Chapter 3 – Research Design

(1)

Master Thesis

University of Groningen - Faculty of Economics and Business Msc BA - Specialization Marketing Management

March 2013

Leonie van den Hoven (1608444) Roelof Hartstraat 17-3

1071 VG Amsterdam 06 16 36 36 10 Leonievdh@hotmail.com First Supervisor: Dr. J.A. Voerman Second Supervisor: Alec Minnema

(2)

2

Preface

The master thesis that lies in front of you, is my last challenge in finishing my study and forms an end to a fantastic study period in Groningen. Writing this thesis began as a hurdle but soon became a challenge and I am very satisfied with the result.

I would like to give my special thanks to my supervisor Liane Voerman, who made me see that writing a thesis is a journey and that SPSS is not so ‘scary’ as I thought. Her guidance, support and constructive feedback has helped me a lot. I would also like to thank Alec Minnema for his feedback in the final phase.

I could not have completed this thesis and my studies without the support of some special people in my life. Therefore, an enormous thanks goes out to my parents as they have supported, motivated and stimulated me all those years and made me believe in myself. Also I would like to thank some close friends that have pushed me, had discussions with me about this thesis, giving me feedback along the way and giving me endless support and motivation. They know who there are! Without them I would not have finished this thesis in such a satisfying way.

Leonie van den Hoven Amsterdam, March 2013

(3)

3

Abstract

The rise of the internet has brought about many changes on the web and offered many new opportunities for advertising. The role of online advertising on this ‘new’ medium has become more and more important and has not only had impact on companies’ way of advertising as well as on consumers’ responses. Online advertisement is mainly measured by companies by the Click-Through Rate, while the changes in consumer responses are often forgotten about. While there are different consumer responses, the focus of this thesis lies on affective responses such as liking, being a subpart of the attitude towards the ad.

In this research, the impact of the liking of online banner advertisements is investigated, with several factors (background characteristics and covariates) influencing this relationship. One of the main experimental variables is congruity with the background website, meaning whether the same product is displayed on the website as on the online banner and whether this has an impact on consumers’ liking of banner ads. The second experimental variable that is investigated is the degree of informativeness. This variable entails textual stimuli, visual stimuli and a combination of textual and visual stimuli, the effect of these three aspects on a consumers’ liking of the banner ad is inspected.

As mentioned, several factors influence the relationship between the congruity and degree of informativeness on the liking of the banner. These entail annoyance with online banners, brand familiarity, earlier experience with the brand, product involvement and a consumers’ style of processing. The following problem statement is formulated and will be investigated in this research:

‘How do congruity with the background website and the degree of informativeness impact the liking of the banner? And how is this relation moderated by the style of processing and the level of product involvement?’

Data is gathered through an online questionnaire, 199 participants responded. A 2 (congruity/incongruity) x 3 (visual stimuli, textual stimuli, visual & textual stimuli) research design was used to test the effects on liking of banner advertisements.

The results of the research indicate that congruity with the background website does not influence the liking consumers have on banner advertisements. The relation between degree of informativeness on a more positive liking of the banner is found.

Furthermore, only two covariates were found to have an effect on banner liking: having earlier experience with the brand strengthens a positive effect on liking. And even though the presence of textual or visual stimuli separately do not generate a more positive effect of product involvement on liking, a significance has been found when both visual and textual elements are present in slightly strengthening the positive effect of product involvement on liking.

This thesis calls for more research in the field of consumer responses on online advertising and a more thorough look on the variable ‘degree of informativeness’ should be conducted. This research shows that consumers do see difference in information stimuli as being more or less effective in their liking towards a banner ad, however this effect could be investigated more in depth.

Keywords: Online Advertising, Liking, Congruity, Degree of Informativeness

(4)

4

Chapter 1 – Introduction

Nowadays, we cannot think of a world without internet: internet users spend more time on the web during daytime than viewing television or tuning on to a radio (Tsao & Sibley, 2004).

Internet enables us to perform routine tasks quickly and efficiently, but it is not only a mechanism for information dissemination. It is also an interactive social platform, where people engage in communication and interpersonal behaviour (Weiser, 2001).

More and more consumers are pushed to be interactive and engage in conversation or some sort of action. Morrissey (2005) describes the internet from the consumers’ viewpoint as an ‘access to more information about the market, being complemented by larger choice sets due to the global reach of the internet, by the ability to exchange information and opinion with peers, to change their own perceptions and behaviour in rapid and largely unchecked manner, and to define brands on their own’. Pires et al. (2006) even argue that there is a changing power relationship between consumers and suppliers. This is mainly seen as a benefit as it enhances the number and quality of value propositions, market knowledge, the ability to search and the ability to take advantage of alternatives (Pires et al., 2006). This has resulted in more customer-centric thinking and customization, which should lead to a more satisfied customer, as they get what they said they wanted.

However, one of the major problems with internet is ‘information overload’ (Edmonds & Morris, 2000). The issue is mostly that, while there is enough information at hand, it is often difficult to obtain useful, relevant information when it is needed (Edmonds & Morris, 2000). This often causes consumers to click away from a webpage or not look at the entire content that is provided to them.

With this big shift towards internet, many companies shifted their communication channels towards the internet and marketeers are increasingly pursuing an integrated multichannel communication strategy to increase advertising effectiveness (Wakolbinger et al., 2009). Cho (1999) states that with new interactive communication media (such as the internet), all the existing theories about advertising that have been applied to the internet is actually dubious, as the internet with its’ two- way communication has different characteristics from traditional one way exposure media.

According to Hollis (2005), online advertising began in 1994. Since then it has grown very rapidly, mainly by dotcom companies promoting their own services online. The growth of online advertising has (had) a major impact on companies’ way of advertising as well as on consumers’ responses.

Attracting individuals’ attention and persuading them remains a critical issue for the practitioner (Yun & Yoo, 2007). Internet provides consumers with a freedom of action, selecting time and place for their website browsing (Gallagher, Foster & Parsons, 2001).

