Ad believability in comparative advertising
“Believability in comparative advertising and the moderating influence of
perceived product risk and brand preference for the comparative brand”
Floris Hendrik Wubs
Master Thesis: Marketing Management
Ad believability in comparative advertising
“Believability in comparative advertising and the moderating influence of
perceived product risk and brand preference for the comparative brand”
June 2012
Floris Hendrik Wubs Kerklaan 2 9717 HE Groningen f_wubs@hotmail.com
Phone: 0622637554 Student: 1551086
University of Groningen, Marketing MscBA Marketing management
Management summary
This study investigates advertising believability within a comparative advertising format. To consumers, ad believability is generally perceived to be lower in comparative ads than in noncomparative ads. Research suggests that ad believability in comparative advertising is essential in persuading customers, and improving brand attitudes and purchase intentions. This study investigates how ad believability can be improved, and what the effects are. Also, perceived product risk, and a brand preference for the comparison brand are included as moderators in the predictive model for brand attitudes and purchase intentions.
After selecting two product categories (toothpaste as low risk product and tablets as high risk product) in a pretest, that are suitable for this study, an online survey accompanied with an ad is used to gain respondents data. As is demonstrated in literature, ad believability is of major importance in comparative advertising. Especially the influence on brand attitudes is considerable. The influence of ad believability on the dependent variables is also contingent to perceived product risk. Under conditions of high perceived product risk, the influence of ad believability is significantly stronger, and under conditions of low perceived product risk, the influence of ad believability is weaker. Subsequently, a significant moderating influence of a brand preference for the comparison brand is not established. Future research can investigate the possibility for this moderator and other possible moderators.
Preface
Writing the thesis to finish the study is the ultimate challenge for a student. For me, it has been a grateful period, in which I felt freedom to investigate a topic of choice on the one hand, and a pressure to deliver an attractive thesis on the other hand.
This gives me the opportunity to thank my supervisor, dr. K.J. Alsem, who has supported me in writing my thesis, triggered my creativity in designing the study, and critically assessed my documents. The feedback I received was always presented in a helpful, open-minded manner, which I have experienced as a pleasant support style.
Also, I want to thank family and friends for support, enthusiasm and interest for the progress of my thesis. Special thanks for the friends and co-students, with whom I spend long hours, days, weeks at the library. By writing our theses simultaneously, we kept each other motivated and inspired, and we could always count on each others help when the writing process slowed down.
Table of contents
1. Introduction………. 7
2. Theoretical Framework..……… 10
2.1 Explanation of comparative advertising 10
2.2 Processing mode in comparative advertisements 12
2.3 Believability in comparative advertising 13
2.4 Fostering advertising believability 17
2.5 The dependent variables influenced by believability 20
2.6 Moderating effect of perceived product risk 21
2.7 Moderating effect of brand preference for comparison brand 23
2.8 Literature summary and conceptual model 24
3. Methodology………. 27
3.1 Research question and hypotheses 27
3.2 Pretest determining product categories 28
3.3 Research design of main study 30
3.4 Measurement 34
3.5 Manipulation checks 39
3.6 Sample and construct descriptive statistics 40
4. Results……… 42
4.1 Main descriptive statistics 42
4.2 Results of main hypothesis 44
4.3 Results concerning moderating effects 45
4.4 The influence of brand preference on ad believability 48
4.5 Manipulation check 49
5. Conclusions and recommendations……… 50
5.1 Discussion of results 50
5.2 Conclusions 52
5.3 Managerial implications 54
5.4 Limitations and future research 56
Overview of tables and figures
List of tables
Table 1: Relative effects of comparative advertising versus noncomparative advertising 14 Table 2: Contingency model of message sidedness and competitiveness of claims 16
Table 3: Pretest results 30
Table 4: Overview of manipulation across the four groups 31
Table 5: Elements manipulating believability 32
Table 6: Origins of the used questions in the questionnaire 35
Table 7: Basic linear regression of brand attitude 44
Table 8: Basic linear regression model of purchase intentions 45 Table 9: Linear regression model including moderating effects 46 Table 10: Linear regression model of purchase intentions including moderators 47 Table 11: Revised linear regression model of purchase intentions 48
