a product/brand that are hard to evaluate objectively. Consumers however, only accept this influence as long as they feel that the independence in the decision-making process is still maintained.

Both intergenerational and peer pressure studies indicate that there are external factors that influence our attitude and behavior towards brands. Possibly, these external factors might also influence the extent to which consumers use brands to identify themselves. For this reason, related to peer pressure, the following hypotheses can be formulated:

Hypothesis 3a: Peer pressure increases the relationship between personality dimension extraversion and the degree of self-identification through clothing brands

Hypothesis 3b: Peer pressure increases the relationship between personality dimension neuroticism and the degree of self-identification through clothing brands 2.2.5 Social networking services Regarding the influence of social networking Services (from here on referred to as SNS), Park, Song, & Ko (2011) found that SNS not only acts as a communication tool among people but also as a promotion channel for brands. For this reason, social media are also considered in this study, as they have an important influence on company’s marketing strategies on social media. In addition, Park and colleagues found that participants that differ in SNS usages also differ in their degree of brand loyalty. In the previously explained literature regarding the incorporation of brands into the self-concept, a similar link with loyalty was found. Due to this possible connection, SNS should be incorporated into this research. This was done in the following manner:

Hypothesis 4a: High activity on social networking services increases the relationship between personality dimension extraversion and the degree of self-identification through clothing brands

Hypothesis 4b: High activity on social networking services increases the relationship between personality dimension neuroticism and the degree of self-identification through clothing brands

2.2.6. Brand personality

Another relevant question is: “How can different brands be distinguished?”. As explained above, brands are thought to have their own “brand personality”. Regarding the definition of brand personality: this can be described as the set of human characteristics associated with a brand (Aaker, 1997). The next question is: how can this be measured? Can this be done by using the Big five principles of human personality? Aaker (1997) states that the Big five used for people cannot be applied to brands. However, Aaker does state that brand personality can be measured in a similar manner by using five other dimensions that categorize brands into types. These five dimensions are: sincerity, excitement, competence, sophistication and ruggedness (Aaker, 1997). There are, however, certain similarities with the Big five personality dimensions for people. For example, agreeableness and sincerity both capture the idea of warmth and acceptance. Furthermore, extroversion and excitement both refer to sociability, energy and activity. Lastly, conscientiousness and competence both encapsulate responsibility, dependability and security. Three dimensions of brand personality therefore resemble three of the dimensions of human personality (Aaker, 1997).

These findings regarding brand personality are of great importance for the marketing of a brand. A person’s preference for a certain brand depends on the overlap between their own characteristics and those of the brands (Malhotra, 1988). One could conclude that if it is clear for a brand what the main personality traits of its target audience are, it (want the brand) could strive to accommodate itself with the matching brand personality. This can for example be done through the use of brand endorsers, the brand name/symbol/logo or by the advertising style (Aaker, 1997). This would lead to a great preference of the target audience for this brand. The idea of brand personality

Personality Sincerity Down-to-earth Honest Wholesome Cheerfull Excitement Daring Spirited Imaginative Up-to-date Competence Reliable Intelligent Succesful Sophistication Upper class Charming Ruggedness Outdoorsy Tough In previous studies, it has been stated that a stronger relationship between a person’s personality and the brand personality results in a stronger preference of that brand over other brands (Malhotra, 1988). According to Aaker (1997), the five brand personality categories can be described as followed. Sincerity is described as down-to-earth, honest, wholesome and cheerful. The excitement category is seen as daring, spirited, imaginative and up-to-date. The brand personality category competence is associated with reliability, intelligence and successfulness. The category sophistication is connected to the upper class and being charming. Lastly, the category ruggedness is seen as outdoorsy and tough (see figure 1). Figure 1. Presents personality components (Aaker, 1997) The above emphasizes that the brand chosen by an individual to identify her or himself with is based on similarity. Meaning that people will choose a brand with a brand personality that is most similar to their own personality. As explained above, the more overlap there is between a persons personality and the personality category of the brand, the more the person will prefer that brand. Therefore, it might be possible that those brands are preferred that are most likely used to identify one-self. Taking brand personality and human personality together leads to the following hypothesis:

Hypothesis 5: People with a certain personality dimension or trait choose brands with a certain brand personality

2.3 Conceptual model

The following figure (figure 2) shows the conceptual model of this study and incorporates the previously stated hypotheses. The main focus of this research is to study if there is a relationship between personality types and the degree to which people use brands to identify themselves. Besides this, if there does exist a relationship, this might be influenced by inter-generational influences, peer pressure and/or activity on social networking services. Additionally, a potential relationship between personality types and brand personality will be investigated.