Fulgoni & Morn (2009) say that even in today’s economically challenging times a shift seems to appear towards online paid advertising: payment requirements when consumers perform some desired action (such as clicking on the ad). Over the years, there have been ups and downs regarding the ‘boom’ of online advertising, which was mostly measured by the Click-Through Rate. Hollis (2005) states that compared to TV ratings and print readership, estimated CTR indicates an active response to the advertising, not simply a probable exposure to it. However, at the same time Dreze & Hussher (2003) state that a click may not be a relevant measure of the actual impact of display advertising; as low click rates are not always evidence that an ad did not have impact on consumer behaviour.

This is also researched by Danaher & Mullarky (2003) as they discussed conflicting opinions on whether to concentrate on click-through or exposure-based figures for the measurement of online advertising. In online advertising, consumers often look at the ads in a pre-attentive level, without looking at them consciously (Dreze & Hussher, 2003). They also claim that traditional brand equity measures as brand awareness and advertising recall should be relied on more and more.

Wakolbinger et al. (2009) support this and say that ‘online advertising has been found effective both in communicating corporate messages and in strengthening brand equity’.

(7)

7 As one of the most influential theories in marketing communications and advertising research is attitude towards the ad (Goldsmith& Lafferty, 2002), the need for more research on the effect it has on consumers’ behaviour, their reaction to certain ads, their attitude towards it; especially in an online context is needed.

In this thesis, research will be conducted to determine how online banner advertising affects consumer response types. As mentioned, the advertising market is shifting towards online execution more and more, but is mainly measured by facts as Click-Through Rates (CTR), while mostly forgetting the change in consumer responses it brings about. This chapter will start with some background information about the subject and will serve as a ‘knowledge domain’ for the following chapter. It will clarify the choice for one type of consumer response that will be pursued throughout this research as well as the choice for banners in online advertisements. Subsequently, the problem statement and research questions are presented. The chapter will end with the theoretical and managerial relevance of this thesis and with an outline of the structure used throughout this research.

1.1. Consumer Responses on Advertisements

Research has shown that there are many forms of consumer responses on ads for measuring advertising effectiveness. As Olney et al. (1991) state, there is more than just one response type, advertisement effects are measured at different levels, varying from behaviour (in sales) to affect (attitude to brand or ad), to cognition (beliefs and judgements), to attention or exposure (readership/click through rates). Much research has been done on actual consumer behaviour, however in this research, the focus will be on the affective response ‘liking’, being a subpart of the attitude towards the ad. Affective responses represent the consumers’ feelings of liking and (un)favorability (MacKenzie et al., 1986). ‘Consumers’ attitudes are considered important to track because they likely influence consumers’ exposure, attention and reaction to individual ads through a variety of cognitive and affective processes’ (Schlosser et al., 1999).

1.2.1. Response type: attitude

Attitude changes are the result of the valence of thoughts, whether these are positive, negative or neutral; together they determine the direction of a consumers’ attitude (Fennis & Stroebe, 2010).

Lutz (1985) defines attitude towards the ad as follows: ‘a predisposition to respond in a favourable or unfavourable manner to a particular advertising stimulus during a particular exposure occasion’.

In another research, he continues that the attitude that is formed towards the ad may influence the attitude towards the brand as well as their purchase intentions (Lutz et al., 1983).

Goldsmith & Lafferty (2002) find that when consumers like an advertisement, they are more willing to develop a liking for the brand and are more predisposed to buy it. They also found a positive attitude towards the ad important because ‘if the ultimate goal of advertising is to form positive attitudes towards the ad and the brand, thus increasing the likelihood of a purchase, then a positive emotional response to an ad may be the best indicator of advertising effectiveness’ (Goldsmith &

Lafferty, 2002).

Brown & Stayman (1992) found that a positive emotional response to an ad enhances positive brand liking, brand attitudes and consumers are even more likely to buy the product. Whereas much research is done in the field of brand cognitions having or not having a significant effect on brand attitude, Brown & Stayman did find a significant effect.

Brown & Stayman (1992) further state that there is increasing interest to affective consumer responses as ‘liking’ of the ad might well be the best indicator of advertising effectiveness. Affective experiences entail for example entertainment/irritation, cognitive experiences consists of for example the informativeness and behavioral entails the utility of online advertising in making decisions. These experiences trigger either a positive, negative or neutral reaction.

(8)

8 The question remains whether the above described research on the attitude towards the ad is generalizable to online advertising. Goldsmith & Lafferty (2002) argue that the only difference between a banner ad and a ‘normal’ broadcast ad, is the way the ad fills up the screen and the way it compels viewers to attend to the ad or to actively tune it out.

So, in this research there will be looked at the effect an online ad has on the consumer response:

liking, as a subpart of attitude towards the ad.

1.2. Characteristics of online banner advertisements

Recent paper discussion described many factors influencing the banner ad, such as animation, the size and position of the banner, the entertainment value and textual vs. graphical only to name a few (Zia-ul (2012), Danaher&Mullarky (2003), Dreze&Husscher (2002), Schlosser et al. (1999), Yun Yoo (2004)). However in this research, I chose to look at the following two variables more in depth:

Congruity of the website background product category with the ad and the degree of informativeness of the banner ad (textual vs. visual or a combination of the two). As much research has been done into congruity with the background website as well as the in visual/textual information processing, but never as a combination and with its effect on liking of the banner ad, this forms an interesting area of research.

The first variable ‘congruity’ will look at the website as a context and the banner on it, after that the degree of informativeness in the form of visual stimuli, textual stimuli or a combination of the two will be used to look at the banner itself as a complexity. The variables will be outlined more thoroughly in the following chapter.

Over the years, many different kinds of online banner ads have developed. Two current forms are:

banner ads and target ads (Cho, 1999). Target ads can also be seen as linked sites derived from the banner ads. In this research, the focus will lie entirely on banner ads. Banner ads form a traditional, noninteractive and passive advertisement, just like a traditional magazine ad, unless they are clicked on. Cho (1999) states that banner ads require action (i.e. being clicked on) for a consumer to be able to process information. The internet is more action-oriented and interactive than traditional media and calls for a more conscious cognitive effort to process the information a banner ad contains.

As soon as consumers voluntarily click on the banner, the communication becomes a two-way interaction and the information processing becomes a more active process. Consumers will then be exposed to more information and they are forced into a more dynamic cognitive learning.

