Index
Abstract 1. Introduction 2. Literature review
2.1. Emotional brand attachment 2.2. Self-congruence
2.2.1. Actual self-congruence 2.2.2. Ideal self-congruence 2.2.3. Social self-congruence 2.2.4. Ideal social self-congruence 2.3. Moderating effect: consumption context
2.3.1. Consumption context and actual self-congruence 2.3.2. Consumption context and ideal self-congruence 2.3.3. Consumption context and social self-congruence 2.3.4. Consumption context and ideal social self-congruence 2.4. (Moderator) control variables
3. Methodology 3.1. Research design 3.2. Research object 3.3. Questionnaire design 3.4. Statistical analyses 4. Results 4.1. Sample description 4.2. Validity & reliability 4.3. Testing the hypotheses 4.4. Additional analyses 4.4.1. Product involvement 4.4.2. Self-esteem 4.4.3. Public self-consciousness 5. Conclusion 5.1. Discussion
5.2. Theoretical and managerial implications 5.3. Limitations and suggestions for future research
1 2 4 4 5 5 6 7 8 9 9 10 11 12 13 15 15 16 16 17 19 19 21 23 25 25 29 33 37 37 39 41
References
Appendix A - Operational table Appendix B – Questionnaire
Appendix C - Pattern matrices of the validity analyses
43 48 52 58
Abstract
This study aims to provide new insights into how the different types of self-congruence influence emotional brand attachment and what the moderating effect of consumption context is on this. It is possible that the ideal image and the public situation have become more important because of the increased use of social media. 159 Dutch respondents participated in an experiment where they were forced into either a private or a public consumption context. They had to answer questions about different types of self-congruence and emotional brand attachment with a certain beer brand in their minds. Additionally, there were some questions about perceived quality, product involvement, self-esteem, and public self-consciousness as (moderator) control variables. The results showed that product involvement was the most important variable to influence emotional brand attachment. After that, perceived quality was the most important. Next to this, the positive effect of actual self-congruence on emotional brand attachment appeared to be strengthened in the private consumption context. Moreover, the main effect of the different types of self-congruence had a positive effect on emotional brand attachment when public self-consciousness and self-esteem were present. This shows on which type of self-congruence or on which other variable a marketing manager has to focus on in which situation to get high emotional brand attachment.
Key words: self-congruence; emotional brand attachment; consumption context; perceived quality; product involvement; self-esteem; public self-consciousness
1. Introduction
When walking through a supermarket, you are exposed to a lot of different brands. Many of them appear to be similar since they offer the same product, like different brands of cola for instance. Therefore, brands have to find a way to stand out and get the attention of customers, instead of losing customers to competition. One way to do this, is to focus on emotional brand attachment which is defined as “the strength of the bond connecting the brand with the self” (Park, MacInnis, Priester, Eisingerich, & Lacobucci, 2010, p. 2). Companies are increasingly focusing on this aspect, because it is required to create a strong brand (Keller, 2001) and research has shown that it has multiple positive outcomes. For example, emotional brand attachment not only leads to resilience to negative information, an act of defending, intentions to recommend, purchase and revisit (Japutra, Ekinci, & Simkin, 2014), but also to stronger brand loyalty and brand performance (Park et al., 2010).
The importance of the self already appears from the definition of emotional brand attachment. Likewise, self-congruity is an antecedent of emotional brand attachment (Malär, Krohmer, Hoyer, & Nyffenegger, 2011) and is even required to occur for emotional brand attachment to take place (Chaplin & John, 2005; Park et al., 2010). This refers to a fit between a consumers’ self-concept and the brand’s personality or image (Aaker, 1999; Kressmann, Sirgy, Herrmann, Huber, Huber, & Lee, 2006, p. 955; Sirgy, 1982).
Next to this, it is important to keep in mind that the effect of self-congruence on emotional brand attachment can differ between customers and contexts. Therefore, this study will focus on the consumption context as a moderator. This refers to whether a consumed product can be seen by others (publicly consumed) or cannot be seen by others (privately consumed) (Bearden & Etzel, 1982). Moreover, this variable has already been investigated as a moderator on the relationship between self-concept and image-congruence (Dolich, in Graeff, 1997; Onkvisit & Shaw, 1987), self-congruence and brand evaluation (Graeff, 1996), and between self-congruence and brand attitude, and purchase intention (Graeff, 1997). Besides, both consumption context and self-congruence are originated from categories that are focused on private and public characteristics. Therefore, consumption context can be an important moderator in this study.
Furthermore, brands are used by people to make a desired self through their possessions (Thompson & Hirschman, 1995), which is also possible via social media. This way, for example people can post photos of themselves with a brand or ‘like’ Facebook pages of a brand (Hollenbeck & Kaikati, 2012). Moreover, the use of social media has increased since the last years (ZIGT, 2018). This may mean that people have become more conscious that others are able to see what they are doing when they post something and that therefore the public consumption context has become more important than the private consumption context. On the other hand, the emphasis can be more on the public self than on the private self, since the opinion of others seems to be more pressing than before. Next to this, the increased use of social media is attended with pressure to perform and the ideal image that people sketch for
themselves and others (Metronieuws, 2017). Likewise, Chou and Edge (2012, p. 1) state that “people tend to present themselves in a favorable way on their Facebook profile”. This may indicate that the ideal image has become more important than the actual image. It is possible that these shifts have also caused a shift in the effects of the different types of self-congruence and consumption context on emotional brand attachment. At this point, it is not clear when a manager has to focus on which congruence in which type of situation to achieve emotional brand attachment. Moreover, self-congruence exists of different types and the influence of some types on emotional brand attachment has not been investigated yet. On the other hand, there are no studies that investigate what the influence of the consumption context is on the relationship between self-congruence and emotional brand attachment. Thus, the research objective of this study is to understand what the influence of self-congruence on emotional brand attachment is and how this effect is influenced by the consumption context (private vs. public).
This research will contribute to the knowledge regarding the different types of self-congruence, emotional brand attachment, and consumption context. First, because it will confirm or reject previous information about the different types of congruence (regarding the actual and the ideal self-congruence) in relationship to emotional brand attachment, since there are contradictory theories and studies about it. Second, it will show if the other types of self-congruence (the social and the ideal social self-congruence) have an effect on emotional brand attachment and what this effect would be. And third, because it will show if the consumption context has a moderating effect on the relationship between the different types of self-congruence and the emotional brand attachment, and if so, what this influence would be. Next to this, it will provide implications for managers about on which type of self-congruence they have to focus when advertising for their brand to achieve emotional brand attachment. So, it will become clear which type of self-congruence has a positive, negative of no effect on emotional brand attachment. Furthermore, the outcomes of the analyses including the moderator ‘consumption context’ will show on which type of self-congruence managers have to focus depending on if they have a publicly or a privately consumed brand to get higher emotional brand attachment.
