Online lifestyle communications:
the effect on brand image, moderated by the
consumer’s psychographics
Rozemarijn de Koomen June 30,2014
Thesis Seminar Business Studies Supervisor: A.C. J. Meulemans MSc Academic year: 2013-2014
Abstract
Online lifestyle communications is a new phenomenon in marketing. It is a marketing tool which communicates a certain lifestyle that is related to a brand through social media, blogs and/or the brand’s website. This research studied whether online lifestyle communications have an effect on brand image, and whether this effect is moderated by the consumer’s psychographics. The findings show that online lifestyle communications have a positive effect on brand image; there even exists a significant positive relationship between the two. This relationship is only moderated by the extent to which someone has the psychographic characteristics of a variety seeker. For variety seekers the effect is stronger than for others. For other psychographic types the effect of online lifestyle communications did not change, there is no moderating effect there. This contribution to the literature is useful for marketing managers, who can now make an educated decision on whether they should use online marketing communications or not.
Contents
Abstract……….………2
Contents………...3
1. Introduction……….……...5
2. Literature review………..7
2.1 Marketing and the internet……….……..……….…7
2.2 The consumer-brand relationship..………..………….9
2.3 Brand image………..…….10
2.4 Psychographics………...12
2.5 Online lifestyle communications………..…13
2.6 The research gap……….14
3. Conceptual framework……….……….16
3.1 Factors of interest………16
3.1.1 Online lifestyle communications………...16
3.1.2 Brand image………..….19
3.1.3 Psychographics………..…19
3.2 Theoretical framework…………...……….………..20
3.3 Boundaries of the framework ..…..………..………20
4. Methodology……….…….….22
4.1 Overall research design……….……….……….……..22
4.2 Research sample………….……….………22
4.3 Procedure……….………...23
4.4 Measures……….………….………...24
4.4.1 Measuring online lifestyle communications…….……….…24
4.4.2 Measuring brand image…….………....25
4.4.3 Measuring psychographics……….………25
4.4.4 Survey construct………..………..…26
4.5 Data analyses………..………..…27
4.5.1 Preparing data and checking reliability……….………..…….…27
4.5.2 Analyzing the relationship between OLC and brand image…….…..….…28
4.5.3 Analyzing the moderator………..……….…….…….28
5. Results……….………..…..30 5.1 Sample characteristics……….……….…30 5.2 Reliability……….…………..31 5.3 Independent-samples t-test……….……….33 5.4 Simple regression……….35 5.5 Multiple regression……….36
5.5.1 Testing for moderation of the leaders rating……….…….36
5.5.2 Testing for moderation of the variety seekers rating………….………37
5.5.3 Testing for moderation of the recognition seekers rating……….….39
6. Discussion………..……….41
6.1 Interpretation of the results……….41
6.1.1 The relationship between OLC and brand image……….41
6.1.2 Moderation of psychographics………..….……….42
6.1.3 Unexpected findings……….42
6.2 Contribution to theory and practice……….………..43
6.3 Limitations……….43
6.4 Suggestions for future research……….44
7. Conclusion………..………...46
8. Bibliography………..47
Appendix………..……….51
A: Online lifestyle communications examples………..…51
B: Scale constructs brand dimensions…….………...58
C: Scale constructs psychographic types……….……….….59
D: Final survey construct……….……..………..60
1. Introduction
Ever since it became possible to interact via the internet, blogs, review sites and social media platforms have grown to become a part of our everyday lives. Everyone can be an author and everyone can have an audience. In the Netherlands the average time a day spent on social media is 41 minutes (Kemp, 2014). In the USA this average is 37 minutes (Adler, 2014). With these high averages, social media beats other major internet activities, making it the biggest internet activity nowadays.
People eagerly follow each other on social media. We post pictures and status updates about what we do all day, how we feel, what we like, what we have accomplished. We communicate our own lifestyle to the rest of the world, and create an image of ourselves on the internet. Friends and family interact with that image, social life has now partially moved to the internet. An interesting development is that some brands have developed their image on the internet in the same way individuals have. They post video’s, photos and status updates, they blog and they interact with (potential) customers. In this way they communicate their ‘brand lifestyle’, which might be appealing to their target market.
According to Holt (1997) lifestyle is about the symbolic expressions of collectivities that map cultural content of important social categories. Literature about online lifestyle communications does not exist yet. Could this be a proper marketing strategy? Could it, just like personal sharing and blogging, develop an image of the brand? A lot of research has been done about brand identity or personality and its effects for marketing purposes. Brand identity might have a similar effect, but still lifestyle is different in the fact that it does not represent one sort of person, but a way of living, which can be appealing to a larger variety of people.
In social media not everyone follows the same people, or likes the same updates. This depends on who you are, what you like, what your personality is and what your values are. It is safe to assume that the same counts for online lifestyle communications: not everyone will be fond of it, it depends on what type of person you are. So if online marketing communications is a proper strategy, would it work for every single consumer?
With online lifestyle communications being such a new phenomenon, it is only normal that there is a gap in the literature concerning the topic itself and the effects it could have for a brand. The aim of this research is to find out what online lifestyle communications are, what its effect is on the way people perceive the brand (brand image), and how this effect can differ for different (psychographic) types of people. The question that will be answered is: what is the relationship between online lifestyle communications and brand image, and how is this moderated by the consumer’s psychographics?
2. Literature Review
In this section relevant literature will be explored. Various topics will be described broadly to find a research gap and define the research question. The first topic to be explained is
marketing and how this discipline is influenced by the internet. The second part will be about the concept of the consumer-brand relationship. Thirdly a segmentation technique called psychographics is discussed. After that the e-marketing tool online lifestyle
communications will be explained. Lastly a summary of the current state of knowledge is presented, together with a narrow research gap that will be addressed by this research.
2.1 Marketing and the internet
Marketing has changed a lot over the last two decades. The main reason for this is the development of the internet. We all use it daily, but how did the internet become so
integrated in our lives? Initially the World Wide Web was created as a platform to exchange information between users (Kaplan & Haenlein, 2010). Back then it was mostly companies and government institutions publishing information. However, when the World Wide Web went through some technological changes, like gaining new functions such as Adobe Flash and RSS, it became possible for users to continuously change and add content and
applications in a participatory and collaborative way. This new type of internet is called Web 2.0 (Kaplan & Haenlein, 2010). It was now very easy for an ordinary user to share content on the internet, or to respond to someone else’s content. According to Kaplan & Haenlein (2010) Web 2.0 is the platform for the evolution of social media. The authors define social media as a group of Internet-based applications that build on the ideological and
technological foundations of Web 2.0, and that allow the creation and exchange of ‘User Generated Content’.