But what are the essentials in a banner ad, what exact elements make a banner ad good? What defines good or bad? Zia-ul (2012) states that ‘interactive settings in banner advertisements appeal to the online advertising. Interactivity could be an effective tool for the online advertisement because it makes a two-way communication possible’. He also says that in online multimedia advertisement environments, both visual and audio expressions play an important role in grabbing a persons’ attention.

Next, Hollis (2005) states that a brand should be communicating in an appealing manner in order to be included in the consumers’ consideration set. Consumers feel something is appealing when it meets their needs, so for example when an ‘acceptable price’ is presented. Olney et al. (1991) found that ads appealing to feelings of consumers exert positive direct influences on ‘pleasure, hedonism and viewing time’.

(9)

9 Some covariates have an effect on how congruity and degree of informativeness influence liking.

These covariates are introduced below and will be explained in more detail in chapter two.

Covariate Brand Experience & Familiarity

Advertising can also play an important role in shaping expectations and in focusing attention on positive aspects of the experience of the ad. The challenge of an ad is that you want the relevant memory to come to mind when people are thinking about the brand. When a brand is already well- known, ads are more used as a strengthening tool by triggering existing mental connections (Hollis, 2005).

Covariate Annoyance

Online advertising also has its dark sides and is not so positively received in every research.

Goldsmith & Lafferty (2002) found that internet advertising can be perceived by consumers as

‘nonsensical, uninformative, unfocused, forgettable and generally ineffective‘. They state that the advertisements mainly lack communication of brand building information, it is often unclear what the product is and why it is needed. Furthermore, they found descriptions of online advertising as ineffective because of low click through rates for banner ads, its lack of useful information, its dullness or lack of interest, its offensiveness and the way it confuses and annoys consumers.

Covariate Style of Processing

Individual differences in processing information is an emerging area of research (Childers et al., 1985). Different people, in different situations may have different goals, skills and prior experience when being confronted with an advertisement. This is supported by Hollis (2005), who states that frequently people are not actively looking for a purchase but that they are seeking for entertainment, information relevant to sports, events, hobbies etc. They experience a more passive encounter with the internet and thus with the ad. The way consumers prefer to process the information they are confronted with, can matter here.

Covariate Product Involvement

Consumers are expected to actively search for and use information to make informed choices.

Nevertheless, according to Zaichkowsky (1985), a great deal of consumer behavior does not involve extensive search for information or an evaluation of alternatives. This has led to the theory of looking at consumers having high or low product involvement in order to understand the purchase and consumption behavior (O’Cass, 2000). This level of product involvement therefore effects the relationship of degree of informativeness and congruity on liking of the banner.

1.3. Problem Statement & Research Questions

As explained, while internet becomes of more importance, companies shift almost automatically towards internet advertising, but have not been looking at the actual effectiveness and the possible effect on the value of the brand and to the attitude of consumers towards the branded ad especially.

The objective of this thesis is therefore looking at the effect of online banner advertisement on the attitudinal response towards the banner advertisement. This has resulted in the following problem statement:

How do congruity with the background website and the degree of informativeness impact the liking of the banner? And how is this relation moderated by the style of processing and the level of product involvement?

(10)

10 The following research questions will provide support in answering the problem statement:

1. How does congruity with the background product on a website impact the banner liking?

2. How does the degree of informativeness (i.e. visual, textual or combination of the two) on a banner ad impact the banner liking?

3. How does a consumers’ style of processing affect the relationship between the banner characteristics and the banner liking?

4. How does the level of product involvement affect the relationship between the banner characteristics and banner liking?

5. What other moderators influence the effect of online banner ads on banner liking?

1.4. Theoretical Relevance

This thesis looks into two ‘layers’ of research, which creates a particular combination of variables that has not been researched in this way before. The two layers consist of first looking at the website-complexity and then looking deeper at the banner-complexity.

First of all, there is looked at the (in)congruity, this entails the website complexity, whether the background website ‘fits’ with the advertised product (Moore et al., 2005). Then the banner complexity itself is looked at, which contains the degree of informativeness. This contains visual stimuli, textual stimuli or a combination of the two. While much research is done in this area, there has never been looked at the effectiveness of the three levels of information jointly, only separately.

Also, they have never been looked at in this particular combination with congruity on liking of the banner.

The variables congruity and degree of informativeness will be elaborated on in more detail in chapter two, where the theoretical background of this research is provided.

1.5. Managerial relevance

During 6 months, I followed an internship at a multinational FMCG company, where one of the great focus points was online strategy. This existed for the largest part out of Facebook and out of online advertising. One of the problems encountered there was that the company invested a large sum of money in online advertising, although they were not really aware of the effects it actually had and whether it paid off at all. This was striking to me, especially when I found out that the Click-Through Rates were pretty low. However, reactions came in on Facebook about our ads, which were mainly positive. So this made us assume that the ads, even though they were not clicked on, might be more effective than the data with CTR’s suggested.

One of my major surprises during this internship was that one of the most successful multinational companies did not even have a set-out strategy for their online activities while investing so much money in it. Not knowing exactly what the effects of their actions and investments are, but still doing (and seemingly being successful at it too) triggered my interest. What is it then exactly that triggers consumers’ in online ads?

1.6. Structure of Thesis

In the following chapter, a theoretical framework is developed, in which the relation between the banner advertisement and the liking of the banner is further outlined. Along the way, hypotheses are presented, which will help building the final conceptual model. This final model will be tested in the third chapter, containing the research design including the research methods, plan of data collection and the plan of analysis. The results coming from the developed survey will be analysed and conclusions will be drawn from it. In the last chapters, recommendations, limitations and possible further research will be presented.

(11)

11 Chapter 2 – Theoretical Background

In this chapter, an academic framework is provided around the variables for which the ‘knowledge domain’ has been described in the previous chapter. From there on, the specific variables will be elaborated on in more detail. The academic framework will discuss two different kinds of complexities: the website where the banner ad is presented on and then the banner ad itself will be zoomed in upon. First, the effect of congruity of the website background product with the product that is presented on the background will be discussed as an effect on liking of the banner ad. Then, the degree of informativeness on the banner will be elaborated on, including three forms of cognitive processing: visual, textual as well as combination of the two stimuli. Lastly, the moderators influencing the relationship between the liking of the banner and banner ads are discussed.

Throughout the academic outcomes, the hypotheses derived from the academic evaluation are presented and the relevant variables will be added to the conceptual model. At the end of the theoretical background, the complete conceptual model will be presented.