First of all, this study focuses on previous studies to see what is already known about emotional brand attachment, the different types of self-congruence, and consumption context. Based upon this literature review, relationships and hypotheses are developed. After this, the method about how to investigate these hypotheses is explained. Consequently, the results are analyzed and discussed, from which a conclusion has been drawn. In the end, implications, limitations and suggestions for future research are provided.
2. Literature review
In this literature review the first focus is on emotional brand attachment. After that, the different types of self-congruence are explained and their main effects on emotional brand attachment are hypothesized. Subsequently, the moderating effects of consumption context on the relationships between the different types of self-congruence and emotional brand attachment are hypothesized. All these hypotheses are brought together in a conceptual model (figure 1).
2.1. Emotional brand attachment
The dependent variable of this study is emotional brand attachment. The definition of attachment given by the Cambridge Dictionary (n.d.) is “a feeling of love or strong connection to someone or something”. Regarding emotional brand attachment, this ‘something’ refers to a brand. Likewise, Keller (2001, p. 29-30) explains the definition of attachment by the use of the following four items: “I really love this brand”, “I would really miss this brand if it went away”, “this brand is special to me”, and “this brand is more than a product to me”. This already shows the importance of the connection between the brand and the person. To use attachment in the marketing context, Malär et al. (2011, p. 36) describes attachment as “the bond that connects a consumer with a specific brand and involves feelings toward the brand”. So, when there is high emotional brand attachment, there is a strong connection between a consumer and a brand and the consumer will have positive feelings towards the brand. When there is a level of low emotional brand attachment, there will be a negative or no bond between the consumer and the brand and the consumer will have negative or no feelings towards the brand. To know how strong this connection is, Thomson, MacInnis and Park (2005) investigated the strength of emotional brand attachment and found that this is determined by the factors: connection, affection, and passion.
Moreover, this construct is important for brands, since it is essential for the category of resonance. Resonance is “the ultimate relationship and level of identification that the customer has with the brand” which is the highest rank of the customer-based brand equity pyramid and part of the latest step to create a strong brand (Keller, 2001, p. 15). Moreover, it can lead to resilience to negative information, an act of defending, and intentions to recommend, purchase and revisit (Japutra et al., 2014), and to stronger brand loyalty and brand performance (Park et al., 2010).
But how to accomplish emotional brand attachment? There are several studies that show different antecedents, such as user-derived benefits, superior marketing characteristics, traditional customer outcomes, socialization, sentimentality or emotional memory (Grisaffe & Nguyen, 2011), reputation, quality, experience, responsiveness (Japutra et al., 2014), self-congruity (Malär et al., 2011), and store-evoked pleasure and arousal on the basis of consumer personalities (Orth, Limon, & Rose, 2010). The independent variable this study focuses on is self-congruence which will be explained below.
2.2. Self-Congruence
To explain the concept of self-congruence, it is useful to look at value-congruence. This is defined as “the similarity between a consumer’s own personal values and perceptions of service brand values” (Zhang & Bloemer, 2008, p. 161). So, it is about the values that a person thinks that he has and if these are equal to the values that he thinks that belong to a brand. If these are similar, it means that there is high congruence. If the values do not match, there is low or no congruence. In this study, the focus is on self-congruence. Like earlier stated, this refers to a fit between consumers’ self-concept and the brand’s personality or image (Aaker, 1999; Kressmann et al., 2006; Sirgy, 1982). So, it is about how a consumer perceives the characteristics of a brand and how a consumer perceives his own characteristics and if these two are similar. Consumers can achieve this congruence by finding a match of personality between a brand and one of their types of self-concept.
The self-concept exists of two different categories: the private self and the public self. The private self is about what a person thinks about himself. This exists of the actual self, which refers to “how a person perceives herself”, and the ideal self, which refers to “how a person would like to perceive himself” (Sirgy, 1982, p. 287). On the other hand, the public self is about what a person thinks that others think of him. This exists of the social self, which refers to “how a person presents himself to others”, so how a consumer thinks that other people see him, and the ideal social self, which refers to how a consumer would like others to perceive him (Sirgy, 1982, p. 287; Sirgy, Rahtz, & Portolese, 2014). Aaker (1999) indicates that consumers want to achieve self-congruence to provide self-expressive or symbolic benefits for themselves. Below, the different types of self-congruence will be explained and their relationships with emotional brand attachment will be hypothesized.
2.2.1. Actual self-congruence
The first private self is the actual self. A congruence with the actual self means that the brand represents the self in the way how a person sees himself or his personal identity (Sirgy et al., 2014). An example for this is ‘The real beauty’ campaign of Dove (n.d.). In this campaign they do not show perfectly shaped models, but women with different postures. This way, people will be able to compare their actual self with the campaign, because they can compare their body shape with one of the models that is similar to them.
The reason why people want to comply to their actual self is self-consistency. This means that people feel uncomfortable if they do things that are not in line with their true self (Sirgy et al., 2014). Therefore, they have the motivation to behave consistently with their view of themselves (Sirgy, 1982). Moreover, Smith (1998, p. 7) states that “people are attracted to, prefer, and support relationships with similar others to maintain balance in their self-identity” which is supported by the similarity-attraction
theory (Byrne, 1971). This way, people will also be attracted to, prefer, and support relationships with brands that are similar to the view of their self-image to comply to their actual self.
Next to this, the study of Malär et al. (2011) already confirms the positive relationship between actual self-congruence and emotional brand attachment. They state that this result may be an antecedent for the success of authentic branding, whereby they refer to Harter who states that “a person’s authenticity is reaffirmed when he or she acts in ways that reflect the “real me” or “my true self” and is discouraged when acting in ways that they feel are phony or artificial” (Harter, in Malär et al., 2011, p. 43). Besides this, they refer to the construal-level theory, because high-level construal (which states for a more concrete brand information-processing perspective, so focusing on the actual self) might bring the brand psychologically closer to the consumer which leads to a more concrete brand-related mind-set. Likewise, Japtura et al. (2014) indicate this positive effect with a qualitative study. They state that people are attached to a brand that is congruent with their actual self, because it reflects who they are and they can show their actual identity to other people by using this brand. This is why the first hypothesis is stated as followed:
H1: Actual self-congruence has a positive effect on emotional brand attachment.
2.2.2. Ideal self-congruence
The other private self is the ideal self. This refers to how a person would like to be seen by himself (Sirgy et al., 2014). An example for a brand that can represent the ideal self is Axe. They have commercials where men are positioned as seducers (Unilever, n.d.). The only thing that a man needs to do to is use Axe deodorant and immediately every women will be attracted to him. When the ideal self of somebody is being a very attractive man or a seducer, he will use Axe to achieve this ideal self.