This development of the internet has brought a lot of opportunities for companies. Web 2.0 and social media can be used as a tool for many marketing purposes. This newest area in marketing is called e-marketing (or internet marketing). E-marketing is ‘achieving marketing objectives through use of electronic communications technology’ (Chaffey, Ellis-Chadwick, Mayer, & Johnston, 2009, p. 9). According to Kotler and Keller (2012, p. 36), thanks to the internet there are six new general opportunities for marketing, to be explained in the next paragraph.
The first one is to use the internet as a channel for information and sales. The brand 7
can use the internet to answer questions and just to present their promotions and offerings. Secondly, marketers can use the internet to collect better information about markets, customers and even competitors, for instance by implementing questionnaires through the internet, or to organize online focus groups through Twitter. The third opportunity is to use social media to amplify the brand message, so to post brand content or get into
conversations, representing the brand’s ideals. Fourthly, marketers can use the internet to facilitate and speed communication among customers through brand advocates and user communities. Some popular strategies are co-creation and viral marketing. The fifth opportunity is to send promotional content to people who requested it or gave permission to send it. This is basically old-fashioned door-to-door advertising which has now moved to the internet. The last opportunity is to perform location based marketing through mobile devices. It is now possible to see where someone is and make internet ads appear in his browser of places near to this consumer (Kotler & Keller, 2012, p. 37).
It is necessary to elaborate on the strategies co-creation and viral marketing, since these kinds of strategies were not possible on this scale before the internet. Co-creation of value together with consumers is interesting, especially now whole brand communities have come to existence on the internet (Cova & White, 2010). Having internet users participate in creation of value creates higher brand loyalty and possibly also word-of-mouth about the brand (Thackeray, Neiger, Hanson, & McKenzie, 2008). Viral marketing, also called ‘buzz marketing’, is a very cheap and efficient way of getting a marketing message across to many potential customers (Thackeray et al., 2008). In just a couple of days your content could spread all over the world, providing you with a tremendous amount of attention for your brand.
All of these new opportunities have something in common. They are interactive. Web 2.0 makes it possible to get people involved with a brand, by generating brand content and then let the internet user generate a response in their turn. This can be challenging for marketers but it can also be a very useful tool.
The largest part of brand published content is meant to gain interest from costumers, for them to eventually buy the brand’s product. This is called marketing communication. According to Kotler and Keller (2012, p. 498), ‘marketing communications are the means by which firms attempt to inform, persuade and remind consumers – directly or indirectly – about the products and brands they sell’. The authors compare marketing communications
with a voice of a company, they are the way to establish a dialogue and to build a relationship with the consumer. The model in figure 2.1 shows the macromodel of the communications process (Kotler & Keller, 2012, p. 502). The major parties in this model are the sender (which would be the brand) and the receiver (the consumer). The sender wants to get the message across, through a certain kind of medium. Depending on the type of consumer that the brand is targeting, it chooses its medium, and the possibilities for the target audience to respond (Kotler & Keller, 2012, p. 502). As this process is repeated, the brand and the consumer get to know each other, developing a consumer-brand relationship.
Figure 2.1: The Macromodel of the Communication Process (Kotler & Keller, 2012, p. 502)
Through Web 2.0 this process has speeded up a lot, since it has become easier and less of a hassle for individual consumers to respond, and the response can be received and processed a lot more efficiently than before (Kotler & Keller, 2012, p. 37).
2.2 The consumer-brand relationship
Like mentioned above, Web 2.0 can help develop the consumer-brand relationship. It is easy for the consumer to find information about the brand, and to provide feedback. For the brand this is an opportunity to listen even more closely to its (potential) customers. An example of an e-marketing concept that develops the brand relationship is brand communities. Research shows that when brand communities are cultivated, community members identify themselves with the community and are committed to it, which leads to brand identification and attachment (Zhou, Zhang, Su, & Zhou, 2012).
According to Blackston (2013), a consumer-brand relationship is the interaction 9
between the consumer’s attitude towards the brand and the brand’s ‘attitudes’ towards the consumer. The author says the development of a successful consumer-brand relationship depends crucially on the consumer’s perceptions of the brands attitudes, since these create meaning out of the brand’s messages. So actually a consumer-brand relationship is much like an interpersonal relationship. This is also because people tend to personify a brand
(Kaltcheva & Weitz, 1999). For the brand it is all about what is going on in the consumer’s mind. The difference between the one-dimensional brand image and a real brand
relationship is that in a relationship, the brand also has an attitude to the consumer. So the important question here is: ‘What do the consumers think that the brands think of them?’ (Blackston, 2013). It is important that the consumer likes the brand, but also feels liked by the brand. By interacting with the consumer, and helping him when necessary, a good brand relationship will come to existence. However, the brand relationship can also be influenced by the consumer’s perception of the brand, which can be very subjective. Research shows that when a person’s self-image and the image of a brand are much alike, this has a positive effect on the quality of the brand relationship (Kressmann et al., 2006). In the next section the concept of brand image will be discussed in more detail.
2.3 Brand image
Brand image, or brand perception, is actually a part of consumer-brand relationship. Like mentioned above, this is a one-dimensional relationship, only about what the consumer thinks of the brand, but it can influence the consumer brand relationship a lot (Kressmann et al., 2006). According to Dobni and Zinkhan (1990), there is no consensus on the definition of brand image. The authors say it is largely a subjective phenomenon, based on the
consumer’s interpretations. There are many definitions of brand image, but a very clear one is: ‘the associations that are activated in memory when people think about a particular brand’ (Shimp & Andrews, 2013, p. 139). The consumer’s perception of the brand can be formed by many factors. It is not necessary for the consumer to have an experience with a product from the brand in order to have a brand image. Rather, brand image is molded by the brand’s marketing activities, context variables and the characteristics of the consumer (Dobni & Zinkhan, 1990). Unlike the consumer brand relationship, the consumer can develop an image of the brand, without any interaction between the two.
There are many elements in brand image. According to Shimp and Andrews (2013, p. 10
139), brand image consists out of two factors: types of brand associations, and the
uniqueness, strength and favorability of these associations. In their framework on customer-based brand equity they present the parts of brand image. This framework is shown in figure 2.2. As presented in the figure, types of associations can be divided into attributes, benefits and an overall evaluation of the brand. Attributes and benefits are made up out of several elements as well.
Figure 2.2: Brand Image in the Customer-Based Brand Equity Framework (Shimp & Andrews, 2013, p. 139).
Most marketing activities are intended to improve or uphold the image of the brand, and thereby maybe the brand relationship. Brand image, like consumer-brand relationship can be improved through the internet. However, for forming a brand image, not that much interaction is necessary. Posting information about your brand can be enough. The
consumers only have to see it and can already form their judgment of the brand. Whether this judgment is either positive or negative will partially depend on the characteristics of the consumer. As mentioned above, when the person’s self-image and the brand image are much alike, this can be good for the consumer brand relationship. So the consumer’s characteristics might play a big role for brand image and the brand relationship. There will be elaborated on these characteristics in the following section.