Table 12: Summary of hypotheses of this study 54
List of figures
Figure 1: Conceptual model 26
Figure 2: Depiction of the four ads used in the surveys 33
Figure 3: Distribution of gender among the four groups 40
Figure 4: Distribution of age 40
Figure 5: Distribution of age among the four groups 40
Figure 6: means of perceived risk of tablet and toothpaste groups 42 Figure 7: means of ad believability, brand attitudes and purchase intentions 43
Figure 8: Means of constructs of four groups 43
1. Introduction
Companies often use advertisements to create brand awareness and to inform people about the quality of the product advertised (Dröge, 1989). Traditionally, only the advertised brand is shown and supported with one-sided messages. However, the last decades, advertising formats have become very diverse for different products and brands. Comparative advertising is a form of advertisement that has become increasingly attractive for companies. Wilkie and Farris (1975) defined comparative advertisement as: “comparative advertising compares two or more brands of the same product or service class and makes a comparison in terms of one or more product service attributes”. Distinctive to this type of advertising is that it includes the brand name or product of a competitor in the advertisement. The increase of popularity for this form of advertising is exemplified in the United States, where the share of comparative advertisements was negligible before 1970, but twenty years later, the share of comparative advertisements had grown to 40% (Muehling, Stoltman, and Grossbart, 1990). A recent content analysis of television ads (Pechman and Stewart, 1990) found that approximately 80% of the advertisements contained comparative claims. However, 75% of the comparative advertisements contained indirect advertising claims (not mentioning the brand name of a competitor), and 25% of the comparative advertisements contained direct comparative claims (mentioning the brand name of a competitor in the ad). So it is evident that comparative advertising has become increasingly popular, but a distinction must be made between direct and indirect comparative advertising, explicitly mentioning competitors in the comparative claims, or not. This study focuses on direct comparative advertising, which is explained in the following chapter.
More countries are making the legislation around comparative advertising more flexible, offering new chances for companies. For example, in Germany comparative advertisement has recently been permitted, creating opportunities for certain companies in a large market (Schwaiger, Rennhak, Taylor, and Cannon, 2007). The European Union is reconciling laws of the individual states, including legislation around media and advertising. Comparative advertising is permitted in the Netherlands, under some conditions, though the use is more common in the United States. Further relaxation of the legal system of the European Union makes research of comparative advertisement even more relevant.
even though other studies show negative results (Dröge and Darmon, 1987; Swinyard, 1981). In these studies however, differences in success are mainly imputed to cultural differences. Although CA is often successfully used to position a product in a market, literature is inconclusive about the actual effectiveness of CA. For instance, Rogers and Williams (1989) investigated 104 studies on CA, and found that 17 studies concluded CA to be more effective than noncomparative advertising, 30 studies concluded CA not to be more effective than noncomparative advertising, and 57 studies had neutral results. Because literature is very inconclusive about its effectiveness, it is required to pinpoint situations when CA is advantageous. The various results found in studies on CA indicate varying effectiveness due to different variables. Variables that have been found to influence the effectiveness of CA are: differences in cultures (Schwaiger, Rennhak, Taylor, and Cannon, 2007; Grewal and Kavanoor, 1997; Donthu, 1998), different product types (Putrevu and Lord, 1994), product category involvement (Schwaiger et al, 2007), preexisting brand knowledge (Dröge, 1989), product experience (Schwaiger et al, 2007), degree of comparing directly (Miniard, Barone, Rose, and Manning, 2006), comparative intensity (Donthu, 1992), psychographic variables (Dutta-Bergman, 2006), and other variables.
More interestingly, Grewal and Kavanoor (1997) developed a hierarchical model that indicates what cognitive, affective and conative variables (on the basis of Lavidge and Steiner’s advertising objectives, 1961) generally are more positively influenced, in comparison with noncomparative advertising. This hierarchical model provides an excellent overview of strengths and weaknesses of CA versus NA.
attitude toward the brand and purchase intentions, CA can become a better tool for advertisers. This is the issue that is investigated in this study.