The motivation of people to meet their ideal self is self-esteem. This refers to “the tendency to seek experiences that enhance self-concept” (Sirgy, 1982, p. 287). So, people want to use and buy brands that have an image similar to their ideal self that helps them to make them feel good about themselves (Sirgy et al., 2014). This is also supported by the self-enhancement theory which means that “people are motivated to increase their feelings of personal worth” (Swann, Griffin, Predmore, & Gaines, 1987, p. 881). Additionally, Escalas and Bettman (2003, p. 346) indicate that “people who are motivated to enhance their self-concept form connections to brands that are used by groups they aspired to belong to”. This may indicate that because of the aspired feelings to the group, people will also have aspired feelings to the brands. Likewise, Smith (1998, p. 7) states that “people are more attracted to, prefer, and support relationships with similar others to reinforce their self-esteem”. So, people who are motivated to comply to their ideal self will be attracted to and make connections to brands that they perceive as similar to their ideal self.
Furthermore, Japtura et al. (2014) found in a qualitative study that the ideal self-congruence has a positive influence on emotional brand attachment, because a respondent told them that she became attached to a brand since it was an aspiration of who she would like to be in the future. On the other hand, Malär et al. (2011) did not find a significant effect of ideal self-congruence on emotional brand attachment.
Although Malär et al. (2011) did not find a significant main effect of ideal self-congruence on emotional brand attachment, there are multiple theories and the study of Japtura et al. (2014) who state the opposite and show a positive main effect. Therefore the second hypothesis is stated as followed:
H2: Ideal self-congruence has a positive effect on emotional brand attachment.
2.2.3. Social self-congruence
The first public self is the social self, which refers to how a person believes how others see or think about him (Sirgy et al., 2014). This is connected to social consistency, which means that “people are motivated to maintain an image others have of them. They feel uncomfortable if they act in ways that are inconsistent with how they believe others see them” (Sirgy et al., 2014, Chapter 3, Self-image, para. 5). For example, the brand Diesel has a campaign of ‘Be stupid’ (Creative ad awards, n.d.). These ads show different photos of people who are stupid, but who also have the courage to do something and have more interesting stories. So, if you think that everybody thinks you are a brave person and a bit stubborn or stupid, because you do not listen to your parents that want you to be smart, you can use the brand Diesel to confirm this image that people have of you.
One way people can accomplish this social consistency is by the use of symbolic consumption which helps “to sustain the self and locate them in society” (Wattanasuwan, 2005, p. 179), because “possessions are part of the social communication system” (Richins, 1994, p. 523). This implies that people can use a product that is similar to their social self to maintain the image that they think others have of them. This will give them a positive feeling, since their social consistency is not violated. Therefore, they may also have positive feelings towards the brand of the product used and feel attached to it.
Likewise, Japtura et al. (2014, p. 622) state that “respondents are likely to use brands in order to communicate their social self or in order to be regarded by other people as who they think the respondents are” and that therefore social self-congruence has a positive effect on emotional brand attachment. Therefore, the hypothesis about the main effect of social self-congruence on emotional brand attachment will be as followed:
2.2.4. Ideal social self-congruence
The last self is the ideal social self and belongs to the public self. This self reflects how a person would like others to see him. The motivation that belongs to this self-congruence is social approval, which means that people want to do things that causes others to admire them and to earn approval from others. When people do not do this, it will cause social disapproval. So they want to gain positive reactions from others by acting consistently with their ideal social self (Sirgy et al., 2014). For instance, the brand Louis Vuitton is a luxury brand that has a sign of distinction, excellence and rarity. Next to this, it represents something truly extraordinary, unique heritage and a brand that is demanding for perfection (LVMH, n.d.). If these values are consistent with the image that a person want others to have of him, he can use the brand Louis Vuitton to display this luxury image and gain social approval. This is different from the ideal self, because the ideal self focuses on enhancing the self-image that you have of yourself, and the ideal social self is about enhancing the self-image that others have of you, so others approve you.
Furthermore, Aaker (1999, p. 45) states that “people who are influenced by social roles are in need for self-presentation”. Self-presentation is defined as the control a person has over how others perceive him, including bringing characteristics of the self to the attention of others, for example by using a certain product, to generate optimal effect (Schlenker & Leary, 1982). The person uses for example products (or brands) to get others to see him how he ideally wants, which is supported by Solomon (1983). He states that people use products to benefit from their symbolic consumption which is related to social behavior. This means that people will search for products or brands that are in line with their ideal social self to give others the ideal social image of themselves so that others approve them, which may result in a feeling of connection between the self and the brand.
Next to this, Japtura et al. (2014) indicate that respondents are likely to use brands in order to be regarded by people in a way that they would like other people to consider them as and that therefore ideal social self-congruence has a positive effect on emotional brand attachment. They support this with an example of a statement from one of their respondents who said that the possession or use of a brand can give you an association with the brand that has an aspirational status. When you have this brand it can make you feel like you belong to the club of the brand. This is why the fourth hypothesis is stated as followed:
2.3. Moderating effect: consumption context
It has appeared from previous research that “people act differently in different situations” (Aaker, 1999, p. 45). That is why it is necessary to take moderators into account to control the relationship between self-congruence and emotional brand attachment in different contexts. The moderator that is included in this study is consumption context. As stated before, regarding this construct, a product or brand can be publicly and privately consumed. A publicly consumed brand means that the consumed brand can be seen by others when being used, while privately consumed brands cannot (Bearden & Etzel, 1982). The importance of the consumption context in relation to self-congruence has already been stated in previous studies. This way, Onkvisit and Shaw (1987) show that the congruence between self-concept and the image of a product is higher when the products are publicly consumed and is lower when the products are privately consumed. But there is no distinction made here between the different types of self-congruence. Below, the results of previous studies about the moderating effect of consumption context on the relationships between the different types of self-congruence and different variables are reviewed. Based on this, hypotheses are developed about the moderating effect of consumption context on the relationships between the different types of self-congruence and emotional brand attachment.
2.3.1. Consumption context and actual self-congruence
One of the studies that already made a distinction between the different types of self-congruence is from Graeff (1996). He shows that there is a positive relationship between the actual self-congruence and product evaluations and that this relationship is strengthened when products are consumed both publicly and privately. However, Dolich (in Graeff, 1997) states that the positive influence on this relationship is stronger for privately consumed products than for publicly consumed products. Next to this, Graeff (1997) notices that the public consumption context has a stronger positive effect on the relationship between self-congruence and brand attitude and purchase intention, but that there is no significant difference between the actual and ideal self-congruence. This means that there are different results regarding to the moderating effect of consumption context on self-congruence in relation to different dependent variables. Although, the positive moderating effect of the consumption context corresponds in these studies.