2.4 Psychographics
Market segmentation is a significant part of marketing. Kotler and Keller (2012, p. 236) define a marketing segment as: ‘a group of customers who share a similar set of needs and wants’. It is up to the marketer to distinguish these segments and to decide which segments to target. There are different ways to divide the market into segments. One can segment markets on the basis of geographic characteristics, demographic characteristics, or
psychographic characteristics (Kotler & Keller, 2012, p. 236). With the internet of today it is possible to target your ads accurately at specific audiences (Chaffey et al., 2009, p. 341), which makes correct market segmenting even more important.
Since psychographic characteristics are the most advanced segmenting factors, further explanation of this concept is necessary. Psychographics is actually a combination of demographics and psychology. Unlike using demographic variables, when using
psychographics, one assumes that people within the same demographic groups can have very different psychological profiles (Kotler & Keller, 2012, p. 248). According to Kotler and Keller (2012), in psychographic segmentation, ‘buyers are divided into different groups on the basis of psychological traits, lifestyle, or values’. A popular method to measure
psychographics is through the VALS framework (Kotler & Keller, 2012, p. 248). VALS stands for Values, Attitudes and Lifestyles. This framework classifies adults into eight groups, based on those three factors. A set of 39 questions determines which group someone belongs to. As shown in figure 2.3, the two dimensions of the framework are consumer motivation (horizontal) and consumer resources (vertical). There are three horizontal factors: ideals, achievement and self-expression. One of these factors is the consumer’s main motivation. People motivated by ideals are guided by knowledge and principles. Achievement-motivated consumers want to demonstrate their success to peers. Consumers motivated by
self-expression want social or physical activity, risk and variety (Kotler & Keller, 2012, p. 249).
Figure 2.3 The VALS Framework (Kotler & Keller, 2012, p. 248)
Since psychographics is a combination between demographics and psychology, which together determine a large part of how people behave and act, lifestyle is a concept that is very closely related to this. Therefore psychographics would also be an important factor for the marketing tool ‘online lifestyle communications’.
2.5 Online lifestyle communications
In the field of marketing communication, lifestyle is a concept that is used very frequently. The reason behind this is that lifestyle influences both consumption patterns and the way that consumers process different forms of marketing communication (Vyncke, 2002).
A new e-marketing tool is online lifestyle communications. This is done in blogging, on social media, but also on companies’ websites. Because this marketing tool is
implemented online, mostly through social media, it is a cheap way of marketing. Content about daily activities, food, clothing, nice locations and entertainment is shared or
published. Basically this content represents a way of living, which is not necessarily centered around the product the company is trying to sell. It actually sells the life around it. The idea behind it is that when people find that lifestyle appealing, they will buy products that go with it, which are offered by the brand. Because it is such a new phenomenon, academic literature about this concept does not exist. However, lifestyle as a separate concept is known in the literature. “Lifestyle” is a cloudy term, which can have many interpretations.
Lifestyle would be defined by Holt (1997) as: ‘the symbolic expressions of collectivities that, through relational differences with other collectivities, map cultural content of important social categories in a particular context’. Another way of looking at it, according to Coreil Levin and Jaco (1985), is that the concept of lifestyle serves as a conceptual bridge between cultural and behavioral patterns, and through that also as a link between idealist and materialist perspectives. A simpler definition is ‘the patterns in which people live and spend their time and money’ (Vyncke, 2002). The last one is more to the point, and thus easier to use for this research. However, it would be good to keep in mind that lifestyle is also about the link between idealist and materialist perspectives. This is because in the end the goal of a marketing tool like online lifestyle communications is for people to make that connection, and to move from the desire for a certain lifestyle to the desire for a certain product.
2.6 The research gap
Taking all of this knowledge into consideration, it is clear that the variables online lifestyle communications, psychographics, the consumer-brand relationship and brand image are interconnected in some way. However, a direct relationship between any of these variables is not found in the literature so far. Part of the reason for this is that online lifestyle
communications is such a new phenomenon.
In summary, the current knowledge concerning these four variables is the following. The internet of today brings many great opportunities for marketing communications, and can be used as a tool to improve the brand image and consumer-brand relationship. Brand image is the one-way relationship part of the consumer with the brand. Psychographics are important for marketing communications to divide the market in segmentations and target specific groups. Lastly, we do not know much about online lifestyle communications, but we do know that it is a new e-marketing tool of the interactive internet, which is already used by many brands.
Given this information, the next step is to find out if this new phenomenon could be used by marketing managers to improve the consumer’s brand image. For managers knowledge on this topic would be valuable, since it will help them make decisions about which marketing tools to use. It is important to know whether online lifestyle
communications has an effect on brand image, whether it is worth using. This brings us to 14
the question: what is the relationship between a brand’s online lifestyle communications and
the consumer’s brand image? Although the answer to this question would already be a
contribution to theory, it would be even better when it was known for which market segments the tool would be the most effective. This way marketers could determine if the segments for which the tool is proven effective also apply for their target market, and based on that decide to use it or not. A second research question is appropriate: is the relationship
between online lifestyle communications and the brand image moderated by the consumer’s psychographics? Once both of the questions are answered, theory on this topic will be
improved with a simple but useful contribution.
3. Conceptual Framework
Now the overall picture of the current state of knowledge is clear, this knowledge can be used to draft a conceptual framework. Due to the scarce theory about online lifestyle communications, it is not possible to state sufficiently grounded hypotheses. Therefore, exploratory expectations will be set. First all factors involved will all be clearly stated. Next, there will be an explanation on the role of each variable in the framework. In subsection 3.2 the framework will be presented and explained, and in the last subsection the boundaries of the framework will be discussed.
3.1 Factors of interest
As discussed in the literature review, the research questions involve three variables: online lifestyle communications (to be abbreviated as OLC), brand image and psychographics. To get complete clarity about the factors being researched, the definitions that will be used are presented below.
Online lifestyle communications: content shared/published online by a brand, concerning the
patterns in which people live and spend their time and money.
Brand image: largely a subjective and perceptual phenomenon which is formed by the
associations that one has with the particular brand.
Psychographics: Personal characteristics concerning attitudes, personality traits, lifestyle and
values.
Now the definitions of these factors are clear, it is necessary to elaborate on each of them, to develop an expectation of their relation to the other factors.
3.1.1 Online lifestyle communications
There is hardly any academic literature to be found about this concept. It is very new, and although it is used by many brands, there is no scientific support for this marketing tool to actually be effective. However, it is possible to study a few examples of brands using this tool.