H 1a Figure 2. Conceptual Model for this research. Personality Dimension: -Extraversion (Extravert/Introvert) Degree of self-identification through brands Moderators: - Inter generational influence - Peer pressure - Social networking services Brand Personality Personality Dimensions and Traits H1b H2/3/4 H5 Personality Dimension: -Neuroticism (neurotic/emotional stable) H1c

3. Methodology

In this section of the paper, the method of data gathering and the sample will be discussed. Followed by the approach of the study, aiming to find an answer to the research questions defined. This will include an explanation of the implemented variables and crucial statistical analyses to substantiate these findings.

3.1 Data gathering

In order to test the previously established hypotheses, quantitative data were collected. These data were acquired through the use of self-created questionnaires. The content of the questionnaires will be elaborated further upon in depth in the paragraphs below. Within this questionnaire, the elements of interest were addressed, which in turn were subjected to statistical analysis. The questionnaires were made with the help of Qualtrics, a program that enables the creation of online questionnaires. The questionnaires were made in English, mostly because the existing measurements that where integrated into the main questionnaire of this study were also in English, therefore making translation unnecessary.

Once the questionnaires were as desired, the website provided a unique link. The link to the questionnaire was distributed via e-mail, Facebook, WhatsApp (a mobile application) or text-message. The recipients were able to complete the questions online whenever suited them best. Qualtrics saved the data automatically.

The participants of this survey were approached directly. Most of the participants recruited were acquaintances. All participants where asked to forward the questionnaire to other potential participants. The thought was that sufficient participants would be recruited digitally, so there would be no need to conduct an

offline survey alongside. During the recruitment, the participants were informed about the nature of this survey.

The desired data were gathered by means of a questionnaire because of its convenience and efficiency. Within a short time, many people can be assessed since the online link is easily distributed online. The respondents can fill out the questionnaire whenever and wherever suits them best. Once all data have been collected, Qualtrics allows easy quantitative analysis via transfer of the data to SPSS (a statistical analysis program). It should be taken into consideration that a questionnaire couldn’t be too lengthy, since otherwise the amount of respondents that complete the questionnaire will decrease. In addition, it requires a substantial amount of time to write a good quality questionnaire that takes validity and reliability into consideration.

3.2 Sample

The sample consisted of participants from the age of 17 and above, all in different phases in life, studying, working, being retired, etc. Nevertheless, it was probable that most of the participants were students, either at University or at a so called “Hogeschool’’ (university of applied sciences). The participants that are currently working are active in various fields. The wide age range and difference in levels of education rules out or at least minimizes the influence of age and education on the outcome of this study. There were no limitations with respect to nationality of participants, which limit the influence of nationality on the outcomes of this investigation. General variables such as age, gender and highest level of completed education were studied by means of descriptive statistics analyses and frequency checks in SPSS.

3.3 Pre-test

Before carrying out the survey, a pre-test was done in order to create a suitable measure for the brand personality variable (Appendix A). If the participants would have directly been asked to which of the five brand personality categories they belonged, they might have given a socially desired or biased response. For that reason, suitable clothing brands were sought-after to represent these different brand personality categories.

The Qualtrics survey started with two general questions regarding age and gender, followed by the five brand personality categories. The participants had to divide sets of two brands over the five categories. Two brands were paired together to enlarge the chance of people understanding what type of category these brands were supposed to represent. Each set of brands could only be used once. In the end, each brand personality category was linked to a set of two brands. The descriptive statistics where first studied with the help of SPSS. Furthermore, the frequencies of brands correctly linked to brand personality categories were checked by the frequency analysis using SPSS. After these frequencies were studied and found to be sufficient, the categories resulting from this pre-test could be inserted into the main questionnaire to collect the data set for the remainder of this study. 3.4 Main Questionnaire

When the necessary pre-test was completed, the main questionnaire could be constructed. The questionnaire consisted of several different components. First, several acknowledged measures were integrated into this survey, as explained above. These measures are implemented because they have been proven to be of great value in previous investigations. In addition, several questions were specifically composed for

this study. These types of questions where used to measure the possible effect of the moderators.