Moreover, it may be important that emotional brand attachment is also linked to these three variables. This way Japtura et al. (2014) show that emotional brand attachment has a positive effect on purchase intention and intention to recommend, which they refer to as positive WOM that exists of evaluations of a brand, and it may have high positive correlations with brand attitude. This may indicate that the moderating effect of consumption context on the relationship between actual self-congruence and emotional brand attachment is present too.
Furthermore, the actual self is part of the private self which means that it is about the images that a person has of himself (Sirgy et al., 2014). This corresponds to the private category of the consumption context, because this situation is about consuming products that cannot be seen by others when being used, so it is all about what the person thinks of himself. Additionally, it is in contrast to the public consumption context where the product can be seen by others when being used (Bearden & Etzel, 1982), because the private self is not about what others think of him. This may indicate that there is a positive moderating effect of the private consumption context and no or a less strong positive effect of the public consumption context. This means that the brand can be more important when it is privately used than when it is publicly used. Consequently, people can be attached more to the brand in the private consumption context, since it helps to comply to and maintain their self-identity, which is important for the actual self-congruence (Sirgy et al., 2014).
Because of these studies and statements, it may be indicated that the effect of consumption context on the relationship between actual self-congruence and emotional brand attachment is positive and even stronger for the privately consumed products than for publicly consumed products. This is why the hypothesis is stated as followed:
H5: The relationship between the actual self-congruence and emotional brand attachment will be stronger in the private consumption context than in the public consumption context.
2.3.2. Consumption context and ideal self-congruence
Like stated before, Graeff (1997) shows that the positive moderating effect of the public consumption context on the relationship between self-congruence and brand attitude and purchase intention is stronger than for the private consumption context. However, he does not find a significant difference between the actual and ideal self-congruence under these circumstances. Meanwhile, during an earlier study, Graeff (1996) indicates that both the private and the public consumption context have a positive effect on the relationship between the ideal self-congruence and product evaluation. Additionally, Dolich (in Graeff, 1997) shows that publicly consumed products have a stronger effect on this relationship than privately consumed products. Graeff (1997) explains that the difference between his results and the results of Dolich (in Graeff, 1997) are due to conspicuousness manipulation of the consumption context in his own study. In this case this means that both “actual and ideal self-image were measured as static cognitive states that are assumed to remain stable over varying contexts and situations, and subjects were evaluating the brand with a specific situation in mind” (Graeff, 1997, p. 66). This is a better way of investigating this relationship than just giving a brand name, because their evaluation will be unrealistic since consumers think of specific consumption situations which would differ between respondents (Graeff, 1997). Again, these studies have contradictory results regarding the moderating effect of the consumption context on self-congruence and several constructs. The
corresponding results they do show is the positive influence of the moderator and that most of the time this effect is stronger for the public consumption context than for the private consumption context.
Because of this and the positive relationship between emotional brand attachment and purchase intention, intention to recommend (which exists of evaluations of a brand), and high correlations with brand attitude as shown by Japtura et al. (2014), it might be stated that both consumption contexts also have a positive moderating effect on the relationship between ideal self-congruence and emotional brand attachment.
Furthermore, the ideal self is just like the actual self part of the private self. So, it is about the images that a person has of himself and not about what others think of him (Sirgy, 2014). This corresponds to the private category of the consumption context, because of the situation that others cannot see the use of the product (Bearden & Etzel, 1982). In contrast to Dolich (in Graeff, 1997) and Graeff (1997), this would indicate that the moderating effect of the private consumption context is stronger than the public consumption context. Therefore, the brand would also be more important in the private consumption context than in the public consumption context. Besides, people would be more attached to a brand when used privately, since it enhances their self-concept, which is important to meet their ideal self-congruence (Sirgy et al., 2014).
Despite the results of previous studies, here the private self is considered as more important. This refers to the statement that for the private self the opinion of others is not important. Therefore, the private consumption context would be more of influence than the public consumption context. This is why the hypothesis is stated as followed:
H6: The relationship between the ideal self-congruence and emotional brand attachment will be stronger in the private consumption context than in the public consumption context.
2.3.3. Consumption context and social self-congruence
Since the relationship between the social self-congruence and consumption context has not been investigated specifically yet, it may be useful to look at the moderating effect of consumption context on the relationship between self-congruence and different variables. As stated before, Graeff (1997) shows that both the public and the private consumption context have a positive moderating effect on the relationship between self-congruence and brand attitude and purchase intention, and that this effect is even stronger for publicly consumed products than for privately consumed products. However, Onkvisit and Shaw (1987, p. 21) indicate that there is a “higher congruence between the self-concept and images of socially consumed products in contrast to a lower consistency of congruence that exists between the self-concept and images of privately consumed products”. Both studies show that the public consumption context has a positive moderating effect. However, the influence of the private consumption context differs.
Besides this, the social self is part of the public self. This means that it is all about how we think or how we would like others to think about us (Sirgy et al., 2014). So here, the images a person thinks others have about him are important, and not the images a person has of himself. This has similarities with the public consumption context, since others are able to see the use of a brand in this situation. On the other hand, it does not correspond with the private context where others are not able to mention the use of a product (Bearden & Etzel, 1982). This may indicate that in this case the public consumption context is more important and has a bigger influence than the private consumption context. Therefore, a brand may be more important when used publicly than privately, because it helps them to maintain and confirm the image that others have of them, which is key for the social self-congruence (Sirgy et al., 2014). Consequently, people can be more attached to the brand. On the other hand, when using the brand in private, there is nobody and so there are no images that others have of you that you have to confirm to. Therefore, it may be less important which brand you use and you can be less attached to it.
Because of these studies and statements, the hypothesis will be stated as followed:
H7: The relationship between the social self-congruence and emotional brand attachment will be stronger in the public consumption context than in the private consumption context.
2.3.4. Consumption context and ideal social self-congruence
Also, there has not been done a lot of investigation on the moderating effect of consumption context in relation to the ideal social self-congruence. Like stated before, Onkvisit and Shaw (1987) show that publicly consumed products have a higher self-congruence, whereas privately consumed products lead to lower consistency of congruence. They explain that the higher self-congruence is possible because of the self-concept that encourages public consumption to, for example, upgrade the consumers’ self-concept. This corresponds with the motivation of ideal social self-congruence, namely to gain social approval by generating optimal effects about the self (Schlenker & Leary, 1982; Sirgy et al., 2014). Additionally, Graeff (1997) did study the situational ideal self-congruence which refers to project an ideal image of the self in a particular situation that is similar to the brand image. The situations used were a meeting with your new boss and new senior executives, and a meeting with two of your friends and their friends who you do not know yet, so both public contexts. So, apparently the opinion of others matter and you want to show an ideal image to get approval from your boss and your friends. Because of this definition and contexts, the situational ideal self-congruence can be seen as equal to the ideal social self-congruence. This experiment shows that the public consumption context has a positive effect on the situational ideal self-congruence in relation to brand attitude and purchase intention. Next to this, symbolic consumption is important for the ideal social self-congruence, because it can generate benefits for a certain self-presentation (Schlenker & Leary, 1982; Solomon, 1983). Solomon (1983, p. 324) indicates that “product symbolism is generated at the societal level but may be consumed at the
individual level”. However, he also states that the symbolic consumption takes place at both the public level because of group pressure and the private level because of making images with yourself which are equal to a form of experience. This means that during this private situation you keep in mind the public situation, because the opinion of other people does matter. For instance, when you are trying on new clothes and looking at it in the mirror, you can think about what others would think of your clothes. So eventually, product symbolism is influenced by both contexts, but more by the public consumption context than by the private consumption context. The results of these studies may indicate that the public consumption context has a positive moderating effect on the ideal social self-congruence. However, the results about the moderating effect of the private consumption context are in conflict.