Nike for instance, is very obviously communicating a certain lifestyle. As you can see in the appendix (page 51), it regularly posts sports related content. The brand communicates
a simple, sporty and close to nature lifestyle, and inspires people to improve their sports performances. The content they post are generally beautiful pictures of scenic environments with someone working out playing a sports match. Most of the times, Nike products are not clearly visible in the picture. The text that goes with the picture is always something
inspirational and performance-oriented. Nike sometimes also posts separate inspirational quotes to motivate their followers.
Another example is Modemusthaves.com, a Dutch online fashion store. On their website and on their social media accounts they post content about their clothes and the life that goes with it daily. In these pictures (see appendix, page 53), their products clearly play a role, but they are mixed with other elements, in a cool setting, like a girl drinking champagne in the sun. Their website is fully adapted to the stylish young women that visit it. For
instance, it has a ‘boyfriend’ section with men’s clothes. The idea behind it is that the men will not shop at Modemusthaves.com, but their girlfriends will do it for them. Also it has a ‘blog’ section, with someone from the company always telling about the newest trends or inspiring sets of clothes or accessories. The ambience is always very happy and stylish. Just like Nike, they sometimes post quotes, mostly about fashion. Even though it is mostly their own products they use in their content, the pictures breathe one certain kind of lifestyle, which is young, girly, sociable, stylish, fashionable and fun.
Lastly Red Bull is also a brand that does OLC. The lifestyle they communicate is a very extreme one (see appendix, page 55). Mostly they communicate about extreme sports. They never show the energy drink, which is actually the product that they are trying to sell. Their posts are mostly pictures and videos, with a comment about the picture or a question to their audience. A quote that is frequently dropped is ‘go big or go home’, which is also exactly the kind of lifestyle they are communicating. Of course there are only very few people who can actually live this way, but still it is for a lot of people (especially young men) a very attractive idea to become more like that.
What all of these brands have in common, is that they communicate content about a certain way of living, which is always inspiring and/or entertaining. People do not only go through the brand’s posts because they need one of their products, but also to just because it is fun and they can imagine themselves being more like that. Probably consumers are not even going to buy anything most of the times they visit the brand’s page, but once they need something in that product category, they will think of this brand first, because they associate 17
those products with a lifestyle they feel drawn to.
How can all of this OLC be related to psychographics and brand image? Like
mentioned before, there is no actual proof that there is a connection between this specific factor and the other two. However, these brands are successful at what they do. Nike, Modemusthaves.com and Red Bull respectively have 18 million, 102 thousand, and 43 million likes on their Facebook pages. Another measure to look at the success of their content sharing is the Klout Score. This is a public score which ‘measures a person’s (or organization’s) overall online influence on a scale of 1 to 100, with 100 being the most influential. Klout’s system analyzes multiple social networks based on your engagement’ (“Klout,” 2013). Klout analyzes your social networks, and based on how often people like, share or respond to your content, you get a score. The average Klout Score is 40. Users with a score of 63 or higher are in the top five percent of online influencers (“Klout,” 2013). The scores of Nike, Modemusthaves.com and Red Bull are respectively 79, 57 and 91.
Modemusthaves.com seems to have a pretty low score compared to Nike and Red Bull, but one should take into account that this company started only 3 years ago, and is only active in The Netherlands. The average Klout Score of the top ten politicians in the Netherlands is 59 (“G20 Influencers,” 2012), so knowing that, 57 is quite high. Of course these statistics do not prove that it is because of the brands’ OLC that they are so influential, and being influential might not necessarily have to do with a good brand image. However, it does point out that the brands are doing something right, and their OLC plays at least a small part in that.
Given the information above, it would be safe to say that for these brands OLC has a positive effect on their marketing performance. Building on from that statement, it is naturally suspected that OLC can have an effect on the brand image. Like discussed in the literature review, brand image is a consumer’s perception and interpretation of a brand. The consumer’s perceptions of the brands attitudes create meaning out of the brand’s messages (Blackston, 2013), so when a brand implements OLC, and thereby inspires potential
consumers who visit their page, it is expected that the brand image of these people will improve. Looking at the literature about lifestyle and the developments in practice, it is
expected that OLC has a positive effect on brand image. Assuming this effect exists and will
be proved, it is also expected that there is a positive relationship between OLC and brand
image.
3.1.2 Brand image
The brand image can be compared to a one-way relationship from the consumer to the brand. It depends on what the consumer thinks of the brand, and how he perceives the brands attitudes towards him. The brand’s goal is to sell its products. So what the brand needs to do to improve the brand image, is to make sure the consumer views the brand as positive. Once the consumer feels ‘liked’ by the brand, through seeing and reading
marketing communications that apply to him, it is expected that the brand image improves. This also applies to OLC. As discussed in section 3.1.1, it is expected for OLC to have a positive effect on the brand image.
Lifestyle communications can represent a kind of attitude towards the consumer. Nike, for instance, motivates consumers to live a more active lifestyle and to push yourself to achieve more. The way consumers perceive this message and ‘attitude’ towards them, can influence their brand image. Does the consumer feel the same way about an active lifestyle; this will probably make him feel more attached to the brand. Does the consumer hate the idea of getting fit; he probably will not feel as connected to Nike.
Whether someone likes sports or not, depends on many variables, such as
personality, attitudes, values and lifestyle. These variables together are psychographics. So the way consumers perceive OLC, can be influenced by their psychographics. This suggests that the effect of OLC on brand image can be influenced by the consumer’s psychographics.
Expected is that psychographics are a moderator of the relationship between OLC and brand image.
3.1.3 Psychographics
Psychographics are a combination of demographics and psychology. Demographic
characteristics are facts like age, family size, gender, income, occupation, education, religion, generation, nationality and social class (Kotler & Keller, 2012, p. 238). Psychological
characteristics are things like personality traits, attitudes, values and beliefs. With psychographics being a combination of the two, each of these examples also belong to psychographics. An extra variable, which does not belong to demographics or psychology, is lifestyle. Lifestyle is a result of demographics and psychology together. Demographics like income, occupation, nationality and social class, together with psychological factors like values, personality and attitudes result in the kind of lifestyle someone chooses to have.
With lifestyle being one of the variables of psychographics, it is likely that the effect of OLC on brand image is influenced by psychographics. For instance, Modemusthaves.com communicates a young, girly and fashionable lifestyle. Some girls do not believe in fashion and are not willing to spend money on that sort of thing. They have a different lifestyle, and their brand relationship will probably not improve after seeing Modemusthave.com’s OLC. It might even get worse after seeing this kind of content. So the consumer’s psychographics will probably not only determine the strength of the effect of OLC on brand image, but also the direction.
To construct further on the expectations aforementioned, psychographics is expected
to have a moderating effect, either positive or negative on the relationship between OLC and brand image.
