How prior brand attitude and information familiarity influence message sidedness effectiveness Vincent Eijsbouts
MSc Business Administration: Marketing 10253610
2 How prior brand attitude and information familiarity influence message sidedness effectiveness
Vincent Eijsbouts Vincentt__@hotmail.com
10253610
MSc Business Administration: Marketing Supervisor: Antoon Meulemans
3 Abstract
This study is concerned with the relation between advertisement message sidedness and
advertisement effectiveness operationalized as customer-based brand equity (CBBE). The objective is to study to what extent and how the relation between advertisement message sidedness and
advertisement effectiveness is influenced by prior attitudes towards the brand and familiarity with negative information. We hypothesize that in general refutational two-sided advertisements (RTSA) are more effective than non-refutational two-sided advertisements (NRTSA), which are more effective than one-sided advertisements (OSA). These differences in effectivity between the three types of message sidedness are hypothesized to hold for (potential) consumers with a negative prior brand attitude and (potential) consumers with a positive prior brand attitude who are familiar with the negative information presented in the two-sided advertisements. For (potential) consumers unfamiliar with the negative information having a positive prior brand attitude it is hypothesized that RTSA is more effective than OSA, which is more effective than NRTSA. Therefore, the final hypothesis argues that the effectiveness of message sidedness is moderated by prior brand attitude and
information familiarity. Hypotheses are tested through the means of an online vignette experiment with a sample of 295 Dutch participants. Most hypotheses are partially supported. The results suggest that prior attitude moderates the effectiveness of message sidedness.
4 Table of contents Page Introduction……….6 Theoretical framework……….8 Message sidedness………10
Advertisement effectiveness: customer-based brand equity (CBBE) model……….11
Attitude towards the brand……….12
Theories……….13
Adaptation-level theory………..13
Inoculation theory………..13
Social judgement theory……….13
Correspondence theory………14
Conceptual model……….15
Message sidedness to CBBE in general………15
Consumers with a negative prior attitude……….16
Consumers with a positive attitude………17
New negative information……….17
Familiar negative information……….19
Moderation of prior attitude and negative information novelty………..20
Methodology………21 Pretest………21 Sample………25 Design……….25 Method………..26 Procedure……….27 Measurements……….28 CBBE………..28 Prior attitude………29 Familiarity………..30 Manipulation checks………..30 Reading accuracy………30 Overall influence………30
Impact of negative information………..31
5
Usage status………..32
Results………33
Computing scales………33
Variables’ descriptives and correlations……….33
Main analyses………36
Message sidedness effect on CBBE……….36
The relation for negative prior attitude………37
The relation for positive prior attitude unfamiliar with negative information………….38
The relation for positive prior attitude familiar with negative information………39
Moderation of prior attitude and familiarity with negative information……….39
Additional analyses………40
Discussion………...41
Key findings……….42
Limitations and alternative explanations………..46
Insignificance………..46
Non-measured variables explaining the results……….47
Methodological limitations………...48
Practical value of the study……….48
Concluding remarks………..49
6 How prior brand attitude and information familiarity influence message sidedness effectiveness
Introduction
Society increasingly demands corporations to be open and clear about their practices and
characteristics which symbolizes the current age of transparency (Tapscott & Ticoll, 2003; Hultman & Axelsson, 2007; Ball, 2009). This is largely due to perceptions of business unethicality. Organizations can engage in transparency initiatives in many different ways and in many different areas (Hultman & Axelsson; Bushman, Piotroski & Smith, 2004). Common amongst all these transparency initiatives is that they are desired by society due to their information asymmetry reducing function which would theoretically lead to better informed decision making (Cho, Lee & Pfeiffer, 2013). Furthermore, transparency is demanded by society in their quest for more socially responsible corporations (Dubbink, Graafland & Van Liedekerke, 2008). For corporations transparency often entails disclosing information they would rather keep private because disclosure would not benefit them and even be potentially harmful (Dingwerth & Eichinger, 2010). For organizations a major question is if there are ways to engage in transparent initiatives so that not only society is benefitted, but their businesses as well. For example this can occur because the initiatives contribute to organizational goals such profits.
Literature shows there are ways in which this desired dual purpose of transparency can be satisfied (Carter & Curry, 2010; Wehmeier & Raaz, 2012; Buell & Norton, 2011). One approach to transparency deals with message sidedness and is the focus of this study. Message sidedness refers to the inclusion of only positive or only negative arguments (one-sided) versus including both positive and negative arguments in a message (two-sided) (Crowley & Hoyer, 1994). Specifically, the
effectiveness of message sidedness in achieving organizational goals is investigated. This has been the subject of much prior research (Crowley & Hoyer, 1994; Etgar & Goodwin, 1982; Brown & Stayman, 1992; Paula, 2014; Kamins & Assael, 1987; Belch, 1983; Muehling & McCann, 1993;
Pechmann, 1990; Bohner, Einwiller, Erb & Siebler, 2003; Kamins & Marks, 1987). The results of these studies are mixed although the majority argues in favor of using two-sided advertisements (Eisend, 2006; Kamins, Brand, Hoeke & Moe, 1989; Bohner, Einwiller, Erb & Siebler, 2003; Crowley & Hoyer, 1994; Etgar & Goodwin, 1982; Allen, 1991; Kamins & Assael, 1987).
What current literature has not sufficiently tested is the factors that the effectivity of the types of message sidedness are dependent upon. Examples of variables for which additional research is needed are prior attitude and informational novelty (Eisend, 2006; Crowley & Hoyer, 1994; Myers, Sen & Alexandrov, 2010). This study tries to contribute to filling these research gaps. The main research question is: to what extent and how is the relation between advertisement message sidedness and advertisement effectiveness influenced by prior attitudes towards the brand and familiarity with negative information?
7 The study contributes to the literature by studying the terms under which certain types of message sidedness are more effective than others by testing the conceptual model. Practically, the findings can help businesses improve their advertisement programs.
An online experimental vignette study was conducted to test the hypotheses for three relatively diverse brands: Nokia, McDonald’s and ING. A Dutch speaking sample was recruited through personal contact, social media and e-mail. This resulted in 295 participants with a
completion rate of 76.6%. The experiment first asked participants about their brand attitude before the manipulated advertisement was presented. Subsequently advertisement effectiveness in
achieving organizational goals, operationalized as an increase in customer-based brand equity (CBBE) according to the five dimensions identified by Lassar, Mittal and Sharma (1995), was measured. Finally, familiarity was measured along with several manipulation checks and demographic characteristics.
The next section discusses relevant literature, research gaps, concepts and theories. Subsequently, a conceptual model is presented and elaborated alongside five hypotheses. Thereafter, the research methodology is mentioned and explained. Following, the analyses are described and the result are presented. The paper ends with discussion of the key results, limitations and alternative explanations, practical value of the study and some concluding thoughts.