CA has been intensively studied (especially in the United States) since the seventies of the previous century, and this has led to many insights. Also, believability of advertisements is a well-investigated topic. Numerous studies identified that believability has a different effect in CA than in NA. This study, however, emphasizes exclusively on the influence of believability in comparative advertising on attitudinal and conative effects (based on Lavidge and Steiner’ advertising objectives of 1961), and investigates the moderating impact of product risk and brand preference for the comparison brand. Separately, product risk and ad believability are two well-researched topics, but the moderating influence of product risk on the effectiveness of ad believability in the setting of comparative advertising, is unique. Moreover, brand preference for the comparison brand is a topic that has not been investigated in relation with effectiveness of believability in comparative advertising. Hence, the moderating influence of this variable is also unique in literature, at this time. In conclusion, identifying the mutual relations of the variables in this study can shed an interesting light upon the current knowledge of the topics ‘comparative advertising’ and ‘ad believability’. Furthermore, this study specifically investigates direct comparison advertising, where in most studies no distinction is made between direct and indirect comparative advertising. The results can aid companies in designing comparative advertisements more optimally, according to their advertising objectives. The research question for this study is:
What is the influence of the believability of comparative advertising on consumers’ attitudes toward the promoted brand and purchase intentions, and what is the influence of perceived product risk and a brand preference for the comparative brand on this relation?
2.
Theoretical Framework
In most studies on comparative advertising, effects are investigated in comparison with noncomparative advertisement. The emphasis is on differences, advantages and disadvantages of CA (comparative advertising), with respect to NA (noncomparative advertising). This study focuses on effects of believability in CA, and its influence on attitudes and behavior. So, relations of believability on classical advertising objectives (attitude toward the brand, purchase intentions) are studied, within comparative advertising. Firstly, literature on CA is discussed. Subsequently, literature on believability and its relation with other variables are discussed.
2.1 Explanation of comparative advertising 2.1.1 Definition of comparative advertising
Due to the increased popularity of comparative advertisement, a large number of studies investigated the effects of CA. Because CA is mainly investigated in comparison to traditional, noncomparative advertising (NA), it is often defined in a way that distinguishes it from traditional advertising. McDougall defined CA in 1978 as “any advertisement that compares, implicitly or explicitly, two or more products and states or implies that information has been obtained, or a test has been conducted on a comparative basis, or that states or implies a particular market standing in relation to other similar products, whether the other products are named or not”. In this early definition on CA, no distinction is made between explicitly naming the competitor in the ad (direct comparative), or implicitly comparing to competitors (indirect comparative). However these days, companies always claim to be better at certain aspects then competition in advertisements. So to some extend, it is inherent in advertising that the advertiser states that their brand is better (than competition). The interesting aspect of comparative advertisement emanates from naming competitors (direct comparative) creating a reference brand or product, involving more complex information processing strategies for consumers. Because by naming competitors, it activates people to process the ad in a different way, involving both attitudes about the comparison brand as well as the sponsoring (Dröge, 1989). Priester, Godek, Nayakankuppum and Park (2004) also confirm that the complex information processing due to explicit brand comparison, can lead to strong attitudes toward the sponsoring brand. And naming a high-share brand can attract attention to the entire ad, which can contribute to an increase in purchase intentions (Pechmann, and Stewart, 1991).
The sponsoring brand is the brand being advertised in the ad. The sponsoring brand is trying to fill a gap between how consumers currently think about the brand, and how the brand is in reference to competitive brands. This brand is often normally seen as an inferior brand, but due to the ad, the brand should seem to be equal or superior to the comparison brand. For example, a chocolate manufacturer X claims its chocolate is at least as tasty as the chocolate of Milka (well-known chocolate manufacturer), but much cheaper. Milka is the comparison brand, manufacturer X is the sponsoring brand, and the comparison (as tasty, but cheaper) should persuade consumers that consuming manufacturer X’s chocolate is a wiser choice.
