The effects of brand communication inconsistency on brand image and brand attitude

June 24, 2013

The effects of brand communication inconsistency

on brand image and brand attitude

By

Max Pool

Qualification: Master Thesis

University of Groningen (RUG) Dept. of Marketing

Supervisor: Dr. K.J. Alsem Second supervisor: Dr. J. Liu

Preface

As a marketing student, I have been interested in the topic of branding ever since I first familiarized myself with the influence of brands, both in practice as well as in theory during lectures. The content of the Brand and Product Management course (during the first semester of this year) raised my interest to an even higher level. As soon as I had to apply for a topic for my master thesis I immediately marked the “Inconsistent branding” topic as my number one choice.

At the start of the thesis (writing) process I thought of it as a challenge. I am very grateful that my supervisor Dr. K.J. Alsem provided guidance during the entire process. At the beginning, Dr. K.J. Alsem helped me to find a specific area of interest within the broad topic of branding. His continuous feedback, along with the support of my fellow thesis-group members, really contributed to the final version of my thesis. Every time there was some uncertainty, my supervisor or any of the group members could provide clarification. I am really satisfied with the entire thesis (writing) process and the way it has been organized. Therefore, I would like to thank Dr. K.J. Alsem and all of my fellow thesis-group members.

Table of content

2.1 Brand image and brand attitude 9

2.2 Brand consistency 10

2.3 Brand consistency and brand image 11 2.4 Brand consistency and brand attitude 13

2.5 Brand involvement 15

2.6 Brand involvement and brand image 16 2.7 Brand involvement and brand attitude 16

2.8 Perceived risk 17 2.9 Conceptual model 18 3. Research design 19 3.1 Experimental design 19 3.2 Stimuli 20 3.3 Participants 22

3.4 Variables and scaling 22

3.5 Experimental procedure 24

3.6 Statistical analyses 25

4. Results 25

4.1 Pre-test/Manipulation check 26

4.2 Descriptive statistics and reliability analysis 27

4.3 Brand image 28

4.4 Brand attitude 29

4.5 Brand involvement and brand image 30 4.6 Brand involvement & perceived risk and brand attitude 31

4.7 Additional regression analyses 32

Summary

Consistency is argued to be the number one rule for creating and sustaining a strong brand. However, this consistency guideline has not been scientifically backed up with empirical findings before. This study investigates the effects of brand communication inconsistency by means of an experiment. Respondents are exposed to either consistent- or inconsistent brand communication, i.e. the stimuli, before their brand image and brand attitude, i.e. the dependent variables, is measured quantitatively.

First of all, the analyses indicate that brand communication inconsistency, in terms of not communicating similar core brand values, has a negative influence on the strength of brand associations in the minds of consumers. These findings are in alignment with the line of reasoning of other authors on the topic. Inconsistencies in communication activities lead to re-evaluation of the brand and ultimately reduce brand image (Buchanan et al., 1999). Moreover, in the case of brand communication inconsistency, high involvement consumers are more likely to re-evaluate their image of a brand in terms of reduced strength of brand associations, than low involvement consumers. This moderator effect is the contrary of what was initially expected. High involved consumers, i.e. those with a high level of personal relevance, appear to be more likely to look at the advertisements/stimuli longer and more carefully (Zaichowsky, 1985). By taking a closer look at the advertisements they will engage in more elaborate thinking (Lee and Schumann, 2004; Meyrs-Levy et al., 1994), which in turn increases the extend to which a consumers re-evaluates the brand (Buchanan et al., 1999).

Results of the analyses indicate that brand communication inconsistency has no significant influence on the brand attitude of consumers. However, results confirm the expectation that high involvement consumers will develop a less negative brand attitude in the case of inconsistency than in the case of consistency. Furthermore, the level of perceived risk appears to have no moderator influence on brand attitude.

1. Introduction

1.1 Background problem

A common used statement, both in marketing courses and in practice, is that a marketing campaign should be “consistent across all activities, channels and over time” in order to build and maintain brand equity. According to some marketers, brand consistency is even argued to be the number one rule of branding (Attitudedesign, 2013). The current amount of (non-scientific) articles published on the internet could be seen as an indication of the perceived importance of the brand consistency guideline in practice. A reason for this popularity is that the concept of consistency, and the positive influence on the effectiveness of a marketing campaign, is straightforward. Aligning all communication activities in such a way that they are consistent, in terms of communicating the same core brand values, does not seem like a complex or difficult task for marketing managers. Consistency has therefore become a ‘safe’ word to use. However, there appears to be confusion in the definition of brand consistency. For instance, a brand and its marketing program can be consistent by means of communicating similar core brand values over time and/or across media. But another more strict interpretation of consistent branding could be to make sure that every single aspect (e.g. colour, font and style) of communication activities are identical over time and/or across media. This study focuses specifically on brand communication consistency, and adopts the following definition: a brand is consistent in its communication if it succeeds

in communicating similar core brand values (i.e. core brand meaning) across its communication activities (Keller, 1998).

Brand consistency has been argued to, partly, determine the ability to build and sustain a strong brand image (Keller, 1993; Park, Jaworski and MacInnis, 1986). Moreover, consistent brand communication is likely to lead to recognition, trust and loyalty from customers (Aaker, 1991). According to Buchanan et al. (1991), besides a brand its positioning being consistent over time, each part of the marketing mix should reinforce this positioning by being consistent as well. The product characteristics, the advertising message, price points and the distribution channels should all be aligned. The idea behind this is the point that inconsistencies in communication activities may lead to re-evaluation of the brand and ultimately reduce brand image (and brand equity). Brand image in light of this study is defined as: the total set of

Even though the positive effects of brand consistency on brand equity have been argued by several authors, it is still a topic that is in need of additional research. The amount of research that has been done on this topic is lacking heavily compared to the way the concept is being applied in practice. At this moment it could be argued that many managerial decisions, which involve the concept of brand consistency, are not sufficiently backed up by scientific research. Additional research would contribute significantly to the clarification of the brand consistency concept and its effects on brand equity. This in turn will allow marketers to better justify their managerial decisions concerning branding in practice. Therefore, the main goal of this research is to enhance the current state of knowledge on the concept of brand consistency. This will be done by investigating what the exact effects of brand communication inconsistency are on brand equity, i.e. to what extend brand inconsistencies actually dilute brand equity (Buchanan et al., 1999).