The ideal social self is, just like the social self, related to the public self. In this case it is also about what we think that others think about us or what we want others to think about us (Sirgy et al., 2014). Likewise, the public consumption context might be more important here, because the images others have of us are more important than the images we have about ourselves and in the public consumption context others are able to see the product use in contrast to the private consumption context (Bearden & Etzel, 1982). Therefore, the public consumption context may have stronger influence than the private consumption context. In this case, a brand can be more important, since it helps to control your self-presentation by symbolic use (Schlenker & Leary, 1982; Solomon, 1983) to make sure others approve you which is the motivation for the ideal social self-congruence (Sirgy et al., 2014). Consequently, people can be more attached to the brand. Meanwhile, there are no images that others have of you in the private consumption context, because there is nobody to notice your product use. Therefore, a brand may be less important, because it does not need to give you an aspirational status and so there might be less attachment to the brand. However, during the private consumption context you may still care about social approval, since you keep in mind the opinion of others. Therefore, the brand can still be important and the private consumption context may lead to attachment.
Because of these studies and statements, the hypothesis is stated as followed:
H8: The relationship between the ideal social self-congruence and emotional brand attachment will be stronger in the public consumption context than in the private consumption context.
2.4. (Moderator) control variables
Malär et al. (2011) show that product involvement, self-esteem, and public consciousness have a moderating effect on the relationship between self-congruence and emotional brand attachment. They show that when the level of these moderators is high, the relationship between the actual self-congruence and emotional brand attachment is strengthened more than when these levels are low. Next to this, ideal self-congruence became significant and so can increase emotional brand attachment when there is a low level of these moderators (Malär et al., 2011). They explain this by looking at the positive and negative
effects of ideal self-congruence on emotional brand attachment. First, the positive effects can be a result of the self-enhancement activities. Second, the negative effects can occur because of the perceived gap between their own reality and their idealistic brand personality of their aspirations. This is linked to the social comparison theory from Festinger (1954) which states that when the compared brand is out of reach this can lead to a need of distancing and decreased emotional brand attachment. This means that when people realize that the aspirational brand personality is impossible to achieve, this may even backfire to emotional brand attachment and weaken it. However, these moderators have not yet been investigated in relation to the social and the ideal social self-congruence. Therefore, it is important to control for these moderators.
Furthermore, emotional brand attachment has several antecedents next to self-congruence. One that is mentioned more often is perceived quality. This way, Grisaffe and Nguyen (2011) show that respondents mentioned superior quality as an important characteristic of marketing as the basis of attachment, because it can provide trust, assurance and confidence. Next to this, Japtura et al. (2014, p. 623) notice that respondents would choose a brand “because they perceived that the brand represents good quality products”. The perceived quality that is mentioned in this study refers to “the consumer’s judgement about a product’s overall excellence or superiority” (Zeithalm, 1988, p. 3). Therefore, this variable will be taken into account as control variable.
These relationships and hypotheses come together in a conceptual framework as presented in figure 1.
Figure 1 Conceptual model. Actual self-congruence Ideal self-congruence Social self-congruence
Ideal social self-congruence Emotional Brand Attachment Consumption context Perceived quality Product involvement, Self-esteem, Public self-consciousness H1+ H2+ H3+ H4+ H5 H6H7H8
3. Methodology
In the paragraph of research design, it is discussed why an experiment was conducted, how the research was conducted, how and to who the questionnaire was distributed, and the ethics and the limitations of this research. Consequently, the research object describes what the research is about and which brands were investigated and why. In paragraph 3.3. the questionnaire design is discussed. This exists of an explanation about which variables were measured with what items. In the final paragraph, it is discussed which statistical analyses were used to conduct the research.
3.1. Research design
The data collection method used for this study was an experiment. This was a suitable method for this research, because the respondents were randomly chosen and randomly selected for a certain consumption context (InfoNu.nl, n.d.). This was to test the causality of the type of consumption context on the relationship between different types of self-congruence and emotional brand attachment. Therefore, there were two versions of the investigation. One questionnaire treated the private consumption context, while the other treated the public consumption context. The experiment was conducted by a questionnaire, because this was an easy way to reach a lot of people.
The questionnaire was distributed online to Dutch people via the internet through the social network of the researcher, like Facebook, LinkedIn, WhatsApp, and email. Gender, level of education, and age did not matter. Except that the respondents had to be above 18 years old, because they had to be familiar and had to have experience with the research object, which was a beer brand. The respondents were given a link to the questionnaire with which they were randomly assigned to one of the two versions. This means that the respondents were independent when they participated to the experiment. This may have increased the validity, since their answers could not have been influenced by an interviewer or observer. Moreover, the respondents were treated ethically. They were remained anonymous and they were able to stop participating to the experiment if they wanted to. Before starting the questionnaire, there was information given about the subject of the research and that people with the age below 18 were not allowed to participate. Next to this, their results stayed confidential and if participants wanted to know more about the research, they could have contacted the researcher by email. The questionnaire lasted about 6 minutes and was distributed in April and May 2018.
A limitation of this research is that the object was consumers and their emotional brand attachment to a beer brand. Respondents had to be familiar and had to have experience with the research object. Therefore, people with the age below 18 were not allowed to participate. Furthermore, because of the focus on a fast moving consumer good, it may be impossible to generalize the results to another sector, like durable consumer goods.
3.2. Research object
The object of this research existed of consumers and their emotional brand attachment to a certain beer brand. For this research this implied the following brands: Heineken, Amstel, Brand, Grolsch, Jupiler, Bavaria, and Hertog Jan. These brands were chosen for this study, because these are the most consumed beer brands in the Netherlands (NRC, 2016). This was to make sure that every participant could choose a brand that he or she knew well enough. Next to this, the choice for beer brands was due to the fact that beer is a product that is often consumed, since it is a fast moving consumer good. Moreover, it is consumed often in private and in public consumption contexts and it focuses a lot on image appeals in its advertising (Graeff, 1997). Because of this focus on image appeals, it is possible to attribute personal characteristics to a brand.