For this research, the definition of Miniard, Barone, Rose and Manning (2006) is used. It is based on the well-accepted definition of Wilkie and Farris (1975). Willie and Farris defined CA as: “comparative advertising compares two or more brands of the same product or service class and makes a comparison in terms of one or more product or service attributes”. Miniard et al, distinguishing explicit and implicit comparative advertisements, define direct (explicit) advertising as follows: “[when] specific competitors are identified explicitly as a reference point for interpreting claims about the advertised brand’s merits” (2006). This is the specific type of advertising that is investigated in this study. Due to this type of advertising, most people will think about how the sponsored brand compares to the comparison brand (Miniard et al, 2006). It allows a brand to engage the consumer to reconsider the current attitudes they have for the typical brand. This triggers people to process the advertisement more intensively, according to Petty, Cacioppo and Schumann (1983), this type of advertising is most likely being processed centrally (Priester, Godek, Nayakankuppum and Park, 2004), which is explained later on.
2.1.2 Application of comparative advertising
convenience goods), because it is more important for consumers to purchase the best suited product, leading to a higher consumers’ involvement.
2.2 Processing mode in comparative advertisements 2.2.1 Elaboration likelihood model explained
Forming of attitudes by consumers due to advertising is highly dependent on the consumer’ information processing mode. Petty and Cacioppo (1983) developed a contingency model, the elaboration likelihood model, which illustrates how processing mode contributes to forming and changing attitudes. The elaboration continuum ranges from low elaboration (little thoughts and processing) to high elaboration (high thoughts and processing). The likelihood of elaborating to a certain degree in the continuum generally indicates the route to persuasion. The central route to persuasion “views attitude change as resulting from a person’s diligent consideration of information that s/he feels is central to the true merits of a particular attitudinal position” (Petty and Cacioppo, 1983). Attitudes are thus based on thorough considerations of a person, and are postulated to be relatively enduring and predictive of behavior. Peripheral route to attitude change occurs “because the attitude issue or object is associated with positive or negative cues, or because the person makes a simple inference about the merits of the advocated position based on various simple cues in the persuasion context”. The attitude formations and changes are postulated to be relatively temporary and not predictive of behavior (Petty and Cacioppo, 1983). Elements in advertisements that are processed centrally should be designed differently than certain elements in traditional advertisement. An element that is very important for people to determine the motivation to intensely process ads is involvement. High involvement often leads to high elaboration, thus are ads most effective with the central route to persuasion.
2.2.2 Elaboration likelihood model and comparative advertising
is expected with CA, that triggers issue-relevant elaboration, which provokes central processing. Finally, central processing depends on the presence and activation of preexisting knowledge structures, and by directly using existing information about familiar brands, consumers are motivated to elaborate the presented information with their current knowledge (Droge, 1989).
2.3 Believability in comparative advertising 2.3.1 Hierarchy of comparative advertising effects
In the model of the hierarchy of comparative advertising effects of Grewal and Kavanoor (1997) , a clear overview is given on the effects of the advertising format (comparative or noncomparative), its effects (cognitive, affective and conative; Lavidge and Steiner, 1961), and the moderators of the relationship between the advertising format and the main effects. Grewal and Kavanoor state that the model is not complete, and that more complicated relations between variables can exist and should be investigated. The model gives an excellent general view of factors involved in advertising format and its effects. Moreover, it is concluded that CA can be advantageous with respect to noncomparative advertising in many situations, and that this form of advertising can be optimized. Grewal and Kavanoor summarize the following advantages. In comparison with noncomparative advertising, on average, CA generates more attention, and brand and message awareness is higher. CA also evokes more elaborate processing, and CA tends to be more informative to consumers.