1.2 Problem statement

As mentioned before, authors of previous research argued and concluded that brand consistency has a positive effect on brand image (Keller, 1993; Park, Jaworski and MacInnis, 1986), but this statement might also be oversimplified. There are most likely many more factors and conditions that determine the actual effects of brand consistency (i.e. moderating effects), which have not or barely been researched so far. Besides this, counterarguments for the positive effects of consistency are existent as well. An example of such an argument is that brand consistency can lead to communication that is consistent just for the sake of being consistent. In other words, the design and implementation of highly consistent communication activities might be limited in terms of creativity, and therefore limited in their attractive power for (potential) customers. Advertisements that are extremely consistent could even become ‘boring’ from a consumer point of view, are less interesting for (high involvement) consumers (Charters, 2009) and are therefore less likely to generate positive attitudes towards the brand and its communications. Thus, the lack of attractive power of communication activities might lead to a negative relationship with the effectiveness of a marketing campaign in terms of failing to generate positive attitudes. Brand attitude in light of this study is defined as: the

degree of positivity or negativity with which a brand is evaluated (Park et al., 2010).

Moreover, it could also be argued that inconsistency in communication activities could have positive, instead of merely negative, outcomes as well. This is in alignment with Charters (2009) his study on brand consistency. His main conclusion was that “with most goods there

is a clear desire for identical attributes, and the consumer will search for consistency from consumption event to consumption event. Indeed, with many goods (such as airplanes or cars) consistent quality is not just desirable but essential for the security of the consumer. However, there is a range of goods where both the nature of production and the expectations of the consumer mean that precise consistency in the product attributes is not only impossible, but also undesirable (Charters, 2009)”. Furthermore, Charters (2009) argues that precise

consistency in product attributes might be undesirable for; creative, aesthetic or ludic brands and high involvement consumers. Building variation and inconsistency into products (brands) can raise the interest and increase the pleasure for these consumers, which would logically result in a more positive brand attitude. “Consistency may actually be an impediment – its

continual “sameness” gets in the way of the consumer’s ability to reshape the brand for their own symbolic or experiential purposes (Charters, 2009)”. Charters his conclusions

concerning brand (in)consistency are fundamental for this study, and will be discussed in more detail in the next chapter.

After addressing the need for additional research on this topic, and explaining the complexity of the brand consistency concept in terms of a possible two-sided effect on brand image and brand attitude, the problem statement can be formulated as:

What is the actual effect of brand communication inconsistency on brand image and brand attitude?

activities (e.g. advertisements and commercials) are still sensitive to the interpretation of consumers. Thus, actual consumer interpretations might differ significantly from the interpretation that would be desired by the organization. Because of that, this research will focus on investigating the effects of perceived brand communication consistency.

Brand equity is too broad of a concept to include in this research as a whole. Instead of including both consumer knowledge effects of brand equity, i.e. brand awareness and brand image (Keller, 1998), this study focuses on the latter. Furthermore, this research will focus on brand attitude, mainly because a key goal of marketing is to create and sustain positive attitudes towards a brand. In most conditions, consumers will rely on these previously formed attitudes in forming judgements about a brand (Buchanan et al., 1999). Marketing researchers argued that both brand image and brand attitude are direct (or indirect) drivers of brand equity (e.g. Aaker, 1991; Faircloth et al., 2001). Therefore, this study focuses on the effects of (perceived) brand communication inconsistency on brand image and brand attitude. The main research questions can be formulated as:

1. What is the effect of perceived brand communication inconsistency on brand image?

2. What is the effect of perceived brand communication inconsistency on brand attitude?

3. Does the level of consumer brand involvement have a moderating influence, on the effects of brand communication inconsistency on brand image and brand attitude?

A second possible moderator is the level of perceived risk in purchasing (and using) a brand its products. Mitchell and Greatorex (1989) found that the extra value offered by a strong brand may be a reduction in the potential risk arising from purchase. Moreover, consistent quality is not just desirable but essential for the security of the consumer (Charters, 2009). Thus, it appears that consistency can reduce the risk of purchasing, by giving the consumer a sense of security. Vice versa, inconsistent brand communication would fail to create this sense of security for consumers, which logically results in a (more) negative brand attitude. This leads to the final research question:

4. Does the level of perceived risk have a moderating influence, on the effects of brand communication inconsistency on brand attitude?

However, more findings and arguments based on previous academic research for these possible moderating influences of consumer brand involvement and the level of perceived risk will be discussed in more depth later. This research will now continue with an extensive review of the existing literature on the topic of brand consistency, which includes relevant findings and arguments of other authors. These insights will contribute to the formulation of researchable hypotheses. Analyzing these hypotheses are ought to provide answers to the proposed research questions. After that, the research design will be discussed in detail, and the results shall be presented after the data has been analyzed. Finally, this research will end with a discussion, recommendations for future research and managerial implications.

2. Literature review

2.1 Brand image and brand attitude

A brand can be defined as “a name, term, sign, symbol or design, or a combination of these,

intended to identify the goods or services of one seller or group of sellers and to differentiate them from those of competitors (American Marketing Association, 1960)”. This definition has

been used by several authors of well-known marketing literature (e.g. Kotler et al., 1994). However, critics argued that this definition is too much focused on an organizational- or marketing perspective. According to critics, it is more important what is in the mind of consumers for a brand. This has lead to several other definitions, such as; a brand is “a

reflection of a consumer’s core values (Sheth et al, 1991)”, or “a personality (Aaker, 1991)”.

In the rest of this research, the focus lies on the consumer perspective. As mentioned before, brand attitude can be defined as the overall evaluation of the brand (Faircloth et al., 2001) or

the degree of positivity or negativity with which a brand is evaluated (Park et al., 2010).

According to Kapferer (1992), the primary capital of organizations is their brands. The value of a brand relates to the next concept, i.e. brand equity. “Customer-based brand equity

(CBBE) is defined as the differential effect of brand knowledge on consumer response to the marketing of the brand. A brand is said to have positive (negative) CBBE when consumers react more (less) favourably to an element of the marketing mix for the brand than they do to the same marketing mix element when it is attributed to a fictitiously named or unnamed version of the product or service (Keller, 1993)”.

1998). Based on these definitions, it is most likely that brand inconsistency has the largest influence on the strength of brand associations. Therefore, this study focuses specifically on the strength of brand image. This leads to the line of reasoning that: the strength of a brand its

image is determined by the strength of the brand associations in the minds of consumers.

2.2 Brand consistency

The concept of brand consistency has been shortly introduced and discussed in the problem statement. As mentioned before, there are a number of contradictory findings on the effects of brand consistency and brand inconsistency. These and other findings will now be discussed more thoroughly. Reviewing the existing literature and findings of other authors on the topic of brand consistency will contribute to the formulation of researchable hypotheses, which are ought to provide answers to the research questions.