3.3. Questionnaire design
Like stated before, there were two versions of the questionnaire. All questions were the same, the only thing that differed between the questionnaires was the priming text to lead a respondent into a private or public consumption context. The operational table (appendix A, table 1) provides information about the variables and items used from previous studies. Appendix B provides the questionnaire the way it was distributed. Next to this, the items with a ‘*’ were negative questions which means that these items were reversed coded. The items from previous articles have been translated and back by five Dutch natives fluent in English.
Choosing a brand. In the beginning of the questionnaire, the respondent had to choose one of the beer brands. It was indicated that the researcher was interested in a relationship between the respondent and a brand and that this relationship could be positive or negative.
Familiarity, Experience, Consumption frequency. When the respondent had chosen a brand, there were some questions asked about their familiarity and their experience with the brand based on the study of Graeff (1997). The scales were adapted from a 7-point to a 5-point Likert scale so it would be in line with the rest of the scales in the experiment. These questions were to control if the respondents knew their brand well enough. Furthermore, a question about consumption frequency was added to check how often they actually consumed the product.
Consumption context. After this, a respondent was primed into a certain context. The participants had to read a message that told them to imagine that they were in a particular consumption situation (private or public) so it would be clear in which situation they were placed. Half of the participants were randomly assigned to the public consumption context and the other half of the respondents were randomly assigned to the private consumption context. The messages were based on the texts of Graeff (1997). They were adapted to shift the focus more on the public and private situation (so whether there are other people present or not), instead of the place where the beer was consumed.
Additionally, examples were given to strengthen the explanation of the situation. This way, the participants were free to imagine a suitable situation, without being forced too much into a specific situation. Next to this, the messages were adjusted so they looked more alike to eliminate confounding effects because of possible associations with certain texts.
Types of self-congruence. Consequently, they were asked about their self-congruence with the brand. The actual and the ideal self-congruence were questioned the same way Malär et al. (2011) did. Meanwhile, there were no existing questions about the social and ideal social self-congruence. Based on the questions regarding the actual and ideal self-congruence in Malär et al. (2011) and based on the definitions of the social and the ideal social self-congruence created by Sirgy et al. (2014), new questions were developed to collect data about the social and the ideal social self-congruence.
Emotional brand attachment. Subsequently, questions regarding the dependent variable ‘emotional brand attachment’ were asked. These questions were also used during a study of Malär et al. (2011) and were based on measures of Thomson et al. (2005). For these questions people had to think about how their feelings toward their brand were characterized in relation to the items that were linked to emotional brand attachment, like affection, connection, and passion. This was measured with a 5-point Likert-scale.
Product involvement, Self-esteem, Public self-consciousness. When the emotional brand attachment was measured, the moderator control variables were tested. The questions regarding product involvement, self-esteem, and public self-consciousness were used from the study of Malär et al. (2011).
Perceived quality. Next to this, the control variable was tested. The statements about perceived quality were used from the study of Keller and Aaker (1992) and were measured on a 5-point scale.
Gender, Age, Education. In the end, there were some questions about the respondent’s demographics. This way, the characteristics of the respondents became clear.
3.4. Statistical analyses
Frequencies and descriptives were requested to know the characteristics of the respondents. Consequently, chi-square tests and an ANOVA showed if there were differences between the groups of context and their gender, age, and education. Next to this, an ANOVA was used to check whether there were differences between the effects of different beer brands on emotional brand attachment and on the different types of self-congruence. Moreover, the reliability analysis and the factor analysis were used to test the reliability and validity of the items of the variables. Furthermore, data about the familiarity, experience, and consumption frequency are shown by descriptives. After this, a regression analysis was first used to analyze the main effects of the independent variables on emotional brand attachment (H1-4). Secondly, this analysis was also used to analyze the moderating effect of consumption context on the relationships between the independent variables and emotional brand attachment (H5-8). Third, the regression analysis was used to test the influence of the moderating control variables ‘product
involvement’, ‘self-esteem’, and ‘public self-consciousness’ on the relationship between the different types of self-congruence and emotional brand attachment. During these regression analyses the control variable ‘perceived quality’ was always included.
4. Results
In this paragraph the data is analyzed and the results are shown. In the first part a description is given about the respondents and how they were divided over the two consumption contexts and the beer brands. After this, the results about the familiarity and experience with the brand and the consumption frequency are given. In the third part the results of the main effects of the different types of self-congruence on emotional brand attachment are discussed (H1-4). After that, the moderating effect of the consumption context is analyzed (H5-8). And at last, effects of the moderator control variables are discussed.
4.1. Sample description
After conducting the experiment, there were 235 Dutch respondents of which 62 did not finish the questionnaire. These respondents were deleted. Next to this, there were 9 respondents that had ‘few experience with his or her chosen brand’, 1 respondent that ‘did not have any experience with his or her chosen brand’, and 4 respondents that ‘did never consume their chosen brand’ who were also deleted. This was done to make sure the participants could respond the questions with sufficient knowledge about their chosen brand. There was no one who was not familiar with his chosen brand. In the end, there were 159 participants left.
This group existed of 61 men (38,4%) and 98 women (61,6%). They were equally divided over the two contexts (80 private consumption context; 79 public consumption context). The age of the respondents ranged from 18 to 58 (M = 27.79, SD = 10.63). A chi-square test showed that the two consumption context groups did not statistically differ from each other in terms of gender (χ2 (1) = .57, p = .45). Also, an ANOVA showed that the two consumption context groups did not statistically differ from each other in terms of age (F (28,130) = .69, p =.87). Table 1 shows how the respondents were divided over the two consumption context groups.
Table 1 Descriptives of age divided over the two consumption context groups. Consumption
Context
N Mean Std. Dev. Minimum Maximum
Private 80 27.7 10.85 18 57
Public 79 27.9 10.46 18 58
Next to this, a chi-square test showed that the two consumption context groups did not statistically differ from each other in terms of education (χ2 (3) = 1.99, p = .58). Table 2 shows how the respondents were distributed over the different consumption context groups. This also shows that WO is the most common
level of education in this study.