2.1. Congruity website background with product

In this paragraph, the concept of congruity of the website background product versus the product on the banner is investigated. Congruity means that the same product category that is presented on the website is presented on the banner ad, while incongruity means the two product categories presented do not match.

Moore et al. (2005) state that attitudes towards the ad are influenced by this (in)congruity of placement in environmental background. The environmental background consists of the website as a context, which Moore et al. (2005) define as ‘the primary product information focus’. So for example the website of Wehkamp.nl’s main focus is fashion information. Congruity then takes place when an ad is placed on a website of the same product category that is presented on the banner.

When there is congruity, consumers are thought to have a positive attitude, since they are already capable to assimilate the information at hand and find it fitting with the website they are visiting. It is a pattern they might be expecting and that is logical in their minds.

Newman et al. (2004), found that cognitive consistency theory suggests differences in affect and congruity of association between a banner advertisement and a website should have a strong influence on consumer attitudes. They state that consumers prefer consistency and avoid inconsistency if they can. A balanced state of consistency will exist once the product class associated with a website and a banner advertisement on the website are perceived to belong together.

Furthermore, Moore et al. (2005) researched the similarity of product categories across the website and banner ad as an enhancing factor of grabbing a consumers’ attention, when it falls in the same category. This is also supported by Lane (2000), who found that optimal cognitive processing occurs when congruency and the brand fit well together.

On the other hand Moore et al. (2004) found that when high incongruity is present, so when the information pieces are inconsistent, consumers may perceive it as novel and it may draw more attention and raise awareness as it comes unexpected and as some sort of surprise for a consumer.

Higher recall and recognition values of the information presented may therefore occur. Mandler (1982) stated however that information incongruency may lead to negative attitudes as consumers have experienced difficulty resolving the contrasting information presented to them. Websites are seen more and more as holistic and therefore one unit of information is perceived better by consumers and more favorable ad attitudes and greater behavioral intentions exist when congruity exists between the website and the banner ad.

This has also been tested on television, when looking at a happy show and seeing a happy commercial and looking at a sad show and seeing a sad commercial, consumers have a more positive perception of the ad as the product doesn’t seem out of place (Moore et al., 2005 and Russell, 2002).

This is further supported by Murry et al. ( 1992), who found that when people like the program they are watching, it positively influences the attitude towards the ad and towards the advertised brand.

(12)

12 It doesn’t feel so disrupting to consumers when it appears to be in place and the ad is in line with the program.

In the online context, support was found by Singh & Dalal (1999), as attitude toward the website and toward sponsored advertising were positively related.

The hypothesis following from this part of academic research done is presented here:

Hypothesis 1: When there is congruity with the environmental background, the liking of the banner is positive.

2.2. Degree of informativeness

In online advertising, the advertiser is limited to things the consumer can only see, so hearing or feeling is not involved. Therefore, advertisers have to rely strongly on creativity (Zia-ul, 2012).

MacInnis and Price (1987) state that although imagery and analytical processing are not mutually exclusive, one mode of information processing tends to predominate. Which one however? Should advertisers focus mainly on informing consumers with text or should they also try to trigger them with graphics and visual stimuli? This will be elaborated more on in this paragraph.

Petty and Cacioppo (1986) have created a model that looks into the ways consumers are persuaded in their buying process. They have developed the Elaboration Likelihood Model (ELM) which states that consumers can choose two routes to persuasion when coming across persuasive communication: the peripheral or the central route. The central route entails careful consideration of arguments while the peripheral route refers to people changing their attitudes without thinking, in a more passive way. This model is based on the ‘notion that people hold correct attitudes but have neither the resources to process vigilantly every persuasive argument nor the luxury of being able to ignore them all’ (Petty et al., 1986).

When consumers are highly involved with the product, they have a high motivation to process information and they are willing (or able) to put much effort into processing cognitively (Cho, 1999).

Hereby central cues as arguments, existing beliefs and information are of importance in persuading the consumer. Consumers are in this case actively processing information. However, when consumers have a low involvement with the product and low motivation to process the information, they are not willing or not able to process it cognitively and will rely more on peripheral cues as animation, graphics and music. In this case, consumers are less active in processing information and more prone on processing cues.

Schlosser (2003) adds to this that according to the ELM, the website design that matches the users’

goals should support cognitive elaboration and thus more favorable brand judgments. Whereas applying the imagery side, the website design evoking vivid mental images should engage in more favorable brand judgments. Both of these theories can occur and can be effective also on banners (not only considering websites).

The two ways of information processing, textual and visual, will be looked at in more detail below, as these two influence the attitude consumers have towards a banner ad.

2.2.1. Textual information processing

According to the English Dictionary, there are several short definitions of ‘informative’ in the sense of processing information: ‘tending to increase knowledge or dissipate ignorance’, serving to instruct or enlighten or inform’ and ‘providing or conveying information’.

As explained above, active processing of information in this research, consists of the processing of textual arguments and textual information.

(13)

13 For information being able to have an influence on a consumers’ behavior, it must not only be made available, but consumers also have to be able to process it. It should be easy enough for all consumers to understand and to process it all in the same manner. Research shows that the better the information processing ability, the easier it is for consumers to get in touch with the information and the bigger the chance that consumers react with a positive affective response for the product (Thompson & Hamilton, 2006). When involuntarily exposed to an ad for example, most banners do not fully activate cognitive processing of the message (Cho, 1999). When it is made difficult for a consumer to access information, for example when they want to read something on a website but are disturbed by a pop-up ad, chances are that their attitude towards that website will be less positive.

When consumers are looking for something specific on a website, a text-based website is said to be more effective (Schlosser, 2003). Furthermore, the match between the mode of information processing a consumer is in and the format of the ad has a strong impact on the effectiveness of the ad (Thompson & Hamilton, 2006).

Text in an ad may be pleasurable to consumers when the text exists of simple readings but meanwhile showing the way to more complex readings. The text itself does not provide pleasure, but the stimulation it provides does (McQuarrie & Mick, 1999). According to Mick (1992), ‘the notion of pleasure-of-the-text is readily linked to the concept of attitude-towards-the-ad. When consumers experience an increased pleasure while processing the text, it is more probable that they see the overall ad as more favorable. Of importance is the motivation of the consumers, what do they search for in an ad?