3.2 Theoretical Framework
Now all of the factors involved are clear, and their expected relation to each other has been explained, it is possible to draft a theoretical framework. In figure 3.1 the framework is presented. As one can see there is a direct positive relationship between OLC and brand image. From above there is an effect from psychographics on the relationship between OLC and brand image. Like suggested in section 3.1.3, it is expected to depend on the
psychographics if this effect will be enhancing the positive relation, or decreasing it.
Figure 3.1: The Theoretical Framework
3.3 Boundaries of the framework
The theoretical framework presented above is quite general, which makes it broadly Online Lifestyle
Communications Brand Image
Psychographics
applicable. There are not many boundaries to it.
Geographically speaking it is useable anywhere in the world. The only requirement is for the consumers and the brand to have access to the internet. However, it is possible that in countries where the culture is completely different from the western culture (China for example) the psychographics are so different that the model is not as relevant, because they might not have the same kind of VALS groups as Europe and North America. In that case the research could be repeated for their psychographic groups.
There is no temporal boundary to this framework. Of course it could not have been used before Web 2.0, but this framework is not relevant for the past. As far as we know now, the internet continues to be a significant part of our lives in the future, and as long as people use the internet, this framework is applicable.
Just like geographical validity, the framework is theoretically valid for all sorts of organizations, but probably more suitable for brands that sell hedonic products, which are more luxurious, have a purpose for pleasure and represent status. Of course it would also be possible to build a lifestyle around paperclips, but that probably would not have the same attraction to it as fashionable clothing or a nice car.
4. Methodology
In this section the research design will be discussed. The methodology used to give an answer to the research questions, to test the theoretical framework and analyze the data will be explained. First the overall research design will be discussed, and the sections after that will elaborate on the research sample, the procedure, the measures of the variables and how the data will be analyzed.
4.1 Overall research design
The research design will be a survey, using questionnaires. The main advantage of this approach is that it allows the researcher to collect standardized data from a sizeable population in an economical way, which is easy for comparison (Saunders, Lewis, &
Thornhill, 2012, p. 178). The sample for a standardized questionnaire can be a lot bigger and more diverse than interviews or case studies for example. This makes the collected data more generalizable, which would be very useful since the research is meant for marketing practices. The questionnaire is also appropriate for this research because in a questionnaire one can ask questions to get an image of the exact knowledge that is desired, so as a
researcher you have a lot of control over the data you are going to collect (Saunders et al., 2012, p. 177). Also, the standardized data can be analyzed quantitatively, using descriptive and inferential statistics (Saunders et al., 2012, p. 178), which would allow for a relationship between variables to be discovered.
A limitation of using a questionnaire is that the data that is collected is not as wide-ranging as other research strategies (Saunders et al., 2012, p. 178). This is because in a questionnaire one cannot ask too many and neither too complicated questions, since the respondent might not finish the questionnaire or not start it at all. For this research this is a disadvantage, because OLC is such a new phenomenon, and it would be nice to ask some more exploratory questions on the topic. This will not be possible because once you start asking open-ended questions, the response rate will be a lot lower due to the loose of goodwill of the respondents (Saunders et al., 2012, p. 421).
4.2 Research sample
The population for this research is all consumers in the Dutch market. The sample has to 22
capture all groups of the Dutch markets in terms of social background, age, beliefs and lifestyles. Obviously, there is not enough time or resources to get a large enough sample size that will actually be representable for the whole Dutch population of 16.9 million people (“CBS,” 2014). At a population size of more than 10,000,000 and a margin error of 5%, actually a sample size of 384 is needed (Saunders et al., 2012, p. 266) . This will not be feasible for this research, but it is possible to analyze the sample as a normal distribution, as long as the item sample sizes are at least 30 (Saunders et al., 2012, p. 54). In order to find out what the relation is between OLC and brand image, there will have to be two groups (a control group and a treatment group), each filling out a different questionnaire (see section 4.3). So 60 should be the minimum amount for the sample, theoretically speaking. However, more is always better for the reliability (Saunders et al., 2012, p. 266). The sample should be around 100 people. This should provide a solid basis for the research, giving a good reliability of the outcomes. Also it gives some room for respondents who fill out their survey
incorrectly. The sample needs to be representative for the whole Dutch market, so it would be best if all ages and demographic backgrounds would be equally covered. Unfortunately, this is partially out of control of the researcher, because the procedure to reach the sample will be internet mediated.
4.3 Procedure
The questionnaires will be web-based, which has the advantage of reaching a large number of people and being geographically dispersed (Saunders et al., 2012, p. 421). Also, it will take a lot less time to process the obtained data in the computer, because this can all be done automatically. On the other hand, a disadvantage might be that the response rate would be low, since it is not a very personal way of asking people to cooperate, which makes it easier for them to ignore the request. If the response rate is too low, another option would be to also physically hand out the questionnaires in public places. Asking people in person to fill out a questionnaire makes it harder for them to say no. However, a disadvantage of this is that the answers would later have to be processed into the computer manually. The time period in which the sample is taken does not really matter for this research. Looking at the research planning, ideally the sample will be taken in May, over a period of three weeks.
To get a good, large sample, the questionnaires will be spread through the internet in various ways. Firstly, Facebook and Twitter will be used to share the link. Friends and family
members will be asked to share the link as well. Furthermore blackboard and e-mail will be used to spread the questionnaire. By doing all this, the sample will be diverse on
demographic factors. The only limitation here is that the people who will share the questionnaire for me, will be related to me in some way, and will probably have some psychographic of demographic factors in common with me, just like their friends have with them. So it would be likely that on some areas the sample will not be as varied as preferred. However, this is the only way to get a large sample without using a reward system, so it will have to suffice.
4.4 Measures
In this research there are three variables that have to be measured. The first one is OLC, the second one is brand image, and the third one is psychographics. This subsection is divided into four parts, the first three discussing the measurement of one variable each, and the last one will present the construct of the questionnaire.
4.4.1 Measuring online lifestyle communications
Like explained before, OLC is a marketing tool. Through sharing lifestyle related content, the brand hopes to gain the consumer’s interest for that lifestyle, and eventually sell their products to the consumer to be part of that lifestyle. It is not possible to accurately measure the extent to which a brand does OLC. The brand either does it or it does not do it.
It is the same for simulating OLC in a questionnaire. It will either be shown or it will not be shown. A dummy variable of 0 and 1 will be used to represent if there is OLC or not. One group of respondents will not get to see the OLC, and the other group will. OLC will be simulated by showing images of a brand’s OLC. Enough images should be shown so the respondent gets a clear idea of these communications.