Advertising function positive Negative Attention Awareness Processing Cognitive effects Informativeness Believability
Affective effects Attitude toward brand Attitude toward ad
Purchase intention
Conative effects
Purchase behavior
Table 1: Relative effects of comparative advertising versus noncomparative advertising
2.3.2 Contingencies to comparative advertising
As is mentioned before, the effectiveness of CA in comparison with NA strongly depends on circumstances in which it is applied. Apart from the in table 1 mentioned advantages and disadvantages of CA on cognitive, affective and conative effects, results of CA can be based on more causes. Research has described that the sponsoring brand and the comparison brand can be confused, characteristics of the comparison brand can be transferred to the sponsoring brand (Dröge and Darmon, 1987), or CA can be a useful tool to distinguish a brand from the comparison brand. Moreover, CA challenges current beliefs of consumers, which can lead to counterarguing. This is more likely to happen to customers who prefer the brand that is being compared (they are feeling attacked, loyalty enables defensive arguing). So in contingency terms, since the comparison brand is often the market leader, the question rises to what extend the ad should be comparing (how hostile can it be), and is it wise to apply this marketing tool in a market with a dominant market leader? On the other hand, Dröge and Darmon (1987) argue that brand attitudes toward market leaders are often more favorable, and through CA, these attitudes can be transferred. So the exact influence of comparing with a market leader is not determined, and needs to be investigated more thoroughly.
Attention is also mentioned in literature as a positive characteristic of CA in comparison to NA, because of the novelty of such an ad, and the way it stands out to other ads. However, since this form of advertising has become much more popular, the attention advantage is likely to diminish.
2.3.3 Definition of believability in advertising
purchase behavior (Swinyard, 1981). In traditional advertising, favorable attitudes and purchase intentions are more dependent upon attitudes that are formed toward the ad, and peripheral cues in the ad (elaboration likelihood model). CA is based on explicit comparisons of product or brand characteristics, and therefore, must be believed to be effective. In literature, advertising believability is described as how consumers perceive the believability of the source of an ad and the truthfulness of the message (Grewal and Kavanoor, 1997; Settle and Golden, 1974). These two elements are inseparable, and therefore believability, as a construct, is measured as one concept (Golden, 1977). A construct, like ad believability or involvement, is an abstract variable that is hard difficult to visualize, and is used for research and theory-building purposes. On the other hand, amount of sales is an objective, easily measured variable (Cooper and Schindler, 2006).
If a comparative advertisement is believed, it is a very informative way of communication for the consumer. However, less believable ads are dealing with counter arguing with the message arguments, statements about the curiosity of the claims, derogating the source of the message and other positive or negative statements about the message (Swinyard, 1981). Consequently, this leads to the formation of less favorable attitudes. The most damaging to believability of the ad is counter arguing, according to Wright (1973). And, based on the attribution theory, Swinyard argues that this can best be solved by utilizing two-sided messages (1981). This way, the ad keeps being advantageous by providing high level of information, but with good ad credibility.
Attribution theory is used as a framework for predicting consumer behavior. Attribution theory describes how consumers think about communication of advertisers, and states that these consumers’ thoughts are an important determinant in accepting or rejecting the advertising message. Consumers decide why an advertiser wants to deliver a message and whether or not the message is true (Gotlieb and Sarel, 1991). Comparative advertising messages can be seen as attempts of advertisers to sell products and create brand liking. This means that consumers discount the believability of the ad (Gotlieb and Sarel, 1991; Chow and Luk, 2006) Eagley, Wood, and Chaiken (1978) identify two types of bias that advertisers have, according to consumers’ perception; knowledge bias and reporting bias. Knowledge bias hold that the advertisers’ claim is incorrect, because the advertiser’ knowledge about the advertised issue is not complete. The reporting bias holds that the advertiser deliberately depicts an advertising claim incorrect (in favor of the advertised brand). Attribution theory can be used to help advertisers remove the bias consumers perceive in advertisements.
depicts the enhancement of believability and the differences of CA and NA in a contingency model (1981), as can be seen in table 2 below.