8 Theoretical framework
Advertising is studied from many different perspectives with many different goals. Some of these studied topics relate to advertisement content (Eisend, 2006; Crowley & Hoyer, 1994; Kamins & Assael, 1987; Brown & Stayman, 1992; Allen, 1991; Wilkinson, Alford & Mason, 1976; Belch, 1983; Pechmann, 1990; Muehling & McCann, 1993; Sawyer, 1973), advertisement aesthetics and
presentation (Bottomley & Doyle, 2006; Kamins, Brand, Hoeke & Moe, 1989; Bohner, Einwiller, Erb & Siebler, 2003; Crowley & Hoyer, 1994; Belch, 1981; Muehling & McCann, 1993), advertisement media and modality (Stern, 1994; Brown & Stayman, 1992; Wilkinson, Alford & Mason, 1976; Belch, 1983; Muehling & McCann, 1993) and advertisement processing models such as the Elaboration Likelihood Model (ELM) (Allen, 1991; Paula, 2014; Brown & Stayman, 1992).
As for the previously mentioned transparency requirements that are demanded by modern society, especially the content aspects of advertising is relevant. Although several of the
aforementioned aesthetic (for example color usage (Bottomley & Doyle, 2006)) and methodological changes (for example type of media used (Wilkinson, Alford & Mason, 1976; Belch, 1983;, Brown & Stayman, 1992; Stern, 1994) and endorser reputation (Kamins, Brand, Hoeke & Moe, 1989)) may potentially induce the perception of transparency amongst customers, it is especially the content of advertisement that can explicitly convince the customer of a company’s, and its advertisement’s transparency. What is meant with the content of advertisement is the argumentation used in
advertisements and its characteristics. It are these transparent arguments that explicitly demonstrate transparency, opposed to the more indirect perceptual effect of many of the earlier mentioned advertisement design, presentation and media effects. Plenty of studies addressed advertisement content. Specifically, prior studies studied the argumentation that is or should be used in advertising to create more transparent, effective or efficient advertisements. Examples of variables studied in this respect are message sidedness (Eisend, 2006; Kamins, Brand, Hoeke & Moe, 1989; Bohner, Einwiller, Erb & Siebler, 2003; Crowley & Hoyer, 1994; Kamins & Marks, 1987; McGuire, 1961; Etgar & Goodwin, 1982; Allen, 1991; Paula, 2014; Kamins & Assael, 1987; Wilkinson, Alford & Mason, 1976; Belch, 1983; Sawyer, 1973), quantity of arguments (Eisend, 2006; Crowley & Hoyer, 1994;), strength of arguments (Eisend, 2006; Crowley & Hoyer, 1994; Pechmann, 1990), type of arguments (Crowley & Hoyer, 1994) and advertisement length (Singh, Balasubramanian & Chakraborty, 2000).
The specific focus of this study is on the effects of message sidedness to induce transparency which thus can be beneficial for both (potential) customers, but also possibly for the businesses themselves (Carter & Curry, 2010; Wehmeier & Raaz, 2012; Buell & Norton, 2011). In this study message sidedness is defined as the extent to which information supporting a proposition is presented alongside information not supporting a proposition. The literature studied many aspects of message sidedness. Examples of variables supposedly influenced by message sidedness are
9 attitude towards and acceptance of the advertisement (Crowley & Hoyer, 1994; Etgar & Goodwin, 1982; Brown & Stayman, 1992; Paula, 2014; Kamins & Assael, 1987; Belch, 1983; Muehling & McCann, 1993), attitude towards the brand (Crowley & Hoyer, 1994; Paula, 2014; Kamins & Assael, 1987; Pechmann, 1990), purchase intentions (Kamins & Marks, 1987; Brown & Stayman, 1992; Pechmann, 1990), credibility (Bohner, Einwiller, Erb & Siebler, 2003; Crowley & Hoyer, 1994; Kamins & Marks, 1987; Kamins & Assael, 1987; Belch, 1983; Pechmann, 1990), counter-arguing (Crowley & Hoyer, 1994; Kamins & Assael, 1987) and supportive-arguing (Kamins & Assael, 1987). However, also many executional aspects that may moderate the effects of message sidedness are studied such as the amount and importance of negative and positive information being included (Eisend, 2006; Crowley & Hoyer, 1994; Pechmann, 1990), message repetition and duration (Bohner, Einwiller, Erb & Siebler, 2003; Sawyer, 1973), attribute quality (Eisend, 2006), placement and order of (negative) information (Eisend, 2006), correlation between negative and positive attributes (Eisend, 2006; Crowley & Hoyer, 1994;, Pechmann, 1990), marketers’ voluntariness in disclosing information (Eisend, 2006; Pechmann, 1990), type of arguments (Crowley & Hoyer, 1994), endorsers (Kamins, Brand, Hoeke & Moe, 1989), modality and media used (Wilkinson, Alford & Mason, 1976; Belch, 1983;, Stern, 1994; Brown & Stayman, 1992), order of information (Crowley & Hoyer, 1994), product trial (Kamins & Marks, 1987) and personality traits and characteristics (Myers, Sen & Alexandrov, 2010; Sawyer, 1973). As illustrated by aforementioned, research has been conducted into the cognitive and affective processing of message sidedness. However, few studies examined the effects of personal differences in this specific advertisement context. More elaborate studies are needed to inspect the influence of personal differences in prior attitude, perceived novelty of information, involvement, general need for cognition, attached importance, as well as personal characteristics amongst others (Eisend, 2006; Crowley & Hoyer, 1994; Myers, Sen & Alexandrov, 2010). The main theoretical contribution of this study is to assist in closing the literature gap concerned with prior attitude, while also addressing novelty aspects.
This study attempts to do so by studying to what extent and how the relation between advertisement message sidedness and advertisement effectiveness is influenced by prior attitudes towards the brand and familiarity with negative information.
Currently, there are few studies assessing the influence of prior attitudes towards the brand on the effectiveness of message sidedness in the context of advertisements. The first is Belch’s (1981), through measuring prior brand preference. This study however, did not yield significant results which could originate in the experimental procedure and measures used. The study relied on television commercials exclusively, and the procedure may have been too cumbersome.
Furthermore, Belch’s study focused on prior preference, which does not entirely measure the construct of prior attitude since you can have a positive or negative brand attitude regardless of your
10 brand preference. Finally, Belch’s hypotheses employed different theoretical underpinnings then in this study, resulting in contradicting expectations. Furthermore, a meta- analysis by Allen (1991) did not yield reliable, nor significant results for the moderation of brand favorability. Other studies often assume prior attitudes rather than measuring them (Etgar & Goodwin, 1982).
In studying the before mentioned relations we refer to literature covering adaptation-level theory (Helson, 1947; Richins, 1991; Crowley and Hoyer, 1994), inoculation theory (Kamins & Assael, 1987; Etgar & Goodwin, 1982), social judgment theory (Sherif, Sherif & Nebergall, 1965; Mallard, 2010; Crowley & Hoyer, 1994) and correspondence theory (Kamins & Assael, 1987; McGuire, 1961). This as to offer (possible) explanations, build expectations, and to assess the generalizability of prior studies in the context of this study. Following, these theories and their possible implications will be shortly explained alongside a further operationalization of the crucial concepts studied in current work. Thereafter, a conceptual model will be built.
Message sidedness
As briefly addressed previously, massages and thus advertisements differ in their sidedness of arguments. The main distinction made within the message sidedness concept is the contrast between one-sided messages and two-sided messages.
One-sided messages, also called supportive appeals, consist of nonvaried message claims and “ignore any competitive arguments and use a completely positive claim format” (Belch, 1981). This implies that an one-sided advertisement claims superiority of the advocate brand or product on all mentioned attributes (Belch, 1983). A one-sided advertisement may thus be defined as one that “presents only those arguments in favor of a particular proposition” (Allen, 1991), which is the most traditional form of advertising (Kamins & Assael, 1987).