To cancel out the effect of people’s preferences for a lifestyle, two opposite brands will be used, which both represent completely opposite lifestyles. This way, the preferred lifestyle of the respondents will not have an influence on their overall rating for brand image. The two brands that will be used are Nike and Jumbo. Nike represents a very active, healthy and outdoorsy lifestyle and Jumbo represents the social, fun, food-centered and relaxed lifestyle. When the questionnaire represents two opposite lifestyles and the focus is not on
which lifestyle the respondent prefers, but on the question whether the images they see or do not see have an effect on their perception of the brand.
4.4.2 Measuring Brand Image
Brand image is a very general concept with many facets. As was shown in figure 2.2, brand image was made up out of different types of brand associations and the uniqueness, strength and favorability of these associations. All of these associations together form a perception of the brand, so all elements in these associations have to be measured. Lassar, Mittal and Sharma (1995) came up with five dimensions which together measure brand equity, but can also be used for analyzing the different dimensions of the consumer’s perceptions on a brand. These dimensions are performance (what is the quality of the brand?), social image (does this brand fit my personal image?), value (do I get value for my money?), trustworthiness (is this brand to be trusted?) and attachment (do I have positive feelings towards this brand?). To measure these dimensions, a survey is used (Lassar et al., 1995). For each dimension there are a couple of statements, and the respondent has to rate these statements on a Likert scale from 1 to 5, 1 being strongly disagree, 5 being strongly agree.
For the questionnaire in this research twelve questions will be used to measure brand image. Only twelve questions are used and not seventeen like in the original
questionnaire, since respondents will have to fill these out twice: once for Jumbo and once for Nike. If the questionnaire becomes too long, the response rate will drop (Saunders et al., 2012, p. 424), so the questions have to be kept to a minimum. A limitation of this is that the information that is obtained is less detailed. In the appendix (page 58) the questions are presented per dimension. Most questions are straight out of Lassar’s (1995) questionnaire, but others are adapted to measure the dimension in a more general way, since not all questions from the original questionnaires can be asked.
4.4.3 Measuring Psychographics
Psychographics is a very complicated variable as well. To measure these, the VALS questions will be used. As explained in the literature review, these are a commonly used set of
questions to measure psychographics. In the psychographics questionnaire, a scale from 1 to 4 will be used, like in the original VALS questionnaire. This is useful because when there is no 25
box for ‘neutral’, people are forced to make a choice between disagree and agree. When testing for moderation this could help, since the respondents psychographics will be more specific this way. The VALS questionnaire consists of 39 statements, but that is too many for this research, so the questions are reduced to only sixteen.
A limitation of the VALS questions is that the company that owns the questions, Strategic Business Insights, does not reveal the way they rate the answers. In the original VALS method, after the respondent completed the survey, the respondent is categorized in one of the 8 VALS types, as discussed in section 2.4. However, it is unknown how exactly they determine this, so the statements used for this research are selected on the basis of common knowledge. These sixteen questions are expected to represent all the 39 questions. It is probably impossible to rebuild 8 groups like in VALS, but the questions selected are based on four more general types: leaders, variety seekers, recognition seekers and narrow thinkers. Innovators are the types that like to lead, that are knowledgeable, and that like personal challenges. Self-expressionists are the types that like variety, like to express themselves through the things they buy, and highly value an exciting social life. The
recognition seekers are the types that like to belong to a group, that seek recognition from their peers, and who buy things for status. Lastly, the narrow thinkers are people who do not like change, who like stability, who are conservative in their way of thinking and who have a narrow focus in their life. In the questionnaire there are a few questions for each type. This way the respondent gets a rating for each type of consumer. The set of statements to measure each type can be found in the appendix (page 59).
After asking the sixteen VALS questions, also a couple of general questions will be asked concerning the respondents’ age, income and sex. These questions are important for getting a general idea of the research sample, and to check whether the sample is diverse enough.
4.4.4 Survey Construct
Now all measures have been discussed, there will be explained how they are captured in the survey. The final version of the survey can be found in the appendix (page 60). The survey tool Qualtrics is used, because this tool provides a lot of options for developing questions, the final questionnaire is easy to use for respondents, and after the answers are processed they can be organized and exported to SPSS.
The questionnaire consists of three parts. The first part measures the variable
psychographics. The sixteen questions will be asked and after that the three questions about age, sex and income. Next is the Jumbo part. The whole sample has to fill out the questions on brand image for Jumbo in this part. However, before they have to do this, half of the respondents has to look at images of Jumbo’s OLC. These are six images of OLC of Jumbo, mostly coming from their Facebook page. The last part is about Nike, and just like the Jumbo part, only half of the respondents gets to see OLC from Nike. Every respondent will have to fill out the twelve brand image questions, whether they have seen the OLC or not.
Before the survey starts there will be a short introduction, thanking the respondent for their help and explaining briefly what my research is about. During the survey the respondent can see what percentage of the survey he has already completed. This prevents people from stopping after they started, because when they see how fast they move
through the questionnaire, they will be reassured it is actually really short, and have less tendency to quit when they get bored. Once they finished all the questions their answers will be processed and they will be thanked for their cooperation.
4.5 Data analyses
The aim of this research is to find out what the relationship is between OLC and the consumer’s brand image, and if this relationship is moderated by the consumer’s
psychographics. In this section will be discussed what analyses will be used to find out the answer to each research question.
4.5.1 Preparing data and checking reliability
Before starting the analyses to answer the research questions, it is necessary to make sure that the data is reliable. All questionnaires that were not fully completed need to be
removed from the data. Also, the internal consistency reliability of the scales measuring the brand image dimensions and the psychographic types need to be checked. Cronbach’s Alpha (α) can be used for this (Pruppers, 2014). For each construct item Cronbach’s Alpha needs to be at least above 0.6 (Pruppers, 2014). If this is the case, the item constructs are reliable and are together a measure of the same dimension. If the Alpha is below 0.6, the construct needs to be changed, or a question needs to be removed from the construct.
4.5.2 Analyzing the relationship between OLC and brand image
First we need to find out what sort of relationship there is between OLC and brand image. Is it a positive one (like expected), a negative one, or is there no relationship? To answer this question the means of the brand image of the control group and the treatment group need to be compared. The control group did not get to see the OLC of the brands, and the treatment group did. If there is a significant difference between these two group means, it would mean there is a relationship between OLC and brand image.
The t-test can be used to test whether two group means are different (Field, 2009, p. 325). Since the respondents who did not see the OLC are not the same as the respondents who did see the OLC, the samples are independent. According to Field (2009, p. 325), when different participants are assigned to each condition, the independent-samples t-test should be used. In this case, the grouping variable would be if the OLC images were shown (dummy of 1) or not (dummy of 0). The total value of the brand image of Jumbo and Nike together will be analyzed. In the results will be checked whether one mean is different from the other, and whether this is significant, using a 95% confidence interval. It will also be interesting to see what effect OLC has per dimension of brand image, so the same analysis will be done for the means of brand image dimensions of performance, social image, value, trustworthiness and attachment.