For an ad to be informative, it must entail certain cues that enable consumers to better achieve their personal set of objectives (Resnik & Stern, 1977). Garcia et al. (2000) found in their research that the position of the text in an advertisement matters much for the attention a consumer pays to it. Text that is placed in the top left quadrant and beneath the illustration, is said to receive the most attention.

Olney et al. (1991) discussed the informativeness of ads with respect to the usefulness of that information. The question remained how new information (novelties) should be assessed in the present advertisement form. On the one hand, the ad differs from others so has a high level of uniqueness, on the other hand, the ad might lose its level of familiarity, as the novelty is dependent on the prior experience of the consumer. Both uniqueness and familiarity of the ad are of importance.

2.2.2. Visual information processing

A well-know saying is: ‘A picture is worth a 1000 words’; however, is this true?

According to the English Dictionary ‘visual’ is defined as (amongst other meanings): ‘done, maintained, or executed by sense only’ or ‘having the nature of or to producing an image in the mind’.

Or as Thompson & Hamilton (2006) put it: ‘Imagery is based on a nonverbal, sensory representation of perceptual information in memory, as opposed to more semantic, reasoned processing’.

As explained above, non-active processing or non-formative processing in this research, consists of the processing of peripheral cues as visual and graphical stimuli. This will be elaborated on more in detail below.

Nowadays, consumers suffer more and more of information overload as the information they see is difficult to find or to recognize and they might even underutilize large amounts of relevant information. Therefore, visual stimuli become more important in attracting attention and having an effect on consumers (Tegarden, 1999). According to McQuarrie & Mick (1999), only recently have

(14)

14 researchers started to treat visual imagery as an ‘essential, intricate, meaningful and culturally embedded characteristic of marketing communication’, instead of just as a peripheral cue.

Visualization might be of large help in problem solving and it can even help in processing unidimensional or multidimensional information as it enables consumers to ‘chunk’ up relevant information (Tegarden, 1999). This is of growing importance in the context of information overload and the growth of internet and thus the increasing availability of more and more information.

Moreover, mental imagery closely resembles an actual experience (versus cognitive elaboration) and therefore is more superior in influencing a consumers’ intention (Schlosser, 2003). When seeing themselves using a certain product for example, it automatically influences their purchase intentions and thus their attitude towards the ad.

The use of visualization can have two primary consumer responses: larger elaboration or greater degree of pleasure (McQuarrie & Mick, 1999). Larger elaboration as consumers are drawn to the image in a discursive and imagistic manner, which triggers cognitive activity. Of course, increased elaboration does not automatically mean consumers are persuaded by the ad. A larger degree of pleasure is enhanced as consumers ‘experience’ the ad more when seeing an image, it becomes more alive and personally relevant by imagination.

As can be drawn from research done in TV advertising, it seems to work well as the ‘pictures’

presented to consumers ensure an effective attitudinal change. Images in TV ads are researched to work well because they can accurately communicate emotions, which can be recognized by the consumers and which makes it easy for them to relate to (Rossiter, 1982).

However, the type of image or visualization that will draw a consumers attention is found to be a personal matter, dependent on their set of objectives and likings (Garcia et al., 2000).

According to Rossiter (1982), the size, placement, animation and used colors are also of importance when attracting a consumers attention and having a chance at making an impact on their attitude.

The bigger the ad is, the more it ‘springs in the eye’ of the consumer and the more likely it is to draw attention to it.

That colours have an impact on a consumer response is supported by Moore et al. (2005), it is well known that colors can influence someone’s mood and can thus influence an attitude. Although this is of course dependent on cultures, Moore et al. (2005) found that use of colors in ads does actually make a difference. Depending on the product and the goal you want to obtain, you might want to choose warm colors (e.g. red) to generate more attention and arousal rather than cool colors such as blue, which will evoke greater relaxation and pleasure. In general cool colors are received more favorably than warmer colors.

The hypotheses following from this part of academic research done, are presented here:

Hypothesis 2: The higher the degree of informativeness, the more positive the liking of the banner.

However, when researching, the relationship between congruity and degree of informativeness on liking is not as ‘clean’ as it is presented above. There are several factors influencing this relationship, called moderators. These moderators are elaborated more on in the following paragraph.

2.4. Moderators

This paragraph will address certain factors that have an influence on the relationship described above and are called ‘moderators’. Based on literature found, the following moderators are discussed in more detail below: annoyance with online banners, familiarity with the product, previous experiences with the brand, style of processing and product involvement.

(15)

15 2.4.1. Annoyance

Several research has shown that an increase in irritation levels can lead to a general reduction in the effectiveness of all advertising (Aaker & Bruzonne, 1985).

An irritating commercial/advertisement is ‘provoking, causing displeasure and momentary impatience’ (Webster’s online dictionary). This is a stronger feeling than just ‘disliking’ something (Aaker & Bruzonne, 1985). There are many factors increasing the factor of annoyance with advertisements, the product category being one of them. Rettie (2001) found that consumers highly value the ‘flow state’ when surfing on the internet, when this is disrupted by banner ads or other forms of online advertising, this often creates negative attitudes towards the ad. This is further supported by Coulter et al. (2001) as according to them, a negative attitude towards ads is caused by the perception that the consumer is seduced to move away from its intended activity.

This flow disruption during internet usage is seen as most annoying when text, sound and images are all competing for attention (Goldsmith & Lafferty, 2002).

The following hypothesis is proposed based on the academic research:

Hypothesis 3: The higher the annoyance with online banner, the more negative the effect on liking of the banner.

2.4.2. Brand Familiarity

The matter of familiarity with the brand may also play a role. This is researched by Brown & Stayman (1992), and they found that ‘novel brands reported significantly stronger relationships between ad attitude and purchase intentions, suggesting that the existence of prior brand attitudes reduces the impact on ad attitudes’. Also, the product type is of importance according to them, as some product categories have a higher ‘liking’ factor than others.

Alba & Hutchinson (1985) describe brand familiarity as ‘the number of product/brand related experiences that have been accumulated by the consumer including direct and indirect experiences such as advertising exposures, interactions with salespersons, word of mouth communications, trial and consumption. Dursun et al. (2011) state that this familiarity with a certain brand can generate a positive (or negative) affect towards a certain brand.