In his book, Keller (1998) introduces the theme of complementarity and consistency as an important aspect of building brand equity in any way. On a tactical level, brand equity can be built through the choice of brand elements, the marketing program and through the leverage of secondary associations. Brand elements include the brand name, logo, symbol, characters and packaging. The marketing program includes the product (attributes), price level, distribution channels and communication activities. Leveraging secondary associations means that the brand is linked to other entities, such as a country (of origin), an event or another company. Keller argues that a high degree of consistency among brand elements and the marketing program is necessary to create the desired level of awareness and type of image (i.e. strong associations) that leads to brand equity. “Consistency ensures that diverse brand

and marketing mix elements share a common core meaning (Keller, 1998)”. This line of

reasoning forms the basis for the definition of brand communication consistency that is adopted in this study, i.e.: a brand is consistent in its communication if it succeeds in

communicating similar core brand values (i.e. core brand meaning) across its communication activities. The common core meaning of communication activities should create a set of

However, referring back to the complementarity and consistency issue of Keller, he argues that complementarity enables different brand elements and marketing activities to compensate for the shortcomings of others, which in turn contributes to brand equity as well. Besides this consistency and complementarity issue (or perhaps trade-off), Keller also describes the importance of balancing continuity and change over time. A brand should evolve in the right direction, and in order to maintain its competitive position a brand might need to revitalize (e.g. changes within the DNA of the brand). This seems logical, since the needs and wishes of consumers are likely to evolve and change over time as well. To conclude, Keller addresses the possible two-sided effect of brand (in)consistency as mentioned earlier in the problem statement of this study. Keller argues the importance of brand consistency and continuity on the one hand, and the importance of complementarity and change on the other hand. Next, findings of other authors, arguing in favour of- and/or against brand consistency, will be discussed.

2.3 Brand consistency and brand image

The main findings and arguments in favour of brand consistency have been shortly discussed in the introduction part, but they will be reviewed again in order to formulate researchable hypotheses. As mentioned earlier, brand consistency has been argued to, partly, determine the ability to build and sustain a strong brand image (Keller, 1993; Park, Jaworski and MacInnis, 1986). Moreover, consistent brand communication is likely to lead to recognition, trust and loyalty from customers (Aaker, 1991). These findings are supported by de Chernatony (2001) and de Chernatony et al. (1995), who also found that brand image heavily relies on consistent communication. Furthermore, authenticity becomes increasingly important as a core brand value for many product categories (Brown et al., 2003), e.g. luxury products, which can also be related to brand consistency in terms of communicating common core brand values over time and across media.

Arguments in favour of standardized advertising in general were identified by additional studies, and include the following:

• Maintaining a consistent brand image • Ease of implementation

• Ability to control execution • Reduced costs

A final argument in favour of standardization is that consumers interact among each other, and it may be helpful if they have the same perception (Katsikeas, Samiee, and Theodosiou, 2006; Taylor and Johnson, 2002; Zhou and Cavusgil, 2002). Even though these findings relate to advertising in general, they could apply to the consistency of brand communication in light of this study. Standardization of advertising, in terms of communicating similar core brand values across communications, relates to brand consistency as well.

On the other hand, the characteristics, needs and wishes of individual consumers (or consumer segments) differ among each other and evolve over time. This would mean that brand communication activities should be customized and/or revitalized according to the needs of specific consumers in order to be most effective. An example of this could be that one advertisement targeted towards older people emphasizes a car its fuel efficiency, while another advertisement targeted towards younger people emphasizes a car its performance (e.g. acceleration and top speed). In this case the advertisements are expected to be most effective in terms of generating sales and creating positive attitudes, since both ads communicate the specific value that is preferred by their target group. However, viewed from a more aggregate level; consumers might get confused about the actual positioning of the brand, since the marketing activities are inconsistent in communicating similar core brand values. By emphasizing the fuel efficiency of a car, a brand could be perceived as conservative and rational. But on the other hand by emphasizing the high performance of a car, a brand could be perceived as outgoing and innovative. In the end, both target groups might have very different (weak) brand associations, resulting in a weak overall brand image. Thus, in this specific case and from this perspective, the brand image dilutes instead of being enhanced.

The previous example clearly indicated the possibility of a two-sided effect of brand communication inconsistency. Inconsistent branding in the form of customized communication activities could have positive effects on profitability, but at the same time negative effects on brand image. Seen from another perspective, customized (inconsistent) advertising could result in short-term profits and long-term losses. However, this was merely an example based on assumptions to illustrate the possible effects of brand communication inconsistency.

besides a brand its positioning being consistent over time, each part of the marketing mix should reinforce this positioning by being consistent as well. The product characteristics, the advertising message, price points and the distribution channels should all be aligned. The idea behind this is the point that inconsistencies in communication activities may lead to re-evaluation of the brand and ultimately reduce brand image (and brand equity). This consumer ‘brand re-evaluation’ shall be explained further. “When expectations are disconfirmed (e.g. in the case of inconsistent brand communication) consumers engage in constructive processing

that can result in revised brand valuations. In contrast, when expectations are confirmed, consumers are less likely to engage in constructive processing and more likely to rely on stored judgements of the brand (Buchanan et al., 1999)”. This is in alignment with findings of

other authors; “when consumers try to make sense of incoherent associations (e.g. in the case of inconsistent brand communication) they will engage in cognitive elaboration and scrutinize

all available information contained in the associations (Lee and Schumann, 2004;

Meyrs-Levy et al., 1994)”.

Thus, re-evaluation of the brand due to inconsistent brand communication is likely to have a negative influence on brand image, i.e. in terms of reduced strength of brand associations. As mentioned in the introduction of this paper, these findings and arguments are straightforward and appear to be widely adopted and applied in practice. The following hypothesis can be formulated:

Hypothesis 1: Brand communication inconsistency, in terms of not communicating similar core brand values, has a negative influence on the strength of brand associations in the minds of consumers.

2.4 Brand consistency and brand attitude

study, but is rather conceptual and based on the findings of previous research. Nevertheless, the conclusions of Charters are very helpful in guiding this research.