To see if OLC could be an explaining model for brand image, a simple regression can be done. According to Field (2009, p. 198), ‘simple regression analysis is a way of predicting an outcome variable from a predictor’, so we can use it to see to what extent OLC explains brand image. The regression analysis will come up with the line that best describes the data. The coefficient B or ‘slope’ of this line will tell how much the dependent variable (brand image) is changed by a change of one in the independent variable (OLC), and the significance value can tell us whether this relationship is significant (< 0.05).
4.5.3 Analyzing the moderator
For the second research question the aim is to find out whether psychographics are a moderator on the relationship between OLC and brand image. To do this, we first need for every respondent their score for each psychographic type. This will be calculated by taking the average score from the questions that measure a type. After this, each type needs to be analyzed for moderation.
To test for moderation, a multiple regression will be done. According to Field (2009, p. 210) this is a good strategy to measure predictor variables for which there is a theoretical reason for them to predict the outcome. In this case the predictor variables for the model would be the OLC, the rating in for the psychographic type and the interaction term, which is the product of the first two. The interaction term is the one that tells us whether
moderation has occurred (Field, 2013, p. 736). When this term makes a significant contribution, moderation is going on.
To avoid multicollinearity, which is a correlation between any of the predictors, all independent variables need to be standardized (“UvA,” 2014). The regression model will be done with the standardized variables. The dummy for OLC will be standardized, the rating for the psychographic type will be standardized, and the interaction term will be the product of those two standardized variables.
Once the regression is done, the results in the SPSS table will tell if the interaction term is significant or not, using a 95% confidence interval (p-value should be smaller than or equal to 0.05).
5. Results
In the previous section the research design and the methodology were explained. The data has been collected and the tests described in the methodology were implemented. This section will present the results of those tests. First a description of the sample and the data in general will be provided. Next the results of the tests for reliability will be explained. After that the three main tests, which are the t-test, the simple regression and the multiple regressions will be discussed in three separate sections.
5.1 Sample characteristics
The total amount of people who started the questionnaire was 135. However, the drop-out rate was 26%, so only 100 people really completed it. From those 100 people, 11 responses had to be deleted, because they left too many questions unanswered. So in the end the sample was narrowed down to 89. The minimum for the sample was 60, so 89 was good enough for a valid and reliable sample which could be normally distributed.
Of this sample, 50.6% was male, and 47.2% was female. Two respondents have not answered this question, so they are the missing 2.2%. The sex ratio of the population of the Netherlands is 49.5% male to 50.5% female (“CBS,” 2013a), which is quite close to the sample. The average age was 32.21. This is much lower than the national average age, which is 40.8 (“CBS,” 2013a). This is probably due to the large portion of respondents between 20 and 25 years old. People of this age are the largest part of my network, so this was also a large part of the respondents. This is was a limitation of using my own network to distribute the sample.
The yearly incomes of the respondents was for the largest part of the respondents lower than the population average, which is 29,400 (“CBS,” 2013b). 43% of the sample earned less than 10,000 euros per year. However, 18 % was above 100,000 per year. The rest was scattered, but because of the highest 18%, the average income category still was 30,000 to 40,000. Furthermore the highest education completed was for the largest part (47.2 %) bachelor, which is only 22% for the population in the Netherlands (“CBS,” 2013c), so on this area the sample scored a lot higher than the Dutch population. This too is probably due to the use of my network for distributing the surveys.
5.2 Reliability
To be able to implement a reliable research, the reliability of the research constructs needed to be checked. In this research five dimensions for brand image were used, and four
psychographic types. The sets of questions used for each dimension/psychographic type, were analyzed with the Cronbach’s Alpha. In the tables 5.1 and 5.2 the Cronbach’s Alpha for each construct is presented.
Table 5.1: Cronbach’s Alpha for brand image dimensions
Dimension Number of items Cronbach’s Alpha (α) Performance 2 0.780 Social Image 3 0.748 Value 3 0.742 Trustworthiness 2 0.755 Attachment 2 0.653
Table 5.2: Cronbach’s Alpha for psychographic types
Type Number of items Cronbach’s Alpha (α) Leaders 4 0.322 Variety Seekers 5 0.412 Recognition Seekers 4 0.521 Narrow Thinkers 3 0.436
As one can see, all the alphas for the brand dimensions are higher than 0.6, so they are reliable. However, in the second table, none of the alphas are above 0.6. This was a problem, so the constructs of the scales needed to be changed. This resulted in only 3 groups. The narrow thinker type had to be dropped since none of the items used to measure this type were reliable to use as a construct together. A reason for this could be that the statements for the narrow thinker type tried to touch upon too many facets. A statement like ‘I like my life pretty much the same from week to week’ and ‘A woman’s life is fulfilled when she can
provide a happy home to her family’, are too different from each other to measure one type of character. The questions for narrow thinkers were now useless, so they were deleted. Also, for the other dimensions, some questions were changed and some were removed. This lead to a sufficient score of Cronbach’s Alpha for each type. Below the questions for each type are presented.
Leader:
1. I have more ability than most people. 2. I must admit that I like to show off. 3. I like to lead others.
Variety Seeker:
1. I like the challenge of doing something I have never done before. 2. I like a lot of excitement in my life.
3. I like doing things that are new and different.
Recognition Seeker:
1. I follow the latest trends and fashions. 2. I like being part of a group.
3. I dress more fashionably than most people.
The theories behind the types are quite self-explanatory. The leader is self-confident, likes to lead, likes to be the center of attention and is not shy to show his skills to other people. The variety seeker likes to do new things, and does not like too much stability and consistency in his life. The recognition seeker wants to belong to a group, and wants to be recognized as a good member. Therefore he is keen to look fashionably, because he needs others to think he looks good and preferably better than other people, so he can be recognized as a favorite member of the group. The table below shows the Cronbach’s Alphas for the new
psychographic types.
Table 5.3: Cronbach’s Alpha for the new psychographic types Type Number of items Cronbach’s Alpha (α) Leaders 3 0.721 Variety Seekers 3 0.751 Recognition Seekers 3 0.663
All scale constructs are now reliable and can be used for the analyses. The next sections will discuss the results of each analysis.
5.3 Independent-samples t-test
One of the expectations discussed in the conceptual framework was that OLC has a positive effect on brand image. To test this, the means for brand image of the control group and of the treatment group were compared. This was done through an independent-sample t-test, with a significance level of 5%. First the t-test for the total brand image will be presented, and after that also the results of the t-test for each brand image dimension.