Table 2 Distribution of respondents over the two consumption context groups based on education Consumption Context No education Secondary education MBO HBO/ Bachelor WO Total Public 0 4 4 25 46 79 Private 0 7 7 21 45 80 Total 0 11 11 46 91 159
Moreover, the research object was consumers and their emotional brand attachment to a beer brand. Therefore, respondents had to choose between a couple of beer brands. Table 3 shows how the choices of the respondents were divided over the beer brands. As presented, Heineken is by far the most chosen brand. On the other hand, Brand and Amstel have not been chosen at all or have only been chosen once. An ANOVA showed that there was no statistical difference between the different beer brands in terms of emotional brand attachment (F (5,153) = 1.59, p = .167). This means that the results ragerding emotional brand attachment are independent from a certain beer brand. Therefore, the results will be more generalizable to other beer brands. Next to this, an ANOVA showed that there were no statistical differences between the different beer brands in terms of actual self-congruence (F (5,153) = 2.07, p =.07), ideal self-congruence (F (5,153) = .70, p = .63), social self-congruence (F (5,153) = 1.36, p = .24), and ideal social self-congruence (F (5,153) = .89, p = .49). This means that the results of the different types of self-congruence are independent from a certain beer brand and will be more generalizable to other beer brands.
Table 3 Amount of chosen beer brands.
Beer brand Amount chosen
Heineken 38 Hertog Jan 19 Bavaria 10 Grolsch 9 Jupiler 6 Amstel 1 Brand 0
Most of the respondents were familiar with their beer brand (M = 4.44, SD = .58). Next to this, most of the respondents did have much experience with their chosen beer brand (M = 4.15, SD = .70). The consumption frequency of most of the respondents was several times a month, but less than once a week (M = 2.58, SD = 1.38).
4.2. Validity & reliability
Before the analyses, some items were reversed coded. These are indicated with a ‘*’. To check if the variables were valid and reliable, multiple factor analyses and reliability tests were conducted. First, there was a factor analyses done with the dependent variable ‘emotional brand attachment’, the control variable ‘perceived quality’, and the moderator control variables ‘product involvement’, ‘self-esteem’, and ‘public self-consciousness’. There were correlations above .30, therefore, there was chosen for an oblique rotation and the pattern matrix was used. This matrix showed that PI_5 was a cross-loader and was therefore deleted. After this, the factor analysis was repeated and again an oblique rotation was done. This time, there were no cross-loaders and the pattern matrix was used to separate the items to the components. This pattern matrix is shown in appendix C, table 1. The items EBA_1 to EBA_6 were merged to the variable ‘EBA’ and the items PQ_1 and PQ_2 were merged to the variable ‘PQ’. Next to this, the items PI_1 to PI_4 were merged to the variable ‘PI’ and the items SE_1 to SE_4 were merged to the variable ‘SE’.
However, it is noteworthy that table 1 (appendix C) also indicates that the variable public self-consciousness should exist of two components. The reliability of PSC_1 and PSC_2 together was acceptable (α = .77), but the reliability of PSC_3 and PSC_4 together was bad (α = .36). However, when the public self-consciousness was taken as one component, then the reliability was bad but almost acceptable (α = .56). The separation of this variable could be explained by the fact that PSC_1 and PSC_2 are about how you come across in presenting yourself, while PSC_3 and PSC_4 are focused on someone’s appearance. The interpretation of these questions may also have changed through time because of the influence of social media. People are inclined to show themselves better than the reality which is defined as self-presentation bias (Arkin, Appelman, & Burger, 1980), which is done a lot via social media. The use of social media has increased since the development of the scales of public self-consciousness by Malär et al. (2011) whereby the mean age of the respondents of this research are the people who use social media almost the most (ZIGT, 2018). This could have led to different results. Therefore, the existing items were maintained and merged to variable ‘PSC’.
After this, another factor analysis was conducted to test the validity of the different types of self-congruence. Also in this case there were correlations above .30 and an oblique rotation was done. Consequently, the factor analysis was done again and surprisingly the pattern matrix showed that the first questions about a type of congruence (α = .85), and the second questions about a type of
self-congruence belonged to each other (α = .75) (appendix C, table 2). This can be explained by the statements of the questions, because the questions ACTUAL_1, IDEAL_1, SOCIAL_1, and IDEAL_SOCIAL_1 are almost the same. This also applies for the questions ACTUAL_2*, IDEAL_2*, SOCIAL_2*, and IDEAL_SOCIAL_2*. On the basis of the conceptual model, another factor analysis was conducted whereby the results were forced to split up into 4 components. Again, an oblique rotation was done, because there were correlations above .30 and the factor analysis was repeated. The pattern matrix in table 3 (appendix C) shows that the items ACTUAL_1 and ACTUAL_2*, and SOCIAL_1 and SOCIAL_2* belong to each other, just like they supposed to. On the other hand, IDEAL_1 and IDEAL_SOCIAL_1, and IDEAL_2* and IDEAL_SOCIAL_2* belong to each other. This may again be explained by the formulation of the questions. Moreover, the scales for the actual and the ideal self-congruence were copied from the article of Malär et al. (2011). Next to this, the reliability analysis shows that if the variables are kept as in the conceptual model the reliability analyses are still acceptable for all variables (actual self-congruence: α = .69; ideal self-congruence: α = .64; social self-congruence: α = .66; ideal social self-congruence: α = .62). Therefore, the items were maintained and merged to the following variables: ACTUAL_1 and ACTUAL_2* to ‘Actual’, IDEAL_1 and IDEAL_2* to ‘Ideal’, SOCIAL_1 and SOCIAL_2* to ‘Social’, and IDEAL_SOCIAL_1 and IDEAL_SOCIAL_2* to ‘Ideal_Social’.
After the factor analysis, the reliability of the remaining variables was tested. The rules of thumb are as followed: >.9 = excellent, >.8 = good, >.7 = acceptable, >.6 = questionable, >.5 = poor, <.5 = unacceptable (George & Mallery, 2003, p. 231). The reliability of the public self-consciousness and the different types of self-congruence are already given above. The reliability of emotional brand attachment was good (α = .81). Also, the reliability of perceived quality was acceptable (α = .78). Next to this, the reliability of product involvement after deleting PI_5 was good (α = .84) and the reliability of self-esteem was acceptable (α = .78). Table 4 shows the Cronbach’s Alpha of all the variables.
Table 4 Reliability analysis of all variables.
Variable Cronbach’s Alpha
Actual .69 Ideal .64 Social .66 Ideal_Social .62 EBA .81 PQ .79 PI .84 SE .78 PSC .56
4.3. Testing the hypotheses
First, a multiple regression analysis was used to test if the different types of self-congruence had a main effect on emotional brand attachment. Perceived quality was included as control variable. The results indicated that the model explained 11.3% of the variance and that the model was a significant predictor for emotional brand attachment (F (3,153) = 5.02, p < .001). There was no main effect of the different types of self-congruence on emotional brand attachment. This means that hypotheses 1-4 are rejected. Meanwhile, perceived quality significantly predicted emotional brand attachment (β = .20, p < .05). The coefficients are presented in table 5.