The main finding and conclusion of Charters his research can be illustrated with a quote: “With most goods there is a clear desire for identical attributes, and the consumer will search

for consistency from consumption event to consumption event. Indeed, with many goods (such as airplanes or cars) consistent quality is not just desirable but essential for the security of the consumer. However, there is a range of goods where both the nature of production and the expectations of the consumer mean that precise consistency in the product attributes is not only impossible, but also undesirable (Charters, 2009)”. Charters argues that even for the

brands where consistency is desired (e.g. airplanes or cars), it is important to realize that “it is

not consistent content, but a consistency of quality and the ability to deliver up to and beyond expectations that is crucial for the brand (Charters, 2009)”. The type of products and brands

that are meant by the second category, i.e. those where precise consistency in product attributes might be undesirable, are mostly; creative, aesthetic or ludic brands which are consumed by high involvement consumers. Building variation and inconsistency into products can raise the interest of (high involvement) consumers. Variation and inconsistencies might surprise consumers, and increase the enjoyment of purchasing and using the product. This could then result in more positive perceptions and attitudes towards brands that are characterized by such inconsistencies, and where variability becomes an attractive part of the product. Moreover, “consistency may actually be an impediment – its continual “sameness”

gets in the way of the consumer’s ability to reshape the brand for their own symbolic or experiential purposes (Charters, 2009)”.

and are therefore less likely to generate positive attitudes towards the brand and its communications. However, there is an important note to be made with this line of reasoning. Charters (2009) argues that inconsistencies (i.e. variations and novelties) can increase the interest of especially high involvement consumers. Thus, the positive effect of brand communication inconsistency is most likely dependent on the level of brand involvement of a consumer. This moderator effect is discussed further in section 2.7.

Contradictory findings from literature on change management indicate that brand communication inconsistency might result in a more negative instead of positive attitude towards a brand. People have a natural fear of change, and when change occurs they feel a loss of control (Evans, 2001). Change removes the ability to know and therefore the ability to manage the future. Furthermore, change removes the sense of control over ones affairs (Harari, 1999). Moreover, Eagle (1999) argues that uncertainty (i.e. fear of the unknown) is an important reason why people do not change (quickly). Fear of change results in a need to ‘cling’ to the present situation (Harari, 1999). The need to cling to the present situation can be translated into the need for consistency. If this line of reasoning is applied to brand communication consistency, then it is likely that inconsistencies will have a negative influence on the attitude of consumers towards the brand (e.g. becoming ‘annoyed’). Since Charters (2009) his arguments for the positive effect of brand inconsistency on brand attitude mostly apply to high involvement consumers in particular, the hypothesis concerning brand attitude will be based on the idea that consumers fear change (i.e. inconsistency) in general. Hypothesis 2: Brand inconsistency, in terms of not communicating similar core brand values, has a negative influence on the brand attitude of consumers.

2.5 Brand involvement

The next relevant concept is brand involvement. To begin with, an important notion to be made is that the level of consumer brand involvement is not equal to consumer brand loyalty. Involvement can be defined as “a person’s perceived relevance of the consumption object

based in inherent needs, values and interests (Zaichkowsky, 1985)”. Perceived relevance is

in order to purchase and consume a product from the brand they are loyal to. Brand loyalty can be divided in behavioural loyalty, which is obviously about actual customer behaviour, and attitudinal loyalty, which is about positive associations with- and attitudes towards a brand. It becomes clear that brand loyalty refers to actual customers. However, brand involvement refers to both customers and non-customers. A non-customer can still be highly involved with a brand. For example, a Porsche fan can have strong perceived relevance and a strong knowledge base (e.g. in the form of developed associations) but no financial resources to actually purchase a sports car. Thus, the consumer brand involvement in light of this study is based on the idea that high involvement consumers (i.e. customers or non-customers) have

a stronger perceived relevance and have developed stronger associations (and attitudes) towards a brand than low involvement consumers.

2.6 Brand involvement and brand image

Recall that Buchanan et al., (1999) argued that brand inconsistencies will cause consumers to re-evaluate the brand, resulting in a dilution of brand image (and brand equity). Fazio et al., (1986) do not directly deny this line of reasoning, but they found that high involved consumers have developed stronger brand associations, for example through direct experience from usage. Therefore, they are less likely to re-evaluate these brand associations (i.e. brand image) than low involvement consumers, which are likely to have weaker brand associations. Berthon et al. (2009) support this, by stating that consumers their current knowledge base plays an important role in how they interpret brand communication, and that the knowledge base can differ significantly between consumers. Thus, high brand involvement consumers are likely to have a larger and stronger knowledge base, i.e. developed brand associations, and are therefore less likely to re-evaluate these associations. Formulated differently, brand communication inconsistency is likely to have a less negative (weaker) effect on the brand image of high brand involvement consumers than on the brand image of low brand involvement consumers. Thus, the following hypothesis can be formulated:

Hypothesis 3: In the case of brand communication inconsistency, high (brand) involvement consumers are less likely to re-evaluate their image of a brand in terms of reduced strength of brand associations, than low involvement consumers.

2.7 Brand involvement and brand attitude

interest. This line of reasoning, which is already discussed in section 2.4, implies that the (strength of the) effect of brand inconsistency on brand attitude, differs according to a consumer his or her brand involvement level. Reviewing previous research concerning consumer brand involvement levels resulted in several supporting findings for Charters his line of reasoning. First of all, several authors found that consumers view brands differently according to their involvement levels (Lockshin et al., 1997; Santos et al., 2006). Furthermore, it is evident from a number of studies (focusing on the wine product category) that for many consumers, particularly those of a higher involvement level, consistency is explicitly to be avoided (Beverland, 2005; Charters and Pettigrew, 2007). Variations, novelties and differences over time may increase the value of a brand from a consumer perspective. Thus, brand inconsistency, in terms of product attributes, is likely to have a less negative (weaker) effect on the brand attitude of high brand involvement consumers than on the brand attitude of low brand involvement consumers. It is interesting to investigate whether this line of reasoning concerning brand inconsistency in terms of product attributes also applies to brand inconsistency in terms of communicating similar core brand values. This leads to the following hypothesis:

Hypothesis 4: In the case of brand communication inconsistency, high involvement consumers are likely to develop a less negative brand attitude, than low involved consumers.