Table 5.4: Independent-samples t-test for total brand image
OLC N Mean Std. Deviation Mean Difference Sig. (2-tailed)
0 45 2.9905 0.37616 -0.25238 0.002
1 43 3.2429 0.371444
The results show that the difference between the means of the group who did not see the OLC (0) and the group that did see the OLC (1) is 0.25. The treatment group rated the brands 0.25 higher on average, which is a significant difference, looking at the p-value. The
presented p-value is two tailed, so for a right-sided test (the expectation is that the mean is higher for the treatment group), the p-value is 0.001. This is smaller than 0.05, so there is a significant difference between the two means. This supports the expectation that OLC has a positive effect on brand image.
In addition, the effect of OLC was on each separate brand image dimension was tested. In the next table the mean differences and the p-values will be presented per brand dimension.
Table 5.5: Independent-samples t-test for each brand dimension
Brand Dimension Mean Difference Sig. (2-tailed)
Performance -0.10685 0.286
Social Image -0.35357 0.002
Value -0.11055 0.233
Trustworthiness -0.27687 0.041
Attachment -0.43204 0.001
Looking at the p-values, the only two brand dimensions from which the means were not significantly different were performance (0.286/2 > 0.05) and value (0.233/2 > 0.05). Social image, trustworthiness and attachment all have a significantly higher mean when the respondents saw the OLC. So for these brand dimensions the expectation that OLC has a positive effect is also supported. However, for performance and value there is no support that this is the case.
Lastly, it is also interesting to compare the means of the total brand image for each separate brand. Although it is not part of the research question to find out whether there is a difference in the effect of OLC on brand image for the two brands, this will provide an insight which might be interesting for marketing managers. While running the analyses, this difference between the brands was coincidentally discovered. In the table below the mean differences for the total brand image and the p-values will be presented separately for each brand.
Table 5.6: Independent-samples t-test of brand image for each brand separately
Brand Mean Difference Sig. (2-tailed)
Jumbo -0.32602 0.004
Nike -0.17873 0.070
These results show that for Jumbo the difference in brand image was much bigger than for Nike. Still, the difference in means is significant for Nike as well (0.070/2 < 0.05), but not as easy as Jumbo, which has a p-value of only 0.002 for a one-sided test. This could mean that the effect of OLC is not as strong for Nike as for Jumbo. There will be elaborated on this in the discussion section.
Looking at all the results, it is clear that for brand image in general OLC does have a positive effect. However, this effect is not as strong on all brand image dimensions, and also between the two brands there was a difference in the effect. The discussion section will go further in discussing these results.
5.4 Simple regression
It has been discovered that OLC has a positive effect on the consumer’s brand image. The next analysis was to test whether there is actually a significant relation between OLC and brand image, and for what percentage OLC explains the variance in brand image. To do this, a simple regression was used. The dependent variable was the score for brand image, the independent variable was the OLC. OLC was expected to predict brand image. In table 5.7 the results for the regression are presented, showing the most important parts of the output from the regression.
Table 5.7: Simple regression model with OLC as a predictor for brand image
Model Summary ANOVA Coefficients
R R Square Std. Error Estimate Mean Square F Sig. (Const ant) B Std. Error Sig. 0.330 0.109 0.37368 1.438 10.299 0.002 2.984 0.259 0.081 0.002 Looking at the model summary, R has a value of 0.330. Since there is only one predictor for
this model, this is simply the correlation between OLC and brand image. R square is more important, since this tells us what percentage of the variation in brand image is explained by the variable OLC. In this case that is 10.9 %, so 89.1% is explained by other factors than OLC. The F-ratio of 10.299 has an associated significance value of 0.002, which is smaller than 0.05, so there is support that OLC has a good degree of prediction for the brand image. However, ANOVA does not provide the individual contribution of each variable in a model. For this you have to look at the B under Coefficients. B is 0.259, and has just like in ANOVA a significance level of 0.002 ( p < 0.05). This means that whenever OLC increases with 1, the rating of the brand image increases with 0.259. This shows that there is a significant
relationship between OLC and brand image: brand image can be predicted (for 10.9%) by the constant of 2.984 added to the product of OLC and 0.259.
5.5 Multiple regression
The two previous sections both discussed analyses done to find the answer to the first research question. This section will show the results of the analyses done answer the second research question: is the relation between OLC and brand image moderated by
psychographics? For each psychographic type a multiple regression was done to see whether the variable of the psychographic type had a moderating effect on the relationship between OLC and brand image. Each of the following three subsections explains the results of the test for moderation of one psychographic type.
5.5.1 Testing for moderation of the leaders rating
Each respondent got a score for the leader type, which was the average of the three questions measuring this psychographic type. Like explained in the method, a multiple regression was done with the OLC, leader rating and the product of these two, the
interaction term (‘moderator’ in the table). The results will be presented in the next tables. Table 5.7a: Multiple regression testing for moderation from the leader psychographics
Model Summary ANOVA
R R Square Std. Error
Estimate Mean Square F Sig.
0.346 0.12 0.88 3.511 3.807 0.013
Table 5.7b: Multiple regression testing for moderation from the leader psychographics
Coefficients
Variable B Std. Error Sig.
OLC 0.339 0.104 0.002
Leader 0.119 0.104 0.256
Moderator 0.059 0.105 0.578
The first table shows that this model explains 12% of the variance in brand image, and the ANOVA shows that with an F-ratio of 3.807, the model is a significant (0.013 < 0.05)
predictor for brand image. However, looking at the coefficients, this is apparently the result of the variable OLC in the model, which we already discovered to be a significant predictor. When looking at the individual contributions, the variable of the leader rating and the moderator are both not significant predictors of brand image. When the moderator is not a significant contributor, there is no support for moderation from this psychographic rating, so
being a leader type does not have a moderating effect on the relationship between OLC and brand image.
5.5.2 Testing for moderation of the variety seekers rating
The variety seekers rating was also tested for moderation. The exact same analysis was implemented, only now instead of the variable for the leader type, the variable for the variety type was used. Tables 5.8 a and b show the results of this multiple regression.
Table 5.8a: Multiple regression testing for moderation from the variety psychographics
Model Summary ANOVA
R R Square Std. Error
Estimate Mean Square F Sig.
0.383 0.147 0.116 4.309 4.821 0.004
Table 5.8a: Multiple regression testing for moderation from the variety psychographics
Coefficients
Variable B Std. Error Sig.
OLC 0.331 0.103 0.002
Variety 0.032 0.102 0.752
Moderator 0.207 0.103 0.047
According to the results, 14.7% of brand image is explained by this model. Looking at the ANOVA, the model is a significantly good prediction of brand image. Naturally there is a significant relation again between OLC and brand image, but this time, also the interaction term makes a significant contribution to the model. The significance value is 0.047, which is smaller than 0.05. The coefficient of the relation is 0.207. The effect of this moderation is shown in the two graphs on the next page. The first one represents the results for brand image depending on to what extent people are a variety seeker type, when they have not seen any OLC, and the second graph show the same, only when the people have seen OLC.