One-sided claims Two-sided claims Comparative claims Rich in relative price information
Low credibility
Therefore, relatively ineffective
Rich in relative price information Moderate credibility
Therefore, very effective
Noncomparative claims No relative price information Moderate credibility
Therefore, moderately effective
No relative price information Moderate credibility or better Therefore, only moderately effective Table 2: Contingency model of message sidedness and competitiveness of claims
Swinyard investigates differences of CA and NA on the basis of message sidedness, but the underlying construct is advertising believability. As believability increases in noncomparative claims, ads are still moderately effective because it lacks in informativeness to consumers. Believable comparative ads can be very effective, due to the comparison in reference to competitors, where low believable comparative ads can prove to be very ineffective.
2.3.4 Believability in comparative advertising and noncomparative advertising
In earlier models of comparative advertising (Wilkie and Farris, 1975), comparative ads were presumed to provoke more believability than NA. However, further research concluded that noncomparative ads are on average more believable than comparative ads. Grewal and Kavanoor explain that the believability is lower, because the claims in comparative ads challenge consumers’ prior beliefs (1997). Gotlieb and Sarel (1991) also argue that lack of believability is the major shortcoming of successful comparative advertising. However, if this shortcoming in the model can be offset, CA becomes an even more effective way of advertising. Thus, investigating how believability can be improved, and what the influence of believability is on brand attitudes and purchase intention is, is important for improving the knowledge about optimizing comparative ads.
Subsequently, research has shown that consumers generally form fewer favorable attitudinal responses toward the advertisement than noncomparative ads (Gorn and Weinberg, 1984; Swinyard, 1981; Putreva and Lord, 1994). Attitude toward the ad is a consumer’s feelings and overall attitude toward the ad (Grewal and Kavanoor, 1997). Consumers form less favorable attitudes toward the ad because comparative ads are more aggressive, intense, and perceived to be less honest, and considered to be an attack on other brands (Grewal and Kavanoor, 1997).
that this relationship does not necessarily hold for comparative ads. It is mentioned earlier that consumers are generally forming more positive attitudes toward the brand in comparative ads than in noncomparative ads. Droge and Darmon (1987) explain that positive attitudes that are associated with the comparison brand (often market leader) are transferred to the sponsoring brand. Also, because comparative ads are clearer and differentiate the brand on certain characteristics, consumers form more positive attitudes toward the brand than consumers do in noncomparative ads (Grewal and Kavanoor, 1997).So the forming of attitudes toward the ad cannot predict the forming of attitudes toward the sponsoring brand. In NA it is important to create an ad that enhances attitudes toward the ad, because it is then expected that brand attitudes and purchase intentions improve. In CA, attitudes toward the ad are very often negative, but brand attitudes are not directly influenced. So in short, a comparative ad can be annoying or less entertaining than a noncomparative ad, but due to the richness in information, it can be more persuasive in enhancing brand attitudes, and creating purchase intentions. For this reason, this study focuses on brand attitudes, and not on attitudes toward the advertisement. Forming of brand attitudes is considered the main dependent variable that should increase by improved ad believability. Purchase intentions, which result from positive brand attitudes, are also investigated and expected to be subject to the positive influence of ad believability. It is plausible that attitudes toward the ad can still be of influence in this theoretical model, but in the scope of this study, it is not incorporated. Not including attitudes toward the advertisement in studies of comparative advertising is in line with earlier research (Grewal and Kavanoor, 1997).
2.4 Fostering advertising believability
The effects of low believability of an ad are that people derogate the source of the message of the ad, people counter argue with the message claims, make statements about the curiosity of the claims, or make other negative statements about the message. These effects are likely to influence consumer’s attitudes about the brand, and their purchase intentions (because the persuasiveness is negated by its lacking credibility). Since the believability is at stake in CA, increasing it should prove to be a fruitful tactic.
2.4.1 Message-sidedness in comparative advertising
print ads. A CAI of 0 means that the ad is noncomparative. The intensity rises 1 level, (1) when the advertisement explicitly names the comparison brand, or in other literature defined as direct comparative advertisement (Miniard et al, 2006). The intensity raises 1 level, (2) when the ad makes specific comparisons on the basis of product attributes (not general claims of superiority, as is common in traditional advertising). Furthermore, the CAI increases (3) when the ad utilizes one-sides messages (not two-sided messages, illustrating both positive and negative differences between brands). Finally, the CAI increases (4) if it spends more than 50% of the time comparing the brands (this only holds for television ads).