2.8 Perceived risk

Reviewing previous studies on the topic of brand consistency also indicated that the level of perceived risk, in purchasing a brand its products, might be influential in the effects of brand inconsistency on brand attitude. The level of perceived risk for consumers is a multidimensional concept, and can be divided in five types of risk, i.e. functional risk, financial risk, social risk, psychological risk and physical risk (Shimp and Bearden, 1982). However, in this research the focus lies on the first four types of risk, since they are most relevant and appear to be related to each other. If the purchase/usage of a product does not meet ones expectations (i.e. functional risk in the case of a utilitarian product and social- and/or psychological risk in the case of a symbolic product), the investment made by purchasing the product could be seen as a financial loss (i.e. financial risk).

essential for the security of the consumer (Charters, 2009). Thus, it appears that consistency can reduce the risk of purchasing, by giving the consumer a sense of security. Vice versa, inconsistent brand communication would fail to create this sense of security for consumers, which logically leads to a (less) positive brand attitude. The level of perceived risk has not been directly related to the effects of brand consistency before, and especially not to the effects of brand communication consistency in particular. Therefore, the level of perceived risk associated with purchasing a brand its products seems to be an interesting additional moderator to include in this research. The earlier mentioned findings indicate that brand communication inconsistency is likely to have a more negative (stronger) effect on the brand attitude of consumers with a high level of perceived risk than on the brand attitude of consumers with a low level of perceived risk. However, the level of perceived risk is not likely to influence the negative effect of brand communication inconsistency on brand image (i.e. strength of associations). Thus, the following hypothesis can be formulated:

Hypothesis 5: In the case of brand communication inconsistency, consumers with a high level of perceived risk are likely to develop a more negative brand attitude, than consumers with a low level of perceived risk.

2.9 Conceptual model

In order to maintain a clear overview, all research hypotheses are graphically displayed in a conceptual model (see Figure 1).

3. Research design

The aim of this section, which consists of six parts, is to describe the research design in detail. First, the overall experimental design shall be discussed extensively. The next part covers the stimuli, followed by a brief description of the participants (target group). Furthermore, the variables and scaling are constructed based on items and measurements developed and/or used in previous studies. After that, the experimental procedure will be described step-by-step in order to obtain an even more clear view on this research. Finally, this section will conclude with a plan analysis, i.e. how the collected data will be analyzed.

3.1 Experimental design

However, the conceptual model of this research also includes two moderator variables, i.e. the level of brand involvement and the level of perceived risk, which are also measured in the quantitative questionnaire. The variance in brand image and brand attitude between the consistent group and the inconsistent group could then be related to the level of brand involvement and the level of perceived risk in order to test whether their moderating influence was significant. This way, both the main effects and the moderating effects are included in the experiment, so that all proposed hypotheses could be tested appropriately. A 2x2x2 design could be applicable to the conceptual model of this research, but there are several reasons why this was not the most viable option. First of all, both the level of brand involvement and the level of perceived risk are measured on a metric level. Using a 2x2x2 design requires these variables to be recoded into nominal variables (e.g. low-high involvement and low-high perceived risk), which obviously reduces the amount of valuable information and the reliability of the research results. To conclude, the experimental design that is chosen for this research is based on a consistent (control) group and an inconsistent (manipulated) group.

3.2 Stimuli

The conceptual model has been investigated by applying it to an actual brand. The reason for this is that the findings and conclusions in the end of this research will be more valuable (and concrete). Moreover, actual brands are likely to contain a mix of associations, which might also differ from each other in terms of strength. Besides this, it is likely that actual brands are more resistant to reconfiguration from an organizational perspective, and also more resistant to re-evaluation from a consumer perspective (Bridges, Keller and Sood, 2000). This increases the complexity of the research, but the results are expected to be more valuable in terms of reliability and generalizability.

have chosen appeared to be relevant. HUGO BOSS its product range offers clothing for every occasion: business wear, leisurewear, sportswear and dinner wear, including shoes and accessories for men and women. However, the results of their research indicated that the core associations, or brand image, of HUGO BOSS in Australia can be described as strongly

masculine (i.e. non-attribute based association) and as mainly offering formal/business wear

(i.e. attribute based association). Especially the masculinity of HUGO BOSS has been supported with empirical evidence by the study of Jung and Lee (2006) as well. Matthiesen and Phau (2004) argue that the brand communication of HUGO BOSS has been influential in the development of such a brand image in the mind of consumers. Referring back to Charters (2009) his study, his arguments and conclusions support the choice for HUGO BOSS for this research, since he claimed they are relevant for fashion- and luxury goods.

Consumers will most likely perceive a HUGO BOSS advertisement displaying women

wearing leisure-wear as a form of inconsistent brand communication, since the

Figure 2: inconsistent- and consistent HUGO BOSS advertisements

3.3 Participants

The experiment has been conducted among 100 participants, which were selected randomly. This number seems common in similar experiments that include two moderator variables (e.g. Faircloth et al., 2010). If a certain gender is heavily underrepresented more participants were included in the experiment, since the aim was to balance the number of males and females in order to control for differential gender effects. Alreck, Settle, and Belch (1982) found that men tend to exaggerate the differences in brands more markedly than women. Men are also found to perceive a masculine brand to be more masculine and a feminine brand to be more feminine. Nonetheless, the use of students as participants of the experiment is considered appropriate, since they are ought to be part of the target group of HUGO BOSS. Besides this, the use of students is common in experimental research (e.g. Faircloth et al., 2001).

3.4 Variables and scaling

Brand communication inconsistency was measured in order to make sure that participants actually perceived consistent- and inconsistent brand communication (i.e. print advertisement) as consistent and inconsistent. The level of inconsistency was measured using a seven-point scale, i.e. “is this advertisement similar to other/previous advertisements of HUGO BOSS that you have seen?” (where 1=”different” and 7=”similar”) and “is this advertisement in accordance with your expectations of HUGO BOSS its advertisements?” (where 1=”unexpected” and 7=”as expected”).

The strength of brand associations (i.e. brand image) of HUGO BOSS was measured by including three items of a seven-point scale, as suggested by Keller (1998). “To what extend do you feel the following values are descriptive for HUGO BOSS as a brand?” (where 1=”strongly disagree” and 7=”strongly agree”).

______ Masculine ______ Business/Formal ______ High quality

Brand attitude was measured using two items of seven-point semantic differential scales, i.e. 1=”good” and 7=”bad” and 1=”like extremely” and 7=”dislike extremely”, adopted from Berger and Mitchell (1989) who found that this scale has a high internal reliability (> .90).