On the other hand, two-sided messages do recognize inferior attributes of a particular proposition, which is reflected in the presentation of one or more arguments against the particular proposition. Although two-sided advertisements still claim superiority for the advertised product or brand on most of the attributes, they thus disclaim superiority on a single or a couple of attributes also. They represent varied massage claims (Belch, 1981). Consistently, Belch (1983) states that in a two-sided message appeal “the advertiser recognized one or more product attributes where the advocate brand is inferior”. Hence, “a two-sided message presents the arguments in favor of a proposition but also considers the opposing arguments” (Allen, 1991).
This study refers to arguments, attributes, information or claims that are in favor of the advertised product or brand, or reflect its superiority, as positive. Arguments, attributes, information or claims that are against the advertised product of brand, or discount its superiority, are called
11 negative. Hence, positive and negative arguments are opposites and are extremes of a continuous dimension.
Within the two-sided messages there is another distinction to be made, that two-sided messages can be refutational or non-refutational. By definition, two-sided refutational messages mention both positive and negative information, however the presented negative information is refuted in that same message. Hence, a refutational two-sided message presents one or multiple counterarguments against the negative information alongside this negative information. Belch (1981) refers to the refutational appeal by stating that it “refutes the opposing view after acknowledging it”. Allen (1991) clearly explains the intent of this presentation-refutation of negative information: “The intent is to introduce the contrary position and demonstrate why this position is inferior to the position advocated by the communicator”. Two-sided non-refutational messages on the other hand only mention the negative information without refuting it (Allen, 1991). Note that one-sided
messages can’t have refutational properties because refutation implies that conflicting information is provided which would make the message two-sided.
As previously became apparent there are many media to advertise through: examples of which are online, print, television, audio, packaging. For all of these media, the three types of message sidedness can be implemented. This study however, focusses on written online advertising exclusively due to methodological, time, and budget constraints.
Furthermore, the content of two-sided advertisements and the arguments used may differ in several aspects. Some arguments are focused on the own product or brand, while other arguments focus on comparing the own product or brand with competitors’, or even focus solely on
competitors’ products or brands (Belch, 1981; Aronson, 1969; Etgar & Goodwin, 1982). Furthermore, these arguments can be different in that they are more oriented towards communicating
functionalities or information, whilst on the other hand they may evolve around evoking emotions (Chitturi, Raghunathan & Mahajan, 2007; Kempf, 1999; Zajonc & Markus, 1982). Also, arguments, and advertisements in general, may have goals at different levels: some are used to raise awareness, while others are more attitude or even behavioral focused (Vakratsas & Amber, 1999).
Advertisement effectiveness: customer-based brand equity (CBBE) model
Because of these differing goals of advertisements, it may be hard to objectively measure
advertisement effectiveness. However, the overall goal of any advertisement is (almost) always to increase customer (potential) lifetime value. This study argues that customer life-time value, also called future customer value, is well captured by the concept of CBBE since this concept considers the potential value of a customer in different dimensions, and within these dimensions to different extents.
12 The CBBE is concerned with the differential effect of marketing efforts by the brand because of the knowledge of the brand, attitudes towards it, or perceptions about it (Lassar, Mittal & Sharma, 1995; Keller, 1993; Keller, 2001). The model recognizes that brand equity may have positive, or negative influences on marketing. The extent to which brand effects are positive for customers’ evaluations, which is called the CBBE of the brand, depends on various dimensions of the consumers associative network (Keller, 1993) and the level its knowledge and attitudes reach on the consumer based brand pyramid (Keller, 2001). These brand equity effects span from simple brand recognition to intense loyalty towards the brand and thus captures the whole range of possible brand effects. According to Lassar, Mittal and Sharma (1995), it is also the driving force for (incremental) financial gains.
For a brand the first step in building CBBE is to create awareness, which consists of being recognized and being recalled by the consumer (Keller, 1993) and may result in the creation of brand salience. According to the CBBE pyramid this represents the first level of brand equity: the identity level (Keller, 2001). Once customers know of your existence, it becomes important who they think you are, and what they believe about you. This depends on the associative network of the customers which is based on their associations with the brand (Keller, 1993). These associations help to reach higher level in the CBBE pyramid to the extent that they are strong, favorable and unique. The second level of brand equity that may be achieved represents infusing the brand with meaning. CBBE may evoke responses from customers, the third level of the pyramid. Ultimately, brands may build relationships with customers and reach resonance, which represents the top level in the CBBE pyramid (Keller, 2001). To reach this level, and thus to build CBBE, it is crucial that customers have strong, favorable and unique associations with the brand (Keller, 1993; 2001).
Attitude towards the brand
This study will define attitudes as “a psychological tendency that is expressed by evaluating a
particular entity with some degree of favor or disfavor” (Eagly & Chaiken, 2007). In current work, the entity in this definition is the brand. In this context, unless stated otherwise, attitudes are thus the consumers’ evaluation of their favorability towards a particular brand. These attitudes are based on mental residue of experiences with the entity which can take any form since they can comprise of a wide range of cognitive, affective, or behavioral associations. Furthermore, attitudes are internal but may be evaluated and expressed by the attitude holder. Examples of associations that likely influence attitudes are: the brand is exiting, the brand is trustworthy, the brand is scattered, the brand is unethical, etcetera. Because there are countless different associations, resulting attitudes may also differ to a great extent. This study distinguishes between positive, neutral, and negative attitudes. An
13 example of a positive attitude is: I like the brand. An example of a neutral attitude is: I find this brand comparable to the competitor. An example of a negative attitude is: I hate this brand.
Theories
A detailed explanation of all theories that are and will be referred to is beyond the scope of this study, however, their essence (in the current context) will be discussed.
Adaptation-level theory
According to adaptation-level theory, people compare new stimuli with their adaptation level. This adaptation level is a subjective weighted average of all past and present stimulations on a particular dimension (Helson, 1947; Richins, 1991). Basically, the adaptation level is the prior attitude or belief towards the context in which the stimuli is relevant. Because this adaptation level is based on all prior stimuli, the new stimuli will also contribute to it. However, if the new stimulus conflicts with the adaptation level, resistance will occur. Because the adaptation level is formed based on the weighted average of all stimuli, a contradicting stimulus will only slightly influence the attitude of the subject and will thus be largely discounted relative to a stimulus consistent with the adaptation level. Crowley and Hoyer (1994) describe this consequence of the adaptation-level theory as the more consistent a message is with the receiver’s attitudinal schema, the less resistance it encounters. Inoculation theory
Inoculation theory states that if a belief or attitude is attacked with mild counterarguments, a subject will be motivated to bolster one’s defenses against this attitude contradicting information and increase the effort to search for attitude supporting arguments. If the counterarguments are too strong to defend against in such a way, these effects of strengthening the prior attitude by attacking them will dissipate (Kamins & Assael, 1987). However if the prior attitude can be successfully defended against (mild) attacks, it may result in the ability to cope with stronger counterarguments in the future (Etgar & Goodwin, 1982). Concluding, Inoculation theory is about attacking prior beliefs, and the subjects’ responses to those attacks.