Given this empirical evidence for the influence of brand familiarity, it is proposed that:

Hypothesis 4: The higher the brand familiarity, the more positive the effect of liking of the banner.

2.4.3. Brand Experience

Kara et al. (2009) researched how the level of previous experiences the consumer had with a certain product influences their perception about the brand. They found that the more experience, the more the consumers perceptions were effected. The greater the knowledge of the product, which leads mostly from a consumers’ previous experiences, the greater the awareness of consumers towards the brand. A positive experience with a product will lead to a positive perception of the brand as a whole and a negative experience with a product will lead to a negative perception of the brand.

The following hypothesis follows from this research:

Hypothesis 5: The higher the brand experience, the more positive the effect of liking of the banner.

2.4.4. Style Of Processing

The way individuals process the information at hand differs per person. Individuals differ in their acquisition of information, the strategies they employ during this acquisition and their utilization of acquired information when forming judgments (Childers et al., 1985). According to Richardson

(16)

16 (1978), the crucial factor in understanding how individuals process information is their tendency to utilize one strategy over a set of alternatives, in this case the choice between processing textual (verbal) vs. visual information.

Whereas much research is already done on verbal information processing, Childers et al. (1985) decided to include visual information processing (mental imagery) in their research as ‘it is a strong facilitator of the type of learning that characterizes consumer information acquisition’.

They divided imaginal processing into three different forms:

- imagery vividness (the clarity of evoking a mental image) - imagery control (the ability to self-generate a mental image),

- imagery style (willingness to engage in imaginally vs. verbal processing).

They developed a new measurement scale called the Style Of Processing scale (or SOP), a 22 item measure, researching an individual’s preferred way of processing information (verbally or visually).

The hypothesis following from this research is presented here:

Hypothesis 6: The less textual elements present in the banner, the stronger the positive effect of Style of Processing on liking.

2.4.5. Product Involvement

Another factor influencing the relationship between banners and liking, is product involvement.

Since products mean different things to different people, consumers form differing attachments to them (O’Cass, 2000).

Considering the Elaboration Likelihood Model (ELM), a consumers’ involvement and level of motivation towards a certain topic and the ability to process the information provided (i.e. being familiar with the topic), affects the recall, recognition and attitude formation (Petty & Cacioppo, 1986). This is further supported by Putrevu & Lord (2003), who argue that consumers’ involvement with the website topic will affect the processing of internet advertising. When consumers have either a high or low involvement, they will limit their attention to and elaboration of the banner ads.

Therefore, the levels of motivation and the ability to process information are of importance (Moore et al., 2005 and Cho, 1999).

The degree of personal relevance and importance determines the level of involvement. When a message is perceived as (personally) relevant or of importance, it is logical that a consumer devotes more time and attention to the message in the ad and that the information is processed at a deeper level (Park & Young, 1986). According to Zaichkowsky (1985), involvement with a product leads to greater perception of attribute differences, perception of greater product importance and greater commitment to brand choice. Product involvement is found to be very personal, the same product has different involvement levels across people. The focus with product involvement should therefore lie on the relevance of the product to the needs and values of the consumer.

O’Cass (2000) describes product involvement in his research as ‘a construct linked to the interaction between an individual and an object and refers to the relative strength of the consumers’ cognitive structure related to a product’.

Cauberghe & De Pelsmaecker (2010) state the product involvement is ‘the consumers’ overall evaluation of how important the product is to his/her life, and is often categorized as situational or enduring’, meaning that it can be evoked from stimuli or cues and staying apparent across situations and conditions. It can be attributed to the storage of personal information within knowledge structure of the product category, such as personal experience. Therefore, Cauberghe & De Pelsmaecker (2010) state, that consumers who are highly involved with the product will attach more meaning to it compared to a product with which they are low involved. They think this implies that the level of product involvement influences the consumers level of motivation to process information, spending more attention, time and effort on highly involved products than on low involved products to process.

(17)

17 Based on the academic research, the following hypotheses have been developed:

Hypothesis 7: The less textual elements present in the banner, the weaker the positive effect of Product Involvement on liking.

2.5. Conclusion

Looking back to the beginning of this chapter, an academic framework is now developed around the introductory chapter 1. First, the relationship between the characteristics of a banner ad and the liking of a banner is described. This characteristics of a banner ad consist of congruity and degree of informativeness, which is further layered into three levels: Textual elements, Visual elements and a combination of Textual&Visual elements. When looking at the effect of textual information processing vs. visual information processing on a consumers’ attitude towards a banner ad, we can conclude that it is not entirely predictable. On the one side it is strongly dependent on the goal of the consumer and on their needs and objectives, on the other side, there is something to say for both stimuli but they are dependent on the level of motivation/involvement the consumer has.

There is not necessarily one better or worse than the other, it depends on the way a person processes information. Therefore, the style of processing is of importance, as this is a very personal trait and different for every person. This is where the moderators come into the picture.

The relationship between the characteristics and liking of the banner is influenced by several factors.

The degree of annoyance consumers have with an online advertisement, the degree to which a brand is familiar to a consumer, the amount of previous experiences with a certain brand, as well as the level of product-involvement, and the style of processing a consumer subconsciously prefer, are all factors of importance in influencing the relationship between banner ads and liking of the banner ad. Annoyance, brand familiarity and brand experience have direct effects on liking of the banner and form background characteristics, while the other covariates (Style of Processing and Product Involvement) have an effect on the relationship between banner characteristics and liking of the banner. As these factors play a dominant role in the testing of this relationship, they can certainly not be ignored. In the following figure, the final conceptual model is presented.

‘Liking’

of the banner Congruity background

website with product category

Degree of informativeness

- Visual - Textual - Visual&Textual

Covariates:

Style of Processing (SOP) Product Involvement (PI)

Background Characteristics:

- Annoyance - Brand Familiarity - Brand Experience

H1

H2

H6, H7

H3, H4, H5 Experimental Variables:

Figure 1: Final Conceptual Model

(18)

18

Chapter 3 – Research Design

Before testing the hypotheses, the research design is set out and choices herein are described.

Thereafter, the population and sample, the data collection, including the operationalization and the design of the questionnaire and its procedure will be explained. Finally, the plan of analysis will be described.