The level of brand involvement has been defined as “a person’s perceived relevance of the

brand its products (Zaichkowsky, 1985)”. Recall from the theory section that a clear

distinction has been made between brand loyalty and brand involvement. Involvement has been measured in previous studies by administering three (seven-point) Likert statements that ought to tap the underlying concept of perceived relevance, i.e. the brand means a lot to me, the brand matters to me, and the brand is important to me (e.g. Traylor, 1981). However, Zaichkowsky developed a well grounded 20-item semantic differential scale in order to measure the level of involvement (see Appendix B). This scale appeared to have sufficient content validity, reliability over time, criterion-related validity and construct validity (Zaichkowsky, 1985), which made it suitable for this research.

likely”). Similar items were included in the experiment of Jung and Lee (2006). HUGO BOSS its products could be seen as mostly hedonic or symbolic, and thus more likely to be associated with social and/or psychological risks instead of functional risks. However, four items of all four risk types were included in order to obtain complete data.

3.5 Experimental procedure

Students were requested to participate in the experiment by means of an e-mail notification. The e-mail contained a hyperlink, which enabled the participant to access the material, i.e. consistent- or inconsistent brand communication and a questionnaire online. See Figure 3 for an example and overview of the questionnaire. Participants were assigned randomly to either the consistent group or the inconsistent group. First, participants were exposed to the consistent- or inconsistent brand communication of HUGO BOSS. Afterwards they were asked to fill in a questionnaire (see Appendix C), containing multiple questions for each construct that needed to be measured: brand image, brand attitude, level of brand involvement and the level of perceived risk. In order to disguise the goal of the experiment, i.e. the effects of brand communication inconsistency, the HUGO BOSS brand and its identity were not introduced, and the questionnaire included a distraction element in the form of a video commercial as well. This way the experiment was kept as realistic as possible. After completion, the participants were thanked for their time and participation.

Variable/construct No. items Example question

Stimuli - -

Distraction element - -

Brand image 3 To what extend do you feel the following value is descriptive of HUGO BOSS as a brand?

Masculine 1 2 3 4 5 6 7

Brand attitude 2 What is your overall evaluation of HUGO BOSS as a brand?

Bad 1 2 3 4 5 6 7 Good

Advertisement attitude

2 What is your overall evaluation of these advertisements of HUGO BOSS?

Brand involvement 20 Make a judgement of HUGO BOSS as a brand.

Important 1 2 3 4 5 6 7 Unimportant Undesirable 1 2 3 4 5 6 7 Desirable

Perceived risk 4 According to you, how likely is it that a HUGO BOSS product will not function accordingly?

Very unlikely 1 2 3 4 5 6 7 Very likely

Perceived inconsistency

2 What is your judgement of these HUGO BOSS advertisements compared to previous/other HUGO BOSS advertisements that you have seen?

Very different 1 2 3 4 5 6 7 Very similar

Figure 3: example and overview of quantitative questionnaire

3.6 Statistical analyses

As mentioned before, the experiment was designed in such a way that the collected data were be suitable for analysis of between-group variance (main effects) and regression analyses (main effects and moderator effects). The main reason why a regression analyses is chosen over a 2x2x2 experimental design is that the first type of analysis supports the possibility to include continuous or metric (moderator) variables. Furthermore, a regression analyses is suitable to test the proposed hypotheses since its results indicate whether the dependent variables can be explained from the independent variables. Moderators can be included in the regression model in the form of interaction effects (i.e. X*Z). As explained in section 3.1 and 3.2, the main analysis is based on the stimuli (i.e. control group and manipulated group) as an independent variable. Belonging to a group is then a dummy variable, where the consistent group= 0 and inconsistent group= 1 (an alternative dummy coding of -1 and 1 is used as well in order to double-check the results). However, additional regression analyses are based on the individual level of perceived inconsistency (see section 4.7). This way, results of both approaches can be compared to each other in order to improve the reliability of the findings.

4. Results

4.1 Pre-test/Manipulation check

The pre-test was conducted among 16 participants, of which 56% were female and 46% male. Half of the participants (control group) were exposed to consistent brand communication, and the other half (manipulated group) was exposed to inconsistent brand communication. The inconsistent brand communication consisted of advertisements displaying women wearing leisure-wear (see Appendix A). HUGO BOSS its proposed brand image, as described in the previous chapter, is confirmed by the results of the pre-test. On a seven-point scale, HUGO BOSS was perceived by the consistent (control) group as strongly masculine with an average score of 5.88 (out of 7) and strongly formal/business with an average score of 6.00. Additionally, HUGO BOSS was perceived as a high quality brand with an average score of 5.63. Thus, the designed inconsistent brand communication, which communicates brand values that are the opposite of masculinity and formal/business, appear to be relevant for this experiment.

The main goal of the manipulation check was to confirm that the participants actually

perceive the designed consistent- and inconsistent brand communication as consistent and

inconsistent. The results of the independent-samples T-test, which compared the means of the perceived inconsistency-score, confirm that the designed stimuli significantly differ in terms of perceived consistency/inconsistency. The designed inconsistent brand communication (manipulated group) was significantly perceived as more ‘different’ (t= 2.935 and p= .011) and as more ‘unexpected’ (t= 2.571 and p= .022) compared to previous/other HUGO BOSS brand communications, than the designed consistent brand communication (control group). Thus, the perceived inconsistency in brand communication is significant, which makes the designed stimuli suitable for the experiment.

Furthermore, the results of this pre-test indicated that the items included in the questionnaire are reliable. First of all, the two items that measure the level of perceived

consistency/inconsistency of the designed brand communication (stimuli) appear to be

reliable, with a Cronbach’s Alpha of .838. Secondly, the two items that measure brand

attitude, i.e. good/bad and extremely like/extremely dislike, appear to be reliable as well with

a Cronbach’s Alpha of .823. However, if the two items that measure the attitude towards the

stimuli are included, the four items together have an even higher Cronbach’s Alpha of .831.

obtained with the experiment. The twenty sub-items of Zaichowsky’s Personal Involvement

Index (see Appendix B) had a sufficient Cronbach’s Alpha of .893. To conclude, the results of

the pre-test/manipulation check were sufficient in order to continue with the actual experiment (n= 100).

4.2 Descriptive statistics and reliability

Before the actual experiment is conducted, it is relevant to discuss the descriptive statistics of the dataset (n= 100). The respondents were 51% female and 49% male. Such a balanced sample was desirable in order to control for possible gender-effects as mentioned in section 3.3. Furthermore, the average level of brand involvement appeared to be 4 out of 7, which also indicates that the dataset is balanced.

Variable Consistent group

Mean value Inconsistent group Mean value Result Gender 47% male 53% female 51% male 49% female p= .696

Age 21,35 y/o 22,51 y/o p= .230

Brand involvement 4.098 3.873 p= .236

Perceived risk 2.417 2.776 p= .073

Perceived inconsistency 5.373 4.020 p= .000*

Figure 4: overview of descriptive statistics and independent-samples T-test results *= Significant at .05.