Social judgment theory
According to social judgment theory, people evaluate ideas, messages or events by comparing them subconsciously with current attitudes. An individual assesses how to place new information by weighing it against its present point of view (Sherif, Sherif & Nebergall, 1965). Following, based on this comparison, people (subconsciously) judge whether they believe the message is acceptable, hence, whether they believe it (Mallard, 2010). Theoretically, this is described through the placement
14 of the message by the subject in one of three zones or latitudes (Sherif, Sherif & Nebergall, 1965). When the message is accepted, is falls within the individuals latitude of acceptance. This is the range that one perceives as worthy of consideration and reasonable. If the subject sees the message as objectionable and unreasonable, it falls within the latitude of rejection. Third, one may place the message as neither acceptable nor questionable, this is called the latitude of noncommitment. An implication of social judgment theory is that if someone has a negative attitude towards an entity, including (some) negative information about that entity into a message will increase the likelihood of the message falling into the receiver’s latitude of acceptance. This will result in a higher probability of processing the message and reduce the probability of message rejection or counterarguing (Crowley & Hoyer, 1994).
Correspondence theory
Correspondence theory is concerned with subject attributions of the causes of an event. According to correspondence theory someone may attribute the cause of an event to be internal (correspondent), or external (non-correspondent). If the subject perceives the cause of an event to be correspondent, it perceives the event to reflect the true feelings of the correspondent and thus be genuine. Non-correspondent attributions however perceive the event to occur because of external pressures or situational influences. Hence, external attributions do not reflect honest dispositions and are less credible as to their genuine motives. Furthermore, the theory implies that behavior deviation from the norm, are more likely to be attributed correspondently. This because it opposes the norm, and thus external pressure to conform, and thus are more likely to reflect the actor’s true disposition (Kamins & Assael, 1987).
15 Conceptual model
Based on the concepts just explained, the following conceptual model (Figure 1) is constructed to study to what extent and how message sidedness influences customer-based brand equity (CBBE) for different (potential) consumers. First, the relation between message sidedness and CBBE is
discussed. Subsequently, the relationship between these two concepts will be discussed for people with a negative prior attitude towards a brand, and people with a positive prior attitude respectively. Finally, an expected conclusion covering potential moderating factors will be voiced.
Figure 1: Hypothesized Conceptual model
Message sidedness to CBBE in general
Following correspondence theory, consumers attribute advertisement claims either to actual products characteristics (correspondent attribution) or to the advertiser’s desire to sell the product (non-correspondent attribution) (Eisend, 2006). One-sided advertising, being the more traditional advertising format, will receive a more non-correspondent attribution (Kamins & Assael, 1987). Furthermore, the inclusion of negative information in two-sided advertisements leads the (potential) consumer to believe the advertiser is telling the truth since such inclusion typically reduces profits (Eisend, 2006). Take for example Apple that communicates in an mostly positive advertisement that their products are produced under inhumane circumstances. Consumers will tend to believe this message because they do not see any ulterior reason why Apple would communicate something negative about themselves which is not true.
Thus, one-sided advertisements (OSA’s), compared to two-sided advertisements will have lower credibility and higher levels of source derogation (Kamins & Assael, 1987). Moreover, because subjects become receptive to message content only to the extent that a message is perceived as credible, (potential) customers will generally be less receptive to the content of one-sided messages.
Customer-based brand equity Familiarity with
negative information Prior brand attitude
16 Two-sided messages are more likely to reach the threshold of credibility needed to become positively influential for the customer’s brand equity (Eisend, 2006; Belch, 1983).
Opposing, the negative information included in two-sided advertising has a direct negative effect on attitudes towards the brand which implies that two-sided advertisements are not
necessarily more persuasive (Eisend, 2006;). Back to the Apple example. If they communicate negatively about their own production circumstances, (potential) consumers will know this and be less attracted to the brand because of it. Supported by prior studies (Eisend, 2006; Kamins, Brand, Hoeke & Moe, 1989; Bohner, Einwiller, Erb & Siebler, 2003; Crowley & Hoyer, 1994; Etgar &
Goodwin, 1982; Allen, 1991; Kamins & Assael, 1987) however, it is argued that the positive effect of enhanced credibility of two-sided advertisements is larger than the negative effect of including a small amount of negative information. Therefore is expected that two-sided advertisements relative to OSA’s, will result in a higher CBBE.
The same expectations result from social judgement theory (Sherif, Sherif & Nebergall, 1965; Mallard, 2010; Crowley & Hoyer, 1994). Since two-sided advertisements include arguments from both sides, the probability of these advertisements to fall in (potential) customers’ latitude of acceptance is higher than for OSA’s.
In distinguishing the effectivity of non-refutational two-sided advertisements (NRTSA’s) from refutational two-sided advertisements (RTSA’s), it is expected that refutational properties will diminish the negative effects of negative information. This will result in a relatively stronger positive net effect of RTSA’s on brand equity compared to NRTSA’s. If Apple would counter to the bad prodction conditions, they could say this is something they only recently found out and is something they will address before the end of the year. This will reduce the impact the negative information has, because it is a problem that will be solved in the near future. Hence, the first hypothesis is as follows:
H1: Advertisement message sidedness influences customer brand equity as such that (a) TSRAs have a stronger positive influence than TSNRAs, (b) which have a stronger influence than OSAs.
Consumers with a negative prior attitude
For people with negative prior attitudes towards a brand, the key lies in making them believe the information in the advertisement represents the truth and making it credible because these people will otherwise discount the content of the message. Therefore, following the same line of reasoning, two-sided advertisements have a higher chance to receive a correspondent attribution, making them more credible. Furthermore, social judgement theory implies that the inclusion of negative
17 people with negative attitudes. Hence, because two-sided advertisements are perceived as more credible, the content, which is still mostly positive, will be accepted as genuine and will therefore adjust the adaptation level to be more positive. Hence, two-sided advertisements will increase brand equity for people with a negative attitude. Furthermore, because the overall content of the
advertisements for both the one- and two-sided advertisements are mostly strongly
counterattitudinal for negative attitude holders, they will be unsuccessful in building defenses against this counterattitudinal information. Hence, propositions from inoculation (Kamins & Assael, 1987; Etgar & Goodwin, 1982) theory will not be relevant in this context. As for the Apple illustration an OSA may be discounted by Apple-haters for being one of those overly positive advertisements. However, when some negative information is included it catches their attention and they take the context in which is placed, which is mosly positive, more seriously as well.
Since two-sided non-refutational and refutational advertisements both have these credibility and judgement effects that allow them to be accepted and believed by the negative attitude holders, the difference between their effects on brand equity will originate from the refutational properties. Since the RTSA will diminish the negative impact of the negative information by refuting it, it will lead to an net effect on brand equity that is more positive. This leads to the second hypothesis:
H2: For the target audience with a negative prior attitude towards the brand, advertisement message sidedness influences CBBE as such that (a) TSRAs have a stronger positive influence than TSNRAs, (b) which have a stronger positive influence than OSAs.
Consumers with a positive prior attitude New negative information
(Potential) customers that have a positive attitude to the brand, especially for OSA’s, are more prone to perceiving an advertisement as genuine and accepting compared to (potential) customers with a negative prior attitude. This will lead the positive attitude holder to be more accepting of both the one-sided and two-sided advertisements’ content since the content (largely) supports the attitudes they already hold. To illustrate consider an Apple-lovers, if they see an advertisement for a new Apple product they are interested in it and take the advertisement seriously. For them the advertisement does not much become more engaging and credible if there is a bit of negative information in it because they would buy into the OSA as well. For readers with a positive prior attitude towards the brand, the main difference between one- and two-sided advertisements is thus not determined by credibility and acceptance. What causes the differential effect of message sidedness for positive attitude holder is the impact of the included negative information in the two-sided advertisements.