3.1. Research Method

For the data collection an online survey will be developed in order to generate primary data. The causality of this study tries to explain the relationships among certain variables, that have been outlaid by the hypotheses. Cross-sectional studies involve the collection of information from any given sample of population elements only once, they are carried out at once and represent a snapshot of one point in time (Malhotra, 1999).

This research is a 2 (congruity/incongruity) x3 (visual/textual/visual&textual) between-subject factor design (see table 3.1.) (Malhotra, 1999).

3.2. Participants in Sample

As can be seen in Table 3.1., the group of participants to which the questionnaire is sent out is randomly assigned to one of the six conditions. Each of these six conditions must contain approximately thirty persons (Franckeal et al., 2011). Each group will see a different banner/background website combination based on randomization. In a between-subject factor design, each participant will only be shown one condition (Malhotra, 1999). The design process of the background website and the banners will be described below in paragraph 3.3.

Total n=199 Visual (n=67)

Textual (n=67)

Visual & Textual (n=65)

Congruity -0 (n=98)

Group A N=32

Group B N=34

Group C N=32 Congruity -1

(n=101)

Group D N=35

Group E N=33

Group F N=33

Table 3.1.: Between-subject factor design

In total, 223 respondents participated in the research. Of these 223, some cases were incomplete and were deleted from the sample. The remaining 199 respondents remained in the testing sample.

The average age of the sample is 37 years (37.45yrs.), ranging from 17 to 74. There were 56 male participants and 142 female participants. As there are many more female then male participants, it can be imagined that this may influence the representativeness of this research, just as age groups might influence this. Therefore, a small check is conducted to see whether the results of this research have been coloured by the sample. From the 2-way ANOVA that was conducted, no significant value is found for either the age or gender the participants had (See Appendix A). This means that males and females, or the difference in age groups do not differ in terms of their liking scores.

(19)

19 3.3. Experimental Design

In order to measure the experimental variables congruity and degree of informativeness, one product had to be chosen with two different background websites that contained distinctive differences as well as distinctive similarities. As a product, Nike sneakers were chosen and as background website, a car website for testing incongruity and a fashion website to test congruity.

To test the degree of informativeness, three different banners were developed: one with only visual elements, one with only textual elements and one with both visual and textual elements. These were applied on each of the two websites, but they were precisely identical to each other in order to keep it clean and as comparable as possible. An overview of the different groups in testing is provided in table 3.2. below, including the six conditions. The six different websites and banners used in the questionnaire as a whole, can be found in Appendix B.

Congruency with background website

Degree of informativeness

Congruity Incongruity

Zalando website Topgear website

Visual Banner with only visuals:

3 sneakers and the Nike

‘Swoosh’ logo

Group A Group D

Textual Banner with only text:

‘De nieuwste Nike

sneakers, bestel ze NU!’

Group B Group E

Visual

&

Textual

Combination of the two stimuli above:

Both the Visual &

Textual elements

Group C Group F

Table 3.2. Overview of the six conditions of the experimental design

(20)

20 3.3.1.Procedure

Participants were approached mainly via social media (Facebook) and by email and were provided with a link, guiding them to the questionnaire. Subjects were randomly assigned to a condition.

Once opened, the online questionnaire started off with an introduction, explaining what the research is about and what is asked of the respondent. Immediately after the introduction, the participants were shown one of the six different website/banner combinations and were then asked to fill out a questionnaire that took about ten minutes to answer. The questions were the same for every participant. As a last question, some personal data was asked such as age and gender. Finally, the participants were thanked for their participation.

3.4. Operationalization of the Constructs

In creating the questionnaire, scales were selected that have been used successfully in earlier research. The scales used come from the secondary data derived in chapter two and were carefully evaluated in usefulness. Each variable is shortly operationalized in this paragraph and the internal consistency is tested by means of the Cronbach Alpha¹.

An alpha of 0.6 or higher is needed on each question asked to show that every question measures the same construct (Malhotra, 1999). All questions that meet this condition can then be combined into a new variable.

In table 3.3., an overview is given of the different constructs, the measurement scales, the corresponding research and the Cronbach Alpha values, which are explained in more detail afterwards. In the questionnaire, the questions per concept are presented in a different order than has been explained in this research, caused by the length of some of the measurement scales and the fear of not gathering the necessary data. For that reason the Style of Processing scale, which consists of 22 items, has been moved to the end. The complete questionnaire that was sent out, can be found in Appendix C.

o Liking (Dependent Variable)

In order to measure liking, a scale is used based on an analogy on television program liking by Murry et al. (1992). It originally exists out of three statements, seven-point Likert scale with 1=Strongly Disagree and 7=Strongly Agree. Respondents are asked to respond to the statements: ‘I’m glad I had a chance to see this banner’ and ‘This banner invites me to visit the click-through website again’.

However, one statement was not so useful in this particular research so it was left out of the questionnaire. This was the statement: ‘If I knew this banner would have been on this website, I would look forward to seeing it’.

Cronbach Alpha

The two statements scored positively on the Cronbach (α=0.853) and can therefore be combined.

1In order to start tests in the database, some statements needed recoding. This entailed the following statements: Q5.2, Q5.3, Q5.4, Q5.7, Q5.9 (PI); Q6.4, Q6.5, Q6.6 (DOI); Q7.3.(Congruity); Q8.3, Q8.6, Q8.19, Q8.20, Q8.21 (SOP)

(21)

21

Constructs Source Items Cronbach Alpha

Liking Murry et al. (1992) 2 statements, 7 point Likert scale

0.853

Congruity Newman et al. (2004) 5 statements, 7 point Likert scale

0.759

Degree of Informativeness

McQuarrie & Mick (1999) 8 statements, 7 point SD scale

F1: 0.872 F2: 0.722

Annoyance factor

Aaker & Bruzonne (1985) 1 statement, 7 point SD scale

----

Brand Familiarity

Dursun et al. (2011) 3 statements, 5 point Likert scale

0.693

Brand Experience

Kara et al. (2009) 3 statements, 5 point Likert scale

0.818

Style of Processing

Childers et al. (1985) 22 statements, 5 point Likert scale

0.695

Product involvement

Zaichkowsky (1994) 10 adjectives, 7 point SD scale

0.941

Table 3.3.: Measurement scales and constructs

o Congruity (Independent Variable)

The scale used to do the manipulation check on congruity/incongruity was developed by Newman et al. (2004), who generated and extended it from the research done by Lane (2000). This scale contains five statements, on a seven-point Likert scale, where 1= Strongly disagree and 7= Strongly agree. The statements entailed information about the fit of the website and banner as well as the enhancement of credibility they involved. With this scale, it is measured whether a congruent or an incongruent construct is actually recognized as such by the respondent.