Reliability analysis

Construct/Variable Items Cronbach’s Alpha

Brand image  Masculine  Formal/Business

.701

Brand attitude  Dislike/Like  Bad/Good

.738

Advertisement attitude  Dislike/Like  Bad/Good

.866

Brand involvement  Zaichowsky’s 20 item Personal Involvement Index

.921

Perceived risk  Functional risk  Social risk

 Psychological risk  Financial risk

.833

Perceived (in)consistency  Different/Similar  Unexpected/Expected

.833

Figure 5: Cronbach’s Alphas of multi-item variables

In order to conduct the conclusive analyses, the multiple items should be merged into singular variables. Reliability analyses indicated whether the Cronbach’s Alpha was sufficient (> .60) for these items. Fortunately, all items had a Cronbach’s Alpha larger than .60, meaning they could be merged into singular variables. See the table in Figure 5 above for the individual Cronbach’s Alphas.

4.3 Brand communication inconsistency – brand image (main effect)

Consistent group Inconsistent group

Brand image 5.902 (out of 7)* 4.796*

Brand attitude 5.520 5.378

Advertisement attitude 5.588* 4.714*

(High-quality association) 5.84 5.57

Figure 6: overview of mean scores (dependent variables) *= Significant difference (p< .05)

An additional ANCOVA indicated that brand communication inconsistency did not negatively influence the strength of high-quality association. The difference in the strength of this association between the consistent group and the inconsistent group appears to be insignificant with p= .170 and F= 1.908. A regression analysis confirms these results, with p= .170. This is a relevant and expected finding, since Charters (2009) argued that it is consistent quality, and not consistent content that is important to focus on. However, this finding will be discussed further in the conclusion and discussion section.

4.4 Brand communication inconsistency – brand attitude (main effect)

Another ANCOVA was conducted in order to test hypothesis 2, i.e. the main effect between brand communication inconsistency and brand attitude. The brand attitude of the consistent group did not significantly differ from the brand attitude of the inconsistent group, since p= .453 and F= .566 is relatively small. A regression analysis confirms these results, with p= .453. Thus, hypothesis 2 is rejected. Brand communication inconsistency does not have a negative influence on brand attitude.

Advertisement attitude

regression analysis confirms these results, with p= .000 and B= -.874. Thus, brand communication inconsistency has a negative influence on the attitude towards the advertisement. If hypothesis 2 is recalled, this finding was expected.

4.5 Brand involvement – brand image (moderator effect)

In order to test hypothesis 3, the level of brand involvement has been included in the regression model (as a direct- and interaction effect). This way it is possible to determine whether the moderator influence on the main effect of brand communication inconsistency on the strength of brand image is significant. First of all, the model is overall significant with p= .000. However, the R Square (.290) is rather low. Furthermore, the regression indicates that the moderator effect of brand involvement is significant with p= .006 and B= -.634. Brand involvement strengthens the negative main effect of brand communication inconsistency on the strength of brand image. This finding is the contrary of what was expected, since brand involvement does not weaken the negative main effect of brand communication inconsistency on the strength of brand image, as proposed in the hypothesis. Thus, hypothesis 3 is rejected. This moderator effect is displayed graphically in Figure 7 below. The slope of the green line, which represents high involvement respondents, is steeper than the other line (i.e. low involvement).

4.6 Brand involvement and perceived risk – brand attitude (moderator effect)

Hypothesis 4 and 5 could be tested simultaneously by including both the level of brand involvement and the level of perceived risk in the regression model. However, it appears that these two variables are correlated (-.435), which means that there is (most likely) an issue of

multicollinearity. In order to solve this, each moderator variable will be included in the

regression model separately. If brand involvement is included, the regression model is overall significant with p= .000. The R Square (.443) is decent. Furthermore, the regression indicates that the level of brand involvement does not significantly influence the main effect of brand communication inconsistency on brand attitude (p= .114). Thus, hypothesis 4 is rejected (for now). However, as expected the positive (B= .823) direct effect of brand involvement on brand attitude is significant with p= .000.

If the level of perceived risk is included, the regression model is overall significant with p= .000. However, the R Square (.264) is rather low. Furthermore, the regression indicates that the level of perceived risk does not significantly influence the main effect of brand communication inconsistency on brand attitude (p= .182). Thus, hypothesis 5 is rejected. However, as expected the negative (B= -.343) direct effect of perceived risk on brand attitude is significant with p= .009.

Advertisement attitude

Since the main effect of brand communication inconsistency on advertisement attitude appeared to be significant (see section 4.4), it is relevant to test whether this effect is influenced by the level of brand involvement and the level of perceived risk. If brand involvement is included, the regression model is overall significant with p= .000. However, the R Square (.226) is rather low. Furthermore, the regression indicates that the level of brand involvement does not significantly influence the main effect of brand communication inconsistency on advertisement attitude (p= .499). However, as expected the positive (B= .456) direct effect of brand involvement on advertisement attitude is significant with p= .019.

4.7 Additional regression analyses

An additional regression analysis, which is based on the individual level of perceived

inconsistency, instead of a dummy variable where (0= consistent and 1= inconsistent), will

contribute to the reliability of this research since both types of regression analyses will compensate for the shortcomings of each other (further discussed in section 5.3). The results of all analyses are compared to each other in Figure 8.

Relationship/Effect Regression series 1:

Consistent/Inconsistent

Regression series 2:

Perceived inconsistency

H1: Brand inconsistency  brand image p= .000 B= -1.106 p= .000 B= -.285

H2: Brand inconsistency  brand attitude p= .453 B= -.142 p= .697 B= -.023

H3: Brand inconsistency  brand image  Moderator: level of brand involvement  (Direct*: level of brand involvement)

p= .006 B= -.634

p= .070 B= .336

p= .196 B= -.87 p= .333 B= .291

H4: Brand inconsistency  brand attitude  Moderator: level of brand involvement  (Direct: level of brand involvement)

p= .114 B= -.258

p= .000 B= .823

p= .047 B= -.089 p= .000 B= 1.022

H5: Brand inconsistency  brand attitude  Moderator: level of perceived risk  (Direct: level of perceived risk)

p= .182 B= -.227

p= .009 B= -.343

p= .898 B= .006

p= .015 B= -.503

Brand inconsistency  advertisement attitude p= .000 B= -.874 p= .012 B= -.180

Brand inconsistency  advertisement attitude  Moderator: level of brand involvement  (Direct: level of brand involvement)