18 In accordance to inoculation theory (Kamins & Assael, 1987; Etgar & Goodwin, 1982), the negative information has the potential to stimulate the building of defenses against negative
information, improving brand equity. Whether these defenses will be built however, depends on the context in which the information is placed, its strength, and its newness (Crowley & Hoyer, 1994) . If there is no defensive response from the positive attitude holder, according to adaptation-level theory (Helson, 1947; Richins, 1991; Crowley and Hoyer, 1994), the negative information is likely to adjust the (in this case positive) adaptation level to become (slightly) less positive, decreasing brand equity. In studying the defensive responses in the context of message sidedness, people with a positive attitude for whom the negative information is new, are distinguished from people with a positive prior attitude familiar with the negative information. For these two groups of positive attitude holders the expectations are as following.
The (potential) consumers with a positive attitude that are not aware of the negative
information contained in a two-sided message, the negative information will be perceived as strongly counterattitudinal since it is not (yet) included in the consumers’ adaptation level (Crowley & Hoyer, 1994). Following, in line with inoculation theory (Kamins & Assael, 1987; Etgar & Goodwin, 1982), this information may represent a too strong attack on the consumers’ attitude for them to defend against (immediately). This implies that inclusion of such negative information will negatively influence the consumers’ prior positive attitudes. Hence, OSA’s, by not including negative information, will yield more positive effects on brand equity than NRTSA’s that include negative information perceived as negative by consumers with a positive prior attitude. Now, return to the Apple-lovers. If they are drawn into the new Iphone advertisement and are not aware of Apple producing them under inhumane working conditions, the inclusion of this information might impact them negatively because it takes them by surprise and it might be too strong for them to build a defence against. Hence, the Apple-lovers that do not know the negative information yet are better off seeing the advertisement without negative information.
Whereas NRTSA’s leave refutation and the building of defenses up to the (potential)
consumers themselves, RTSA’s enhances the defense building by attempting to inoculate (e.g. aiding in refuting information) (Eisend, 2006). By showing the consumer how to possibly defend their positive attitudes against the negative information, RTSA’s weaken counterarguments and thus strengthen the capabilities and desires to bolster defenses for positive attitude holders for whom the counterattitudinal information is new (Kamins & Assael, 1987). Hence, these refutational
advertisements may help consumers to build defenses against attacking arguments, especially when these arguments may seem too strong on first sight as described before (Etgar & Goodwin, 1982). Furthermore, Kamins and Assael (1987) state that refutational advertisements will increase support argumentation and effort to search for arguments supporting the existing position. Take for example
19 the Apple-lovers that wasn’t familiar with the inhumane production conditions. Communicating to them that Apple will address the issues very quickly because they just found out about it will help and stimulate them to shield their perceptions about Apple from getting damaged by the negative information. Thus, including negative information perceived as novel accompanied by a refutation into an advertisement, according to inoculation theory (Kamins & Assael, 1987; Etgar & Goodwin, 1982), increases the brand equity of the consumers with a positive prior attitude. Therefore, for consumers with a positive prior attitude for whom the negative information is novel, RTSA’s are expected to have a stronger positive influence on brand equity than OSA’s. This constitutes the third hypothesis:
H3: For the target audience with a positive prior attitude towards the brand but unfamiliar with the negative information placed in advertisements, advertisement message sidedness influences CBBE as such that (a) TSRAs have a stronger positive influence than OSAs, (b) which have a stronger positive influence than TSNRAs.
Familiar negative information
On the other hand, (potential) consumers with a positive prior attitude may be familiar with the negative information portrayed in a two-sided advertisement. Since the positive attitude holders are familiar with the negative information, it is assumed they have not ignored, rejected or discounted it. The information thus has already influenced their adaptation level and brand equity negatively if they were not able to defend against it. Despite, these consumers still hold a positive attitude towards the brand implying that a subsequent exposure to the negative information will have no negative influences anymore (Crowley & Hoyer, 1994). If anything subsequent exposures represent additional opportunities to elaborate on building defenses against such counterattitudinal
information, emphasizing the propositions of inoculation theory (Kamins & Assael, 1987; Etgar & Goodwin, 1982). Therefore, it is expected that for (potential) customers already familiar with the negative information presented in NRTSA’s, the effects on brand equity will be more positive than for OSA’s. Returning to the Apple illustration, if Apple-lovers are familiar with the inhumane conditions during production, they have had the chance to adapt their frame of reference accordingly and apparently still hold a favorable attitude. Now that an Apple advertisement remembers them about this negative information, they are not surprised and affected by it anymore. If anything they might praise Apple for being honest towards their (potential) customers.
As for the influence of refuational two-sided advertisements on (potential) consumers with a positive attitude familiar with the negative information portrayed, the reasoning is similar to the effects on consumers with a positive attitude unfamiliar with the information. Remember, the main additional influence of the refutation over the non-refutational two-sided advertisement, is that it
20 helps consumers to confirm, build, extend, or explore (new) ways of possibly defending existing attitudes. As mentioned this has positive consequences for these consumers’ brand equity. Following, it is expected that RTSA’s will have a stronger positive influence on brand equity than NRTSA’s. For the Apple example this implies that counter arguing the poor production conditions by mentioning that it has high priority and will be solved quickly might help Apple-lovers who were already familiar with the negative information improve or justify their defenses. This leads to the fourth hypothesis:
H4: For the target audience with a positive prior attitude towards the brand and familiar with the negative information placed in advertisements, advertisement message sidedness influences CBBE as such that (a) TSRAs have a stronger positive influence than TSNRAs, (b) which have a stronger influence than OSAs.
Moderation of prior attitude and negative information novelty
Consistent with prior reasoning, message sidedness is expected to have a differential effect on CBBE for people with a positive prior attitude towards the brand than for people with a negative prior attitude. Specifically, for people with a negative prior attitude, OSA’s will be clearly less effective than both types of two-sided advertisements, as previously mentioned. Whereas for people with a
positive attitude these differences between one- and (especially non-refutational) two-sided advertisements may be more ambiguous and dependent on other factors. Namely, it is predicted that the effect of message sidedness on CBBE of (potential) customers with a positive prior attitude, differs as to their familiarity with the negative information presented. Thus, in line with previous hypotheses, the final hypothesis is as follows:
H5: The relation between advertisement message sidedness and CBBE is (a) moderated by the target audience’s prior attitude towards the brand, and (b) for people with a positive prior attitude towards the brand this moderation is moderated by the perceived newness of negative information about the product.
21 Methodology
Pretests
Before an online experimental vignette study could be conducted to test the hypotheses, two pretests had to be run. The primary goal of the first pretest was to identify brands for which the public opinion is distributed in the desired proportions. The main goal of the second pretest was to find appropriate negative information to include in the two-sided advertisements. Both pretests will be elaborated subsequently.