Table 5 Coefficients of the regression analysis for H1-4. Model Unstandardized
Coefficients
Standardized Coefficients
t Sig. Collinearity Statistics
1= Public B Std. Error Beta Tolerance VIF
(Constant) 1.256 .377 3.330 .001*** PQ .216 .084 .201 2.561 .011* .914 1.094 Actual .132 .099 .144 1.337 .183 .485 2.061 Ideal .039 .106 .045 .364 .716 .366 2.732 Social .058 .083 .065 .697 .487 .641 1.559 Ideal_Social .065 .097 .077 .673 .502 .426 2.345 * p < .05.
** p < .01. *** p < .001.
To test the hypotheses 5-8, another multiple regression analysis was done including the moderator. Interaction variables with the consumption context and the different types of self-congruence were developed. This means that there were 10 independent variables involved in this regression analysis. Therefore, 130 respondents were needed (N = 50+8*10) and the analysis could be conducted, because of the 159 participants in this study (SPSS Handboek, n.d.). These variables were first mean centered, because of multicollinearity. This means that the tolerance was smaller than .1 and the VIF above 10, as shown in table 6. These interaction variables were analyzed with the other variables like in table 7. In table 6, the data has been merged to show only the necessary information in a structured way.
Table 6 Multicollinearity of the interaction variables self-congruence and consumption context Collinearity statistics Tolerance VIF Actual_Context_notcentered .021 48.028 Ideal_Context_notcentered .017 59.862 Social_Context_notcentered .029 34.248 Ideal_Social_Context_notcentered .020 49.683
After mean centering the variables, there was no more multicollinearity and the regression analysis was conducted again. The results of this analysis showed that the model explained 16.3% of the variance and that the model was a significant predictor for emotional brand attachment (F (10,148) = 4.08, p < .001). Again, there were neither significant main effects of the different types of self-congruence, nor a significant main effect of consumption context on emotional brand attachment. However, perceived quality significantly predicted emotional brand attachment (β = .19, p < .05). Next to this, the interaction effects of the actual self-congruence with the consumption context (β = -.31, p < .01) and of the ideal self-congruence with the consumption context (β = .25, p < .05) significantly predicted emotional brand attachment. This means that the relationship between the actual self-congruence and emotional brand attachment is stronger in the private consumption context than in the public consumption context and that hypothesis 5 is failed to be rejected. Meanwhile, hypothesis 6 is rejected, because the relationship between the ideal self-congruence and emotional brand attachment is stronger in the public consumption context, instead of in the private consumption context. The interaction effects of the social self-congruence with the consumption context and of the ideal social self-congruence with the consumption context did not significantly predict emotional brand attachment. This means that hypotheses 7 and 8 are rejected. The coefficients of this multiple regression analysis are presented in table 7.
Table 7 Coefficients of the regression analysis for H5-8. Model Unstandardized
Coefficients
Standardized Coefficients
t Sig. Collinearity Statistics
1= Public B Std. Error Beta Tolerance VIF
(Constant) 1.128 .371 3.038 .003** PQ .200 .082 .186 2.424 .017* .902 1.109 Actual .152 .098 .165 1.555 .122 .469 2.131 Ideal .021 .106 .025 .199 .843 .345 2.896 Social .044 .082 .050 .545 .587 .632 1.583 Ideal_Social .105 .100 .124 1.047 .297 .380 2.629 Context .145 .101 .106 1.437 .153 .980 1.021 Actual_Context -.573 .195 -.311 -2.944 .004** .474 2.111 Ideal_Context .423 .212 .247 1.993 .048* .346 2.891 Social_Context .169 .163 .095 1.034 .303 .632 1.582 Ideal_Context -.306 .198 -.180 -1.546 .124 .389 2.570 * p < .05. ** p < .01. *** p < .001. 4.4. Additional analyses
Next to the stated hypotheses, additional analyses were tested. First, product involvement was tested as a moderator. After this, self-esteem, and consequently public self-consciousness were included as moderating variables.
4.4.1. Product involvement
A multiple regression analysis was conducted whereby the product involvement was included as a moderating variable. Next to this variable, only one of the different types of self-congruence was included, because of the limitation of the amount of respondents. With only one of the four types of self-congruence included, there were 6 independent variables in the analysis. Therefore, there were enough respondents (N = 50+8*6 = 98). All the variables had to be mean centered because of multicollinearity (table 8) to make the interaction effect of one type of self-congruence and product involvement. The interaction variables in table 8 were analyzed separately of each other, just like how they were analyzed
as shown in table 9-12. The data has been merged here to only show the necessary information in a structured way. After this the regression was conducted again.
Table 8 Multicollinearity of the interaction variables self-congruence and product involvement. Collinearity statistics Tolerance VIF Actual_PI_notcentered .025 40.582 Ideal_PI_notcentered .024 42.544 Social_PI_notcentered .023 44.085 Ideal_Social_PI_notcentered .025 40.424
First the multiple regression analysis was done with the actual self-congruence. These results showed that 37.7% of the variance was explained by the model and that the model was a significant predictor of emotional brand attachment (F (6,152) = 16.96, p < .001). The coefficients, as presented in table 9, show that perceived quality is no longer a significant predictor of emotional brand attachment. However, the main effect of actual self-congruence has become a significant predictor of emotional brand attachment (β = .17, p < .05). This means that the actual self-congruence has a positive effect on emotional brand attachment. Next to this, the interaction effect of actual self-congruence with consumption context is a significant predictor of emotional brand attachment (β = -.19, p < .01). This indicates that the relationship between the actual self-congruence and emotional brand attachment is stronger in the private consumption context than in the public consumption context. Also, the main effect of product involvement is a significant predictor of emotional brand attachment (β = .50, p < .001). This means that product involvement has a positive effect on emotional brand attachment. On the other hand, the main effect of context and the interaction effect of actual self-congruence and product involvement are no significant predictors of emotional brand attachment.
Table 9 Coefficients including product involvement and actual self-congruence. Model Unstandardized
Coefficients
Standardized Coefficients
t Sig. Collinearity Statistics
1= Public B Std. Error Beta Tolerance VIF
(Constant) .898 .314 2.858 .005** PQ .134 .070 .125 1.926 .056 .941 1.062 Actual .154 .060 .168 2.555 .012* .196 1.091 Context .170 .087 .124 1.962 .052 .993 1.007 Actual_Context -.357 .116 -.194 -3.069 .003** .987 1.013 PI .428 .057 .496 7.507 .000*** .902 1.109 Actual_PI -.001 .075 -.001 -.013 .989 .994 1.006 * p < .05. ** p < .01. *** p < .001.
Consequently, the same multiple regression analysis was conducted, but with the focus on ideal self-congruence instead of the actual self-congruence. The results of the regression analysis showed that 33.3% of the variance is explained by the model and that the model significantly predicted emotional brand attachment (F (6,152) = 14.15, p < .001). This time only the main effect of product involvement significantly predicted emotional brand attachment (β = .51, p < .001). This means that product involvement has a positive effect on emotional brand attachment. The coefficients are shown in table 10.