The questionnaire started out with asking participants several general questions, such as their age, gender and highest level of completed education. For the latter the choice ranged from high school to a PhD. These general questions were followed by questions regarding their personality. As previously explained the personality scale will be based upon the work of John and Srivastava (Appendix B). Next, the scale to measure the extent to which people use clothing brands to identify themselves was inserted (Appendix C) “Brand engagement in self-concept (BESC)” by Sprott, D., Czellar, S., Spangenberg, E. (2009). This scale formed a perfect bridge to ask the question regarding brand personality. Subsequently, the moderating effect of IGI, peer pressure and SNS were measured by means of five self-created questions. The final questionnaire consisted of a total of eight questions of which question 4 consists of 16 sub-questions, question 5 entails of 8 sub-questions and question 7 contains of 5 sub-questions (Appendix D).

3.5 Measurements

Below the involved variables will be discussed into greater detail. Also taking the necessary statistical applications into consideration. Independent variables The independent variable of this study is personality; as explained the specific focus will be on two of the five personality dimensions: extraversion and neuroticism. The aim of this study is to investigate if personality dimensions and their traits have an influence on the extent to which people include brands into their self-concept. To this end, the

approach of the Big five theory is applied.

In order to measure the personality dimensions, the Big-Five trait taxonomy: History, measurement, and theoretical perspectives scale of John, O. P., & Srivastava, S. (1999, p. 102-138) will be implemented (Appendix B). Only the items regarding to extraversion- and neuroticism dimension will be used. In total there are 16 questions. The questions are scored on a five-point likert scale with 1 being ‘strongly disagree’ and 5 being ‘strongly agree’. The questions will be assessed in a random order. Amongst the questions are six counter indicative items. This means that these questions are asked in a reversed way. For example, instead of asking if someone likes ice cream it is asked if someone doesn’t like ice cream. These types of questions are necessary to avoid people going into an affirmative answering mode. Later on, these questions will be reversed again to assure that they are scored in the same way as the other items.

When applying factor analysis, the importance of each question within a scale is measured. Not all questions within a scale are equally important and for that reason weights have to be attached to each question. When factor analysis is applied correctly, these weights will become clear and can be attached to each question, calculating the correct score on the personality scale per participant. Furthermore, factor analysis can be used to decide if a question in a scale should be kept or not. If the weight of a question is found to be relatively low, this question does not add much value to the scale and there can be decided to delete this question. By means of factor analysis, it can also be checked if the questions that are supposed to test the separate personality dimensions indeed do so. It would for example be problematic if there is assumed that certain questions measure the personality dimension extraversion, whilst in reality two of the

survey were subjected to principal components analysis (PCA) using SPSS. Before PCA could be applied, suitability of data for factor analysis had to be assessed. In order to do so, the correlation matrix was studied in SPSS. Once the factors were decided upon using the previous analysis, oblim rotation was performed. This was done in order to take a closer look at the possibilities of these components. The next step was to divide the participants into either the extravert or neurotic dimension groups. This could be done in two different ways. The first approach for dividing the participants over the personality dimensions and traits was performed as follows. Firstly, participants mean scores were calculated for the question sets extraversion and neuroticism, taking the right weight per question into account. Next, the mean scores where compared. The dimension delivering the highest mean represents the dimension that a participant would belong to. For example, if a participant had a mean score of 3 for the extraversion dimension and a mean score of 2 for the neuroticism dimension, this participant would be classified as extravert. If the mean scores both were the same, the participant was excluded from the survey. If this study were to include more than two dimensions, the mean differences between the dimensions would become too vague, making it impossible to create robust groups.

After these groups where made, the participants had to be assigned to either one of the personality traits within this dimensions. This was done by means of studying the mean score of a participant on this dimension. If the participant’s score were equal to or smaller than 3, this participant would be introvert or emotionally stable. On the contrary, if a participant’s mean score was larger than 3 this means that he or she was extravert or neurotic.