The preliminary pretest consisted of an online closed-question questionnaire administered in Dutch amongst 23 Dutch participants of whom 21 completed it. Convenience sampling was used to gather these participants. This pretest measured the participants’ attitudes towards a number of brands. For the seventeen brands that were expected to be most suited for subsequent usage a 5-item scale was deployed using a seven-point Likert scale. This scale corresponds exactly to the prior attitude scale explained later in this section (Spears & Singh, 2004). Furthermore, a one question-based attitude measure was administered for 27 other brands that were expected to yield less usable results. The following question was presented alongside a 7-point Likert scale ranging from “very negative” to “very positive”: “How is your attitude towards the following brands?”. An eighth answer possibility was also included: “I’m not familiar with the brand”. For the brands for which the 5-item scale was used, a mean was calculated for the five items. Subsequently a frequency analysis was conducted on all brands to assess their appropriateness for subsequent usage.
A brand was considered ideal if the ratio of participants with a negative attitude relative to participants with a positive attitude falls between 0.5 and 1. This is appropriate because the final dataset needs to be split into two initial groups for analytical purposes; one with a positive prior attitude and one with a negative prior attitude. Because hypotheses testing requires a further
categorization within the positive prior attitude group, a deviation from the negative/positive ratio of 1 would still be suitable as long as this implies that a relatively larger share of participants has a positive attitude towards the brand. This since the chances for encountering statistically significant results are largest if all analyzed groups are relatively similar in size. Hence, the probability of relatively similar group sizes is highest if the brand negative/positive ratio lies between 0.5 and 1.
Based on the frequency analysis that can be found in table 1, three brands were selected for the second pretest and the actual experiment: Nokia, McDonald’s and ING. These brands were selected for several reasons. First of all, these brands showed a negative/positive ratio within, or close to the ideal range. It is important to note the first pretest yielded other brands that may have been more suitable for selection based on negative/positive ratio. However, the selection decision was not solely based on this ratio. A second, very important reasons concerns the fact that these three brands are very diverse in terms of their operating industries, products versus services
22 orientation, hedonic versus functional properties and concrete versus abstract properties. For
example Nokia represents a brand in the electronics industry, offering mainly products with
functional and concrete properties to consumers. McDonald’s operates mainly in the foods industry, selling mainly hedonic products while also emphasizing services and experiences. Finally, ING offers its experiential services in the banking industry. Because these three selected brands are so different, the studies external validity is strengthened. Third, these three brands were selected because it was expected to be relatively easy to come up with an appropriate two-sided advertisement that would be perceived as intended.
Table 1
Frequencies (%) of Prior attitude for the seventeen brands for which all five questions were asked
Negative (%) Neutral (%) Positive (%) Unaware (%) N/P-ratio
Samsung 4.8 14.2 81.0 0.0 0.059 Apple 9.5 0.0 85.7 4.8 0.111 Nokia 23.8 9.5 61.9 4.8 0.384 Blackberry 71.4 0.0 28.6 0.0 2.497 McDonalds 47.6 4.8 52.4 0.0 0.908 Burger King 38.1 4.8 57.1 0.0 0.667 Dominos 19.0 9.5 66.6 4.8 0.285 Pepsi 33.3 4.8 61.9 0.0 0.537 Coca Cola 9.5 4.8 85.7 0.0 0.111 Internet Explorer 76.2 4.8 19.0 0.0 4.011 ING 28.6 19.0 52.4 0.0 0.546 T-Mobile 23.8 23.8 52.4 0.0 0.454 BP 33.3 23.8 19.1 23.8 1.743 Staatsloterij 47.6 23.8 28.6 0.0 1.664 Smart 19.0 4.8 61.9 14.3 0.307 Google 4.8 0.0 95.2 0.0 0.050 Smarties 0.0 14.3 85.7 0.0 0.000 N=21
The second pretest is also conducted as an online closed-question questionnaire administered in Dutch amongst 23 Dutch participants of whom 22 completed it. Convenience
23 sampling was used once more. The second pretest had three purposes. The main purpose was to find appropriate negative information to include in the two-sided advertisements. Negative information was deemed ideal if it satisfied two criteria. First, the information needs to be perceived as
significantly negative. Second, because hypotheses testing requires differentiation between positive prior attitude holders in terms of familiarity with the negative information, a more-or-less equal distribution of familiar versus unfamiliar participants represents another criterion. To find
information that satisfies these criteria, multiple sources of negative information were presented to the participants for each of the three selected brands. A total of thirteen negative, but refutable arguments were presented. For each of the negative arguments participants were asked two questions: “were you familiar with this information before this survey?” and “how does this
information influence your opinion about the brand?”. The respective answer possibilities were yes or no and a seven-point Likert scale ranging from “very negatively” to “very positively”. Table 2 shows the relevant t-tests and frequency results for each of the thirteen arguments as well as the final decision on which information to proceed with. For Nokia and McDonald’s the criteria were satisfied for multiple negative arguments. In these cases, selection was based on the perceived ease of refutation of the respective negative information.
Table 2
Negative information Familiarity frequencies (%), Means (Standard Deviation) and t-value different from neutral (p-value)
Familiarity (%)
Mean (SD) t-value (p-value) Waar Nokia in 2008 nog ruim marktleider was, is het
nu slechts nog een kleine speler
77.3 2.95 (1.090) -4.498 (0.000) Nokia heeft als laatste smartphonefabrikant een
competitief toestel op de markt gebracht
13.6 4.14 (1.207) 0.530 (0.602) Alle smartphone besturingssystemen die Nokia heeft
geïntroduceerd zijn geflopt
45.5 2.64 (1.049)
-6.100 (0.000) Nokia heeft de afgelopen jaren stevige verliezen
geleden wegens teleurstellende smartphone verkopen
59.1 3.05 (1.133) -3.952 (0.001) In producten van McDonalds zijn chemicaliën te
vinden
54.5 2.32 (1.249)
-6.315 (0.000) McDonalds is dusdanig ongezond dat je, zoals de
documentaire 'Supersize Me' aantoonde, ruim 11kg
54.5 2.91 (1.192) -4.294 (0.000)
24 aankomt en je vetpercentage met 13% toeneemt als
je een maand lang enkel McDonalds eet Er zitten dusdanig veel conserveermiddelen in producten van McDonalds dat er meerdere bronnen zijn die beweren dat het soms wel jaren kan duren totdat ze bederven
68.2 2.64 (1.293) -4.948 (0.000)
Het betalingsverkeer via de ING heeft (in 2013) geruime tijd plat gelegen omdat een cyberaanval het systeem beïnvloedde
63.6 2.41 (1.008)
-7.406 (0.000) Van de Nederlandse banken ontvangt de ING de
meeste online klachten
13.6 2.55 (1.057) -6.456 (0.000) ING heeft van de Nederlandse banken de meeste
storingen in het online bankieren
40.9 2.50 (0.859) -8.189 (0.000) In 2014 was ING van plan betaalgegevens van klanten
commercieel te verkopen
27.3 2.00 (1.069) -8.775 (0.000) De ING app heeft minder diverse functionaliteiten dan
die van grote concurrenten
0.0 3.14 (0.774) -5.231 (0.000) De ING app wordt erg langzaam van nieuwe
functionaliteiten voorzien vergeleken met de apps van concurrenten
4.5 3.00 (0.873) -5.374 (0.000) N=22, a mean of 4 represents neutral, <4 represents negative, >4 represents positive
The second goal of this pretest is to briefly asses the validity and reliability of the prior attitude scale and the customer-based brand equity (CBBE) scale. The pretest included the same 5-item and 17-5-item scales for measuring prior attitude (Spears & Singh, 2004) and CBBE (Lassar, Mittal & Sharma, 1995) respectively. These scales will be discussed later this section. The Cronbach’s alphas from this pretest did not show any reason to change the scales.