Cronbach Alpha

The five statements measuring congruity scored a positive Cronbach value (α=0.759). The Cronbach could be slightly higher (α=0.779) when deleting statement 3, however the correlated item is larger than 0.2 and the difference is so little that the statement is kept based on the argument: the more information at hand, the better.

o Degree of Informativeness (DOI) (Independent Variable)

The degree of informativeness is used for a manipulation check, to see whether respondents recognized the difference in informativeness of the banner depending on the presence of textual, visual or textual and visual elements in the banner.

A seven-point Semantic Differential scale was used to measure the eight item scale, as was used by McQuarrie & Mick (1999). The items are a combination of imagistic values, discursive responses and items to measure the difficulty of comprehension.

(22)

22 Factor Analysis & Cronbach Alpha

At first, an alpha of 0.493 was the result on the eight statements on Degree of Informativeness. As this is not an acceptable score, a factor analysis was run on the questions to get a more clear view on the data. A factor analysis attempts to identify a small set of factors that represents the underlying relationship among a group of related variables (Pallant, 2011).

The factor analysis showed very clearly that Q6.4 and Q6.5 did not belong significantly to either of the two factors (see Appendix D). Therefore, these two questions were deleted from the dataset and the two factors (see for overview, table 3.4. below) were run through a Cronbach test again. This resulted in positive scores for both factors (factor 1: α=0.872 and factor 2: α=0.722). So Q6.1, Q6.2 and Q6.3 were combined in F1 and Q6.6, Q6.7 and Q6.8 in F2.

Factor 1 Factor 2 1. Provokes /

does not provoke imagery ^0,888 ^0,004 2. Vivid/

Dull ^0,854 ^-0,213

3. Interesting /

Boring ^0,930 ^0,003

6. The meaning of the ad is certain/

The meaning of the ad is ambiguous ^-0,071 ^-0,813

7. The ad is easy to understand/

The ad is difficult to understand ^-0,108 ^0,766

8. The ad is straightforward /

The ad is confusing ^-0,136 ^0,818

Table 3.4.: Rotated Component Matrix

3.4.2. Background Characteristics

o Annoyance Factor (Background Characteristic)

The moderating role of annoyance on banner liking in general was measured with one single statement as used by Aaker & Bruzonne (1985). This statement was used as a seven point Semantic Differential scale with 1= No irritation and 5= High irritation.

Cronbach Alpha

As this scale only exists of one statement, the internal consistency can not be calculated and therefore there is no alpha for this scale.

o Brand Familiarity & Brand Experience (BF & BE) (Background Characteristics)

As brand familiarity and brand experience lie very close to each other in definition, both were measured to ensure there was no misunderstanding and no wrong data was gathered. Brand familiarity was measured as done by Dursun et al. (2011) and brand experience as used by Kara et al.

(2009), both by use of a five-point Likert scale on 3 statements with 1= Strongly Agree and 5=

Strongly Disagree. Brand familiarity statements entail ‘I am experienced with the brand’, ‘I have knowledge about the brand’ and I am familiar with the brand’, whereas brand experience is tested

(23)

23 with information about the willingness to buy, whether the respondent would recommend it to their friends and the probability to consider buying brand products in the future.

Cronbach Alpha

For both constructs, three statements were used as a scale.

The statements measuring Brand Familiarity scored a positive Cronbach value (α=0.693) as well as the three statements on Brand Experience, which also showed a good Cronbach value (α=0.818) and can therefore be combined.

3.4.3. Covariates

o Style Of Processing (SOP) (Covariate)

To measure the style of information processing, the 22-item Style of Processing measuring scale of Childers et al. (1985) was used on a 5-point Likert scale with 1= Always True and 5= Not Always True.

This scale measures the extent to which a person prefers to process information in a more textual or a more visual manner.

Cronbach Alpha & Factor Analysis

In order to test the internal reliability, the Cronbach Alpha test is done on each question asked to the participants. The 22 statements measuring the Style of Processing scored an acceptable Cronbach of 0.695 (α=0.695).

It was indicated when removing some statements of the question the alpha could be higher, but this is only a slight difference so it was decided to keep all of the statements in the construct. The more information at hand, the better for the end results.

o Product Involvement (Covariate)

Zaichkowsky (1994) developed an extensive scale in measuring involvement, which she shortened a decade later into a 10-item scale. As product involvement is an important and very interesting variable in this research, this scale was used and was not replaced by a more compact scale.

10 adjectives were measured on a seven-point Semantic Differential scale.

Cronbach Alpha

The ten statements measuring Product Involvement scored a positive Cronbach value (α=0.941).

3.5. Plan of analysis

After gathering all data with the online survey through means of Qualtrics, the data set will be exported to SPSS and an analysis of the effect of the independent variables on the dependent variable ‘liking’ will follow, by using descriptive statistical methods.

1. Cronbach Alpha

First of all, the Cronbach Alpha will be calculated for every variable, to test internal consistency validity. This will show whether the scale used in each question to measure a construct was reliable and whether the results are useful to use. A value of 0.6 or higher is needed to rely on the question and for the internal consistency between two variables to be strong (Malhotra, 1999). When the Cronbach Alpha is not high enough, variables will be deleted to see if it matters and whether the alpha depends on one certain value. When the Cronbach Alpha is high enough, the data is found to be reliable and the questions will be combined.

2. Manipulation Check

Secondly, to check whether the manipulations in congruity and degree of informativeness were correct, a manipulation check will be done.

Congruity will be checked by using an Independent Sample T-Test. This test is used when having two different (independent) groups of people (so in this case people who saw a congruent

Chapter 3 – Research Design

Preface

Abstract

Table of Contents

Chapter 1 – Introduction

Chapter 3 – Research Design