If this approach of dividing the participants over the two dimensions did not deliver a sufficient amount of participants per group, the second approach could be applied. This approach implements a two by two model for the personality traits, including all the participants into both the extravert and the neurotic personality dimensions. Into which personality trait within each dimension the participants are categorized is calculated as follows. The participants were divided by calculating the overall mean score on the dimension of interest. If participants scored above this overall mean score for the extravert dimension, they belonged to the extravert trait group. Participants scoring below the mean level were part of the introvert trait group. Likewise, participants scoring above the overall mean score within the neurotic dimension belonged to the neurotic trait group. Participants scoring below the mean level were part of the emotionally stable trait group.

Due to the two by two model, each trait consisted of two groups (table 1). For the extraversion trait these groups were: extravert/neurotic and extravert/emotionally stable. By accumulating these two groups, the total amount of participants belonging to the extravert trait has been defined. For the introvert personality trait, these groups were: introvert/neurotic and introvert/emotionally stable. The same approach was applied to the neurotic personality dimension. The groups belonging to neuroticism trait were: neurotic/extravert participants and the neurotic/introvert participants. The emotionally stable trait consisted of the following participants: emotionally stable/extravert and emotionally stable/introvert.

Regarding these two approaches, the first approach is preferred, since it enables

a comparison of the two personality dimensions using their score on the degree of self-identification through brands. However, the second approach may be preferred from a statistical angle, since the amount of participants in each group is larger. Table 1 Two by two model, second approach creating groups Personality dimension Neuroticism Neurotic Emotionally stable Personality dimension Extraversion Extravert Extrovert/neurotic Extrovert/emotionally stable

Introvert Introvert/neurotic Introvert/emotionally stable

Note. Two by two model for the independent variables

This second approach of dividing the participants into groups includes all the participants into both the extravert and the neurotic personality dimension. By doing so it is no longer possible to measure the difference between the two personality dimensions, since the same participants are present in both groups. As a consequence, it has to be accepted that hypothesis 1a (Personality types differ in the degree to which they include brands into their self-identification) could no longer be tested.

To substantiate this measurement, statistical analysis will be applied. The statistical implications for these independent variables were first studied by looking at the descriptive statistics using SPSS. Followed by a factor analysis by means of running a principal component analysis (PCA) in SPSS. Once the factor analysis was completed, the groups could be made. In the first approach, Excel was used to calculate the means. The second approach implemented visual binning with the help of SPSS.

Dependent variable 1

The main dependent variable in this study is the degree to which people use clothing brands to identify themselves. The scale that will be implemented to study this variable is described by Sprott, D., Czellar, S., Spangenberg, E. (2009). The scale “Brand engagement in self-concept (BESC)” consists of eight questions. These questions are answered using a seven-point likert scale. The answers vary between 1 = strongly disagree and 7 = strongly agree.

Regarding the statistical implication to validate this variable, firstly the variable descriptive statistics in SPSS were studied. Secondly, the aim of this variable was to measure if there was a difference between the personality dimensions in the degree to which people use clothing brands to identify themselves. This possible difference was measured by means of implementing either a t-test or an ANOVA in SPSS. It was also of interest to evaluate if there was a difference between the personality traits within a dimension, as well as the degree to which people use clothing brands to identify themselves. This was studied by means of an univariate analysis of variance in SPSS.

Moderation variables

When the relationships between different personality traits and the degree of self-identification through brands are measured, it is expected that additional variables will have an influence on this relationship. Variables that are thought to have an influence on an existing relationship are called moderators. In this study, three moderating variables were taken into account: IGI, peer pressure and SNS.

IGI is thought to affect this relationship due to the fact that direct family or other relatives may influence people in their behavior. In this case, this means that if

someone’s family or close relatives are used to include clothing brands into their self-identification, this person is more likely to do so as well.

In the same manner, peer pressure is thought to influence the relationship between personality and the degree to which clothing brands are included into the self-identification. This means that if someone’s peers tend to include clothing brands into their self-concept, this person is more likely to do so as well.

Finally, it is hypothesized that the degree of activity on SNS has an influence on the relationship between personality and the degree to which clothing brands are included into the self-identification. This means that if someone is more active on social media, this person is thought to incorporate a greater degree of self-identification through brands.