The third goal of the second pretest was that it could function as a control group for the actual experiment. This is because it was conducted exactly similar to the final experiment with the exception that no treatment was shown between the prior attitude and CBBE measures. Hence, the second pretest represents the neutral, zero, or not manipulated condition. Instead some additional questions regarding negative information were included at the end of the survey.
25 Sample
For the experiment a convenience sampling approach was used to obtain a sample from the Dutch speaking population living in the Netherlands or abroad. Of the contacted potential participants a total of 295 started the experiment. 69 of the participants did not complete the experiment and dropped out prematurely. Following, 226 participants completed the experiment which implies a completion rate of 76.6%. Of the participants 46.5% was male and 53.5% female. The mean age was 29.21 (SD=13.473) and 98.2% lived in the Netherlands at the time of participation. It stood out that the majority of the sample lived in Noord-Holland (61.5%), followed by Zuid-Holland (9.7%) and Gelderland (6.6%). Furthermore, 58.8% of the participants had a university degree. The second and third most reported highest degrees amongst participants were HBO degree (18.1%) and a secondary education (Havo) or pre-university education (VWO) (18.1%). Based on this information it may be concluded that the sample may not be an accurate representation of the population since the sample seems to be younger, more demographically clustered, and higher educated than typical for the population. Incomplete values and double responses were removed from the data before analysis. Missing data was excluded for analysis. No impossible values were found.
Design
The research design can be seen as a 3(Nokia vs McDonald’s vs ING)x3(one-sided vs non-refutational two-sided vs refutational two-sided) between-subject design. The independent, categorical variable of message sidedness consists of three categorical manipulations for each of the three brands that represent the three experimental conditions of this study (Figure 2). The first manipulation comprises a one-sided advertisement (OSA). These types of advertisements consisted of three or four positive written arguments about the respective brands. The second type of treatment for the three brands consisted of a non-refutational two-sided advertisement (NRTA). These were approached similar to the OSA’s except that the selected negative information for each respective brand was presented after the positive arguments. The third treatment concerned refutational two-sided advertisements (RTSA). These advertisements were similar to the respective NRTSA with the exception of refuting the included negative information by including a counterargument at the end of the advertisement.
26 Figure 2: Experimental conditions
For each of the three brands information was gathered from corporate websites and other online sources. This information was used to construct plausible arguments and advertisements as described previously. All arguments used in the treatments consisted of factual information that could be objectively evaluated rather than emotional, subjective arguments that may be perceived differently between participants. This was important to ensure that the arguments had a relatively identical influence on all participants. Furthermore, short bold headers were placed in front of all arguments to summarize and clarify the argument to be made subsequently. The Dutch RTSA’s presented for Nokia, McDonald’s and ING can be found in appendix 1, 2 and 3.
Method
Data was collected through the means of an online experimental vignette study. An experiment is the most suitable way to analyze and test whether the conceptual model holds in practice. The method allows the researcher to manipulate variables to a certain extent and observe the direct effects of these manipulations on participants’ actual perceptions. Confirming causal relationships would be substantially more difficult and less reliable when other methods such as surveys, case studies or literature reviews are used (William, Shadish, Cook & Campbell ,2002). Furthermore, the
Non-refutational two-sided advertisement condition Refutational two-sided advertisement condition One-sided advertisement condition Control condition No advertisement Positive information Positive information Positive information Negative information Negative information Counter argumentation
27 specific context of written advertisements requires a vignette approach. Finally, time and budget constrains required an easy, cheap and quick way for data collection. An online approach is most suitable to achieve this (Weible & Wallace, 1998). Qualtrics was used as the platform to conduct the experiment.
Procedure
Participation invitations were granted to potential participants during a time span of two weeks. Potential participants were approached in one of three ways: face-to-face, social media or e-mail. This invitation was accompanied with a description of some practical information regarding the experiment such as the duration of participation.
When participants decided to participate they were shown a more extensive informative introduction to the experiment. The purpose of this introduction was fourfold: to introduce the researcher, to provide practical information about the procedure, to prepare the participant for the effort required from him or her, and to ensure confidentiality and anonymity.
Following, each participant was randomly assigned to one of the three conditions for two of the three brands. For example, participant A could be included in both the OSA condition for Nokia and the RTSA condition for ING, while not being included in any of the McDonald’s conditions. The choice to include each participant in two out of the three brands was made because including them in one condition for all three brands would likely cause the experiment to lose focus and
concentration due to experiment duration. Using only one brand per participant on the other hand would likely yield too few responses per condition per brand too find significant results.
Each brand condition started with measuring the prior attitude of the participant towards that respective brand to get an objective measure before any manipulation took place. This was based on five questions presented in a random order. Following, the randomly selected treatment for that respective brand was shown. To increase the chance of advertisement processes three measures are taken. First, participants are explicitly reminded that the advertisement can be seen only once and that they need to examine it closely. Second, participants are unable to proceed from the advertisement until a minimum of fifteen seconds have past. Finally, participants are asked to declare to have properly studied the advertisement before proceeding. Hereafter, participants CBBE for the respective brand was measured through seventeen statements that were presented in a random order. Subsequently, this procedure repeated for the second brand. The order in which the brands appeared was randomized to reduce any biases. For example the chance that a participant assigned to Nokia and ING was shown with the Nokia questions and treatment first, was equal to that participant being shown the ING questions and treatment first.
28 Following the questions and treatments for both brands, the participants were asked several control questions relevant for the presented manipulations. Also, some demographic control
questions were asked.
Finally, another informative message was shown to thank for participation, to explain the studied relationships in more detail and to provide the opportunity to ask questions and make comments via e-mail.
Measurements CBBE
CBBE is the main dependent variable in this experiment and literature suggest many
conceptualizations and measurements of this concept. In this study a measure adapted from Lassar, Mittal and Sharma (1995) is translated and used. Lassar, Mittal and Sharma (1995) define CBBE in terms of five dimensions: performance, social image, value, trustworthiness, and attachment. These five dimensions are each measured by three or four items. For each dimension one item was reversely posed for validity purposes. In total CBBE is measured by a seventeen-item scale (table 3). Each item consists of a statement some of which were slightly adapted to fit the brand in question (for example ‘product’ was replaced by ‘service’ for ING). Participants are asked the following: ”Considering (the respective brand), to what extent do you agree with the following statements?”. An example of a statement is “from this brand, I can expect superior performance”. The answer possibilities consist of a seven-point Likert scale ranging from (1) strongly disagree, (2) disagree, (3) slightly disagree, (4) neutral, (5) slightly agree, (6) agree to (7) strongly agree.
This approach to measuring CBBE was selected due to its proven validity and wide
acceptance. Furthermore, although financial estimation of brand equity might yield an even more accurate and continuous representation of the concept, this requires advanced behavior modelling and econometric estimations (Swait, Erdem, Louviere & Dubelaar, 1993). This is beyond the scope of this study to measure due to time, budget and methodological constraints. Also, there are other conceptualizations such as those by Keller (1993). However, it is very difficult, time consuming or costly to measure and quantify these types of (mostly) qualitative conceptualizations. Thus, the selected CBBE measure represents the most suitable measure of the value that customers represent to the brand given the research budget, time and methodology. Increasing this value or equity is ultimately the goal of all advertisements.