p= .499 B= -.160

p= .019 B= .456

p= .631 B= -.033 p= .098 B= .505

Brand inconsistency  advertisement attitude  Moderator: level of perceived risk  (Direct: level of perceived risk)

p= .996 B= .001 p= .138 B= -.248

p= .600 B= -.035 p= .604 B= -.143

Main effects

The table above shows that, for the main effects, the results of the regression analyses that were based on the individual level of perceived inconsistency are corresponding with the results of the regression analyses that were based on the consistent (control) - and inconsistent (manipulation) group. First of all, both regression analyses indicated that brand communication inconsistency has a negative influence on the strength of brand image. Thus, there is relatively strong evidence for hypothesis 1. Secondly, both regression analyses indicated that brand communication inconsistency does not have a significant (negative) influence on brand attitude. Thus, hypothesis 2 should clearly be rejected. Third, both regression analyses indicated that brand communication inconsistency negatively influences the attitude towards the advertisement. Even though this relationship was not hypothesized this specifically beforehand, the evidence is relatively strong.

Moderator effects

First of all, the moderator effect of the level of brand involvement on the main effect of brand inconsistency on brand image appeared to be insignificant p= .196 and B= -.87, while the first regression analysis indicated that it was significant. Thus, the evidence for this moderator effect appears to be rather weak if all results are taken into account.

Secondly, the moderator effect of brand involvement on the main effect of brand inconsistency on brand attitude appeared to be significant with p= .047 B= -.089. Thus, brand involvement appears to weaken the negative main effect of brand inconsistency on brand attitude (recall that the main effect was insignificant). This result is interesting, since the first regression analysis indicated that this moderator effect was insignificant (p= .114 and B= -.258). Nonetheless, the moderator effect of brand involvement is displayed in Figure 9.

Third, both regression analyses indicated that the level of perceived risk does not influence the main effect of brand inconsistency on brand attitude. However, the direct negative effect of perceived risk on brand attitude is significant. This effect is logical and expected.

Figure 9: plot of brand involvement moderator effect

5. Conclusion

This section aims to evaluate the results of the previous section, and to come up with valuable conclusions that provide grounded answers to the research questions. Each hypothesis will be discussed again. Even though the results indicated several relationships that are insignificant or unexpected, it is still relevant to provide a (theory-based) explanation for these findings.

5.1 Brand image

Secondly, in the case of brand communication inconsistency, high (brand) involvement consumers were more likely to re-evaluate their image of a brand in terms of reduced strength of brand associations, than low involvement consumers. This moderator effect was the contrary of what was expected (see Figure 7), thus hypothesis 3 was rejected. In order to explain this, it is important to recognize that the results also indicated that high brand involvement consumers did not have stronger brand associations than low involvement consumers. This assumption was supported by Fazio et al. (1986) and formed the basis of hypothesis 3. So, a different line of reasoning than the one discussed in the theory section is required in order to explain the moderator effect of brand involvement. A logical explanation is that high involved consumers, i.e. those with a high level of personal relevance, are more likely to look at the advertisements/stimuli longer and more carefully (Zaichowsky, 1985). By taking a closer look at the advertisements they will engage in more elaborate thinking (Lee and Schumann, 2004; Meyrs-Levy et al., 1994), which in turn increases the extend to which a consumer re-evaluates the brand (Buchanan et al., 1999). In other words, high involvement consumers are more susceptible to the (negative) influence of inconsistent brand communication. Low involvement consumers are less susceptible to the influence of inconsistent brand communication, because they might look at the advertisements quickly and without extensively thinking about it (due to less personal relevance). However, it is important to take into account that the evidence for the moderator influence was not extremely convincing, since only one of two regression analyses found a significant effect.

5.2 Brand attitude

Besides this, the level of brand involvement appeared to have a significant moderator influence. In the case of brand communication inconsistency, high involvement consumers are likely to develop a less negative brand attitude, than low involvement consumers. Thus,

hypothesis 4 can be confirmed. However, the evidence is weak, since only one of two

regression analyses found a significant moderator effect. The plot of this effect (see Figure 9) even indicates that high involvement consumers will not (at all) develop a more negative brand attitude in the case of inconsistency than in the case of consistency. This finding is in alignment with Charters (2009) his line of reasoning that especially highly involved consumers will appreciate variations and novelties since it can raise their interest level. If both moderator influences are taken into account, it appears that high involvement consumers are more likely to re-evaluate their brand associations and at the same time appreciate these changes (i.e. re-evaluations). So, the reduced strength of their brand image, due to brand inconsistency, does not also negatively influence their overall evaluation of the brand. Finally, low involvement consumers will develop a more negative brand attitude in the case of inconsistency than in the case of consistency, which is expected (natural fear of change).

The level of perceived risk appears to have no influence on the main effect of brand inconsistency on brand attitude. The results were clearly insignificant. However, the direct

negative effect of perceived risk on brand attitude was significant. Thus, consumers with a

Advertisement attitude

As mentioned before, the main effect of brand communication inconsistency on brand attitude appeared to be insignificant. However, if the focus shifts to the attitude towards the advertisement specifically, the negative effect of brand inconsistency appears to be significant. The line of reasoning discussed in the theory section is still applicable to this relationship, since advertisement attitude could be seen as part of the brand attitude (more aggregate level). Thus, it appears that inconsistent advertisements provoke negative feelings towards the advertisement, since consumers have a general tendency to fear change (e.g. Evans, 2001; Eagle, 1999).

Besides this, the moderator effect of brand involvement appears to be insignificant, but results do show signs that are corresponding with the moderator effect (see Figure 9) that was found for the main effect on brand attitude. Furthermore, the level of perceived risk does not appear to influence the negative main effect of brand inconsistency on advertisement attitude either. The explanation that was given in the previous section applies to this case as well. Finally, the positive direct effect of brand involvement on advertisement attitude seems logical.

5.3 Limitations

Even though the aim was to minimize the limitations for this study as a whole, there are still some drawbacks that should be taken into consideration. First of all, the experiment was based on a consistent (control) group and an inconsistent (manipulation) group that were either exposed to consistent- or inconsistent brand communication (i.e. stimuli). The limitation of this approach is that the control group might contain respondents who perceive the consistent advertisements as inconsistent, and vice versa. In order to partially overcome this, the level of perceived inconsistency was measured on an individual level as well (and analyzed again). However, this second approach has a major drawback as well, since it might be unrealistic to directly as a respondent about their level of perceived inconsistency (unconscious process). Nonetheless, both approaches compensate for the shortcomings of each other.