Each of these moderators where individually assessed by asking participants to rate the following questions:

• Inter-generational influence

o Your family members incorporate clothing brands into their self-identification

o Your family influences the degree of clothing brand inclusion into your self-identification

• Peer pressure

o Your friends incorporate clothing brands into their self-identification o Your friends influence the degree to which you include clothing brands

into your self-identification • Social networking services

o You are very active (more than 2 hours a day) on social media platforms such as Facebook, Twitter, or Instagram

All questions could be rated on a five-point likert scale with 1 being strongly disagree and 5 being strongly agree. After these measures are made, their effect on the main relationship between personality traits and the degree of self-identification through clothing brands can be assessed.

To study the possible influence of these moderators on the relationship between personality traits and the degree to which people include brands into their self-concept, a univariate analysis of variance was conducted in SPSS. This is the same type of analysis that was used to study the relationship between personality and the degree of self-identification for brands. This time, however, it is subsequently checked if one of these moderators influences this relationship. Additionally, the descriptive statistics of the moderators are studied in SPSS.

Dependent variable 2

The second dependent variable incorporated into this research is brand personality. Aaker (1997) states that there are 5 different brand personality categories. In order to measure this variable without directly asking to which category participants feel they belonged, two representative brands were linked to each individual brand personality category. This was done by means of a pre-test, which will be elaborated upon in the following chapter. When the pre-test was done, the brands chosen could be incorporated into a question in the main survey. Participants were asked to choose one of these brand sets. They had to choose the set that represents them best. The brand set chosen could be reduced to a brand personality category.

Regarding the statistical implication to validate this variable, firstly the descriptive statistics and frequencies were studied in SPSS. Secondly, the possible

relationship between personality traits and personality brands were studied by means of a cross-tabulation in SPSS.

4. Results

In this chapter, the results of the empirical study will be presented, starting with a discussion on the pre-test. This is followed by the study of the main data, beginning with the descriptive statistics, followed by the division of the participants into groups. Lastly, the hypotheses will be tested.

4.1 Pre-test

As explained above, a pre-test had to be done in order to find representative clothing brands for the brand personality categories. The results of the first pre-test were as follows. There were 29 participants of whom 28% was male and 72% female. The age of the participants varied between 23 and 28 years. The mean age was 25 years, with a standard deviation of 1.09 years. The median of age was 25 years, which is not a surprise as most of the participants were students. The percentages of brands correctly linked to the respective categories were as shown below in table 2. Table 2 % Correctly linked clothing brands to brand personality categories Clothing brand categories Excitement (H&M/Zara) Sincerity (Benetton/Hema) Ruggedness (Levis/The north face) Sophisticated (Chanel/Hugo Boss) Competence (Tommy Hilfiger/America Today) Sample Size n = 29 n = 29 n = 29 n = 29 n = 29 % Answered correctly 72% 69% 97% 100% 45%

It is notable that the percentage of brands correctly linked to the category competence is only 45%. Due to this finding another pre-test was performed using different brands (Appendix E).

In this new pre-test, the clothing brands belonging to the brand personality categories Excitement and Competence were changed with the intention to increase the percentage of brands correctly linked to those categories. The category Excitement was no longer linked to H&M/Zara, but to Forever 21/Zara. The Competence category was no longer represented by Tommy Hilfiger/America Today, but by H&M/Nike. In the second pre-test, twenty-two people participated, of which 27% was male and 73% female. The age of the participants varied between 22 and 28 years. The mean age was 24 years with a standard deviation of 1.26 years. The median for age was 25 years again. Competence again had the lowest correspondence of all categories (see table 3). This was mostly due to the fact that participants considered the category competence similar to the category sincerity. The different brands and new outcomes were as follows: Table 3 Note. The percentage correctly linked brands is still fairly low for the sophisticated category % Correctly linked clothing brands to brand personality categories Clothing brand categories Excitement (Forever 21/Zara) Sincerity (Benetton/Hema) Ruggedness (Levis/The north face) Sophisticated (Chanel/Hugo Boss) Competence (H&M/Nike) Sample Size n = 29 n = 29 n = 29 n = 29 n = 29 % Answered correctly 54% 77% 100% 100% 46%

In this second pre-test, the correct score for excitement was 18% lower than in the first pre-test. The scores for sincerity increased with 8% and the score for ruggedness increased with 3%. However, the aim of this pretest was to increase the score on Competence, which only increased by 1%. Based on these findings, it was chosen to include the brands from the first pre-test into the main questionnaire, taking into account that the brands representing competence are not as robust as desired. In the discussion this note has been included. This modification will be elaborated upon in the discussion section. 4.2 Descriptive statistics General implications sample

The aim of this research was to obtain at least 30 participants per personality trait, meaning 60 participants per personality dimension, leading to a desired total of at least 120 participants. The final amount of participants was 174. Of these participants 19.5% was male and 80.5% female. This clearly led to skewing in the distribution of gender, with skewness of -1.55 (SE = 0.18) and kurtosis of 0.41 (SE = 0.37).