29 Table 3
Seventeen items used to measure Customer-based brand equity Performance Van dit merk kan ik superieure producten verwachten
Het is waarschijnlijk dat producten van dit merk defecten vertonen tijdens gebruik
Producten van dit merk zijn zo gemaakt dat zij probleemloos functioneren Producten van dit merk werken zeer goed
Social image Dit merk past niet bij mijn persoonlijkheid
Ik zou trots zijn op het bezit van producten van dit merk Dit merk zal in de smaak vallen bij mijn vrienden
De status en stijl van dit merk komen overeen met mijn persoonlijkheid Value Dit merk is goed geprijsd
De prijs die ik moet betalen voor het merk ligt hoger dan de waarde die ik ervoor terugkrijg
Ik vind producten van dit merk een koopje gegeven alle voordelen die ermee gepaard gaan
Trustworthiness Ik zie het merk en de mensen erachter als zeer betrouwbaar Dit merk houdt rekening met de zorg en het welzijn van de klanten Ik geloof dat dit merk misbruik maakt van klanten
Attachment Het gebruik van producten van dit merk maakt dat ik mij aan dit merk hecht Ik heb positieve gevoelens bij dit merk
Ik zal nooit een warm gevoel krijgen bij dit merk Prior attitude
The moderating variable in this study is prior attitude towards the brand and will be measured using a properly validated and widely used five-item scale adapted from Spears and Singh (2004). These five dimensions are concerned with the extent to which the participants perceive the brand as unappealing versus appealing, bad versus good, unpleasant versus pleasant, unfavorable versus favorable and unlikable versus likable. The participants will be asked the following: “select the answers that best represent your feelings towards (the respective brand)”. Seven-point Likert scales are presented as answer possibilities. An example of one of the five scale’s answer possibilities ranges from (1) very unfavorable, (2) unfavorable, (3) slightly unfavorable, (4) neutral, (5) slightly favorable, (6) favorable to (7) very favorable. The other four items use scales constructed likewise. An eighth answer possibility is added to identify the participants that have never heard of the brand.
30 The eighth answer possibility is “I don’t know this brand”. Including this option seems more
appropriate and valid than to force those unaware participants into a response to reflect on a nonexistent attitude.
Familiarity
To categorize participants as being familiar or unfamiliar with the negative information presented in a respective manipulation, participants are asked the following: “Before this experiment, were you familiar with the fact that (the respective negative information)?”. The two categorical answer possibilities are either “yes” or “no”.
Manipulation checks
Five questions are designed to assess manipulation success and to classify and categorize participants more accurately. Manipulation success is evaluated through frequency analyses and t-tests.
Reading accuracy
The extent to which the reader read the advertisement in detail is measured by asking a truthful answer on the following question: “How well have you read the advertisement of (the respective brand)?”. Three ordinal answer possibilities are provided: “not at all”, “just scanned it” and “read it in detail”.
This question makes it possible to assess the extent to which the participant was involved in the presented advertisement. Reading the advertisement accurately will likely enhance the chance that the message gets across as intended and thus enhance the effects of the treatment. For the Nokia advertisements, 97 participants indicate to have read it in detail, 52 report to have just scanned it and one reports not to have read it. For the McDonald’s advertisements, these respective frequencies are 105, 38 and two. The frequencies for the ING advertisements are 108, 48 and one respectively.
Overall influence
To assess whether all advertisements were perceived as positive as intended, their overall influence on participants is measured. This is measured by asking the next question: “How did this
advertisement of (the respective brand) influence your opinion towards (the respective brand)?”. The seven-point Likert scale answer possibilities range from (1) very negatively, (2) negatively, (3) slightly negatively, (4) neutral, (5) slightly positively, (6) positively to (7) very positively.
This question allows for checking whether the used advertisements are perceived positive as intended. This might represent an early indication of how the advertisement may change the
31 participants’ CBBE. As expected, participants perceived most advertisements to have a positive effect on their CBBE. This was the case for the one-sided (t(54)=7.195, p<0.001), non-refutational two-sided (t(44)=2.562, p=0.014) and refutational two-sided (t(52)=7.485, p<0.01) Nokia advertisements, the one-sided (t(42)=3.469, p=0.001) McDonald’s advertisement, and the one-sided (t(57)=4.715,
p<0.001), non-refutational two-sided (t(56)=4.905, p<0.001) and refutational two-sided (t(44)=3.677, p=0.001) ING advertisements. Contradicting expectations, the non-refutational two-sided (t(54)=-0.830, p=0.410) and refutational two-sided (t(49)=0.843, p=0.404) McDonald’s advertisements were seen by the participants as not influential on their participants’ CBBE.
Impact of negative information
To judge whether the negative information is indeed perceived as negative, participants are asked the following: “How did (the respective negative information) influence your opinion towards (the respective brand)?”. The seven-point Likert scale answer possibilities range from (1) very negatively, (2) negatively, (3) slightly negatively, (4) neutral, (5) slightly positively, (6) positively to (7) very positively.
Participants reported to perceive the negative information as indeed negatively influential for both Nokia (t(96)=-3.713, p<0.001) and McDonald’s (t(103)=-5.806, p<0.001). Unexpectedly, the negative information presented for ING (t(100)=-1.593, p=0.114) was not perceived as negatively influencing participants brand perceptions. However, further examination reveals that the ING negative information was perceived negative in the non-refutational two-sided (t(56)=-2.631, p=0.011), but neutral in the refutational two-sided (t(44)=0.829, p=0.412) condition. It is therefore assumed that the refutation of this information caused participants to discount the negative information’s negativity or it caused them to not attribute the consequences to the ING brand. Concluding, manipulation of the ING negative information was successful despite the early contradicting signs.
Success of counter argumentation
Success of the counterarguments in reducing the specific negative effects of negative information is measured by asking the subsequent to the participants: “To what extent is the effect of (the
respective negative information) on your opinion towards (the respective brand) neutralized by the fact that (the respective counterargument)?”. Four ordinal answer possibilities are provided ranging from (1) not at all, (2) slightly, (3) strongly to (4) very strongly.
For all brands, participants indicated a neutralizing function of the counter argumentation; Nokia (t(52)=9.341, p<0.001), McDonald’s (t(49)=6.547, p<0.001) and ING (t(44)=10.797, p<0.001). The displayed counterarguments successfully refuted the negative information for all three brands.
32 Usage status
Another control variable used is whether the participant is customer or user of the respective brands. To categorize participants along this dimension they are asked the successive question: “Are you a customer/user of (the respective brand)?”. Two categorical answer possibilities are provided: “yes” and “no”.
For Nokia participants were more likely to be non-users (90.1%) than users (9.9%). Amongst participants the majority consumed McDonald’s (71.3%). The distribution customers (52.4%) versus non-customers (47.6%) was more equal for ING.
Other control variables
Participant demographic control variables such as gender, age, education and place of
residency are collected. Furthermore, Qualtrics gathers some relevant control variables automatically one of which is the time spent on the experiment by the participant.