The age of the participants ranged between 16 and 78 years. The mean age was 32.60 years with a standard deviation of 14.29 years. The median of age was 25 years. In addition, age was non-normally distributed with skewness of 1.13 (SE = 0.18) and kurtosis of -0.11 (SE = 0.37). This is due to the fact that there is a high frequency of participants that are around either 25 years old or 55 years old (Figure 3).

Figure 3. This histogram shows a non-normal distribution with a large peak around the age of 25 and

again a smaller peak at the age of 50 years.

The participants where asked to select the highest level of education that they had completed. Amongst the participants 17.2% completed high school, 16.1% completed higher education at the level of applied sciences (‘HBO’), 23.6% had a university bachelors degree, 41.4% had a university masters degree and only 1.7% a PhD degree. This leaded to a relatively normal distribution for education with skewness of -.50 (SE = 0.18) and kurtosis of -1.07 (SE = 0.37). There are neither missing values nor outliers for the variables age, gender or highest level of completed education. Fr eq uen cy Age

Independent variables

For the variables extraversion and neuroticism 4 values were missing from 174 participants. These were deleted pair-wise during further analysis. This manner of dealing with missing data was applied to all missing data in this survey. Therefore, this will not be mentioned beyond this point. The questions measuring extraversion were found to be reliable (8 items; α = .78). The mean score was 3.87 with a standard deviation of 0.55. The minimum score on this scale was 1.67; the maximum score was 5 and the median 3.94. The extraversion dimension was non-normally distributed with skewness of -0.76 (SE = 0.18) and kurtosis of 1.57 (SE = 0.37). Extraversion was found to be negatively skewed, meaning that the mass of the distribution was concentrated on the right side of the figure (Figure 4). This indicates that the greater deal of the participants scored fairly high on the extraversion scale.

The questions measuring neuroticism were found to be reliable (8 items; α = .79). The neuroticism dimension was normally distributed with skewness of 0.26 (SE = 0.18) and kurtosis of -0.38 (SE = 0.37). For this dimension a mean score was found of 2.51 with a standard deviation of 0.68. The minimum score on this scale was 1 and the maximum score was 4.22; the median was 2.36.

Participants Scores on Extraversion Scale Figure 4. This histogram shows a slightly skewed distribution of the scores on the extraversion scale. Dependent variable 1 The variable “degree of self identification through clothing brands” missed 4 values out of 174. The scale was found to be highly reliable (8 items; α = .92). The mean score for this measurement was 3.34 with a standard deviation of 1.26. The minimum score on this scale was 1 and the maximum score was 6.25; the median was 3.25. This variable is normally distributed with skewness of 0.08 (SE = 0.19) and kurtosis of -0.95 (SE = 0.38). Fr eq uen cy

Moderators

When studying the descriptive statistics of the moderator IGI, a mean score was found of 3.03 with a standard deviation of 1.33. The minimum score on this scale was 1 and the maximum score was 6; the median was 3. The scale consisted of two questions ( = .59). A relatively normal distribution was found with skewness of 0.52 (SE = 0.19) and kurtosis of -0.68 (SE = 0.38). For the moderator peer pressure the values where 3.58 for the mean with a standard deviation of 1.33. The minimum score on this scale was 1 and the maximum score was 6, the median was 3.50. Regarding the distribution, the skewness was -0.01 (SE = 0.19) and kurtosis -0.68 (SE = 0.38), also representing a normal distribution. The scale consisted of two questions ( = .64). On the contrary, the SNS variable demonstrated a relatively non-normal distribution with values of skewness of 0.10 (SE = 0.19) and kurtosis of -1.46 (SE = 0.38). The scale consisted of only one question, with a large peak for value 2 for the category ‘disagree’ and the value 6 for ‘agree’, respectively (figure 5). This was due to the fact that participants either agreed or disagreed as a result of the formulation of the question. The question was “I am very active (more than 2 hours a day) on social media platforms such as Facebook, Twitter, or Instagram”. This type of question formulation automatically excludes other answers besides agree or disagree. A mean score was found of 3.60 with a standard deviation of 1.96. The minimum score on this scale was 1 and the maximum score was 7; the median was 3.50. The abovementioned moderators all had 10 missing values out of 174 participants; these where deleted pair-wise as well.

Participants Scores on Social Media Scale Figure 5. This histogram shows a non-normal distribution on the activity on social media measure. Dependent variable 2

The second dependent variable brand personality had 9 missing value’s out of 174 participants. The brand personality measure consisted of only one question. Participants had to chose a clothing brand of preference. The results were as follows: Fr eq uen cy

Table 4

Based on the table above it becomes clear that the variable was assessed to be non-normal with skewness of 0.34 (SE = 0.19) and kurtosis of -1.55 (SE = 0.38). This was the result of almost 50% of the participants choosing the first clothing brand option. 4.3 Factor analysis Before the groups for the personality dimensions extraversion and neuroticism could be created, the impact of each question of the personality scale had to be studied. In order to do so factor analysis was applied. The data were found suitable for factor analysis. The correlation matrix showed a sufficient amount of coefficients that were equal to .3 or above. Additionally, the Kaiser-Meyer-Oklin (1970, 1974) value was .80, which exceeds the recommended value of .6. Furthermore, Bartlett’s Test of Sphericity (1954) reached statistical significance (P = .000), ensuring the factorability of the matrix.

The Principal component analysis revealed the presence of four components with % Chosen Categories by Participants Clothing Brand Categories % chosen H&M/Zara 48% Benetton/Hema 3% Levis/The north face 14% Chanel/Hugo Boss 19% Tommy Hilfiger/America Today 16%

studying the scree plot (figure 6), it could be decided to retain only two components since it showed a clear break after the second component. These two components capture much more of the variance than the remaining other components. Table 5 Component Matrix Component 1 2 3 4 NQ7.4 ,692 ,354 Q7.11 -,612 ,404 NQ7.10 ,600 ,370 ,334 Q7.16 ,582 ,445 -,378 Q7.8 ,575 ,523 Q7.5 -,561 Q7.2 ,530 ,320 NQ7_14 ,518 ,326 ,441 Q7.7 -,506 ,372 Q7.1 -,406 ,657 -,340 Q7.15 -,448 ,598 NQ7.9 -,452 ,550 ,402 NQ7_3 -,369 ,378 ,539 NQ7.13 -,458 ,538 Q7.6 ,384 -,368 ,638 Q7.12 ,314 ,324 -,353 ,559 Note. Unrotated loadings of each of the items on the four components

Component Number

Figure 6. This Scree plot shows the amount of variance explained per component. There is a clear brake

after the second component, likely implying involvement of two relevant components.

A total of 42.9% of the variance was explained by the two-component solution, with component one contributing 25.3% and component two 17.6%.

Oblim rotation revealed a simple structure, with both components showing a number of strong loadings. Additionally, all variables loaded substantially on only one of the components. The items representing the personality trait extraversion strongly loaded on component 1 and the items representing the trait neuroticism loaded strongly on component 2. Furthermore, there was a weak negative correlation between the two factors (r = - .14). The result of this analysis supports the use of the extraversion personality trait and neuroticism personality trait as separate scales. Ei ge nv alu e

When studying the communality values for this two-factor solution, the items NQ7.3, Q7.6, Q7.12 and NQ7.13 where found to only explain only a very small quantity (less than .3) of the variance (see table 6 for precise values). Besides this, these items also had a relatively low loading onto their component. For this reason it was chosen to exclude these questions from this scale. The new percentage of variance explained by this two-component solution is increased to 52.1%. Component one contributed 31.1% and component two 21%. As a result, the overall factor loadings of each variable, the correlation between variables and factors and the variance each item explained slightly increased (see table 7). Lastly, the new negative correlation between the two factors was r = - .17, which is slightly higher, but nevertheless still very low. Due to the deletion of these questions, the Cronbach’s alpha for the remaining 6 questions for extraversion increased from .78 to .81. For neuroticism, Cronbach’s alpha slightly decreased from .79 to .78. Both question sets were still found to be reliable.