The CAI theory confirms that CA can perform better than noncomparative advertising on certain dimensions, or perform worse when the comparisons made in the ad are too intense, measured on the CAI scale. However, the theory also contains several elements that have already been discussed in literature. In the definition of CA, it is captured that this research only employs direct comparative advertisements (thus according to CAI theory, level +1 for comparative ad intensity). Since television ads are not used in this study, that level becomes overdue. So, one-sided or two-sided, and specific or generic comparison will determine the CAI. The previous research determined that CA are most effective just one or two levels beneath the most intensive ads (Donthu, 1992) and that double-sided messages are most important in increasing the believability, especially with high cognitive elaboration (Chow and Luk, 2006). So in an ad where high cognitive elaboration is expected, using the most intense comparative advertisement except employing two-sided messages instead of one-sided messages, will lead to optimal results, as is confirmed by other studies.
According to research of Barrio-Garcia and Luque-Martinez (2003), when the CAI is higher, the ads tend to be more informative, but the believability tends to decrease. Because Donthu (1992) and Chow and Luk (2006) conclude that a very high level of CAI decrease positive attitudes toward the ad, ads should be designed to have a moderate to high comparative intensity. Moreover, arguing on the basis of the attribution theory, utilizing two-sided messages will lead to the consumer’ believe the advertiser is informing the consumer, instead of selling a product (Iyer, 1988). Attributing the intentions of the advertiser as informing, increases the believability of the ad. This concludes that two-sided messages should be utilized to attain better ad believability.
2.4.2 Establishing credible sources
Gotlieb and Sarel argue in their study that ad believability is a critical factor in forming favorable brand attitudes (1991). Based on the attribution theory, source credibility (the second element in increasing believability according to Grewal and Kavanoor, 1997) must be high for consumers in comparative advertising (Golden, 1979).
Gotlieb and Sarel suggest that perceived expertise and trustworthiness are the underlying dimensions of source credibility (1991). Perceived expertise is whether a consumer thinks the source is a knowledgeable person (source as advertiser, endorser, or research agency). Trustworthiness is whether the consumer thinks the source’s opinions are unbiased. For example, the message of an average consumer may be trustworthy (he is objectively telling the truth), but the expertness can be very low. The trustworthiness of a researcher of the advertising company may be perceived as an expert, but his trustworthiness is low (because he can be biased in telling the truth). So establishing credible sources, by using expert and trustworthy sources (for instance a research agency with well-known credentials, or using objective measurements), can increase the believability of the ad. Eagley, Wood, and Chaiken suggest that information from highly credible sources is perceived as providing a more accurate perception of reality (1978).
2.4.3 Substantiating the claim
Golden (1979) investigated the influence of substantiating a comparative advertising claim, with independent test results, on believability. The results of the study are not unequivocal, but they suggest that believability increases, if the claims in an advertisement are supported by independent test results, rather than being unsupported by test results. It seems acceptable that substantiating claims take away counter arguing, and bolster the believability. Grewal et al (1997) also suggest that this is a factor that determines believability of an advertisement. Therefore, in an effort to increase ad believability, advertising claims are substantiated.
2.4.4 Attending to the message content
Edell and Stalin (1983), and Iyer (1988) argue that attending to the content of the message, by comparing information with factual information instead of evaluative information, contributes to the believability of an ad. Factual information, compared to evaluative information, is objective in nature, and therefore elicits fewer counterarguments (Edell and Staelin, 1983). Evaluative information, or subjective information, can be questioned by consumers because of its nature.
previously described characteristics of CA (activating information searching strategies) suggest that this is better suited. But in this study, all comparisons are drawn upon factual information. So the contribution of this effort is not incorporated in the study design. Grewal and Kavanoor (1997) investigate the use of both factual and evaluative information, but the results are not conclusive. 2.5 The dependent variables influenced by believability
The variables investigated in this study that are influenced by increased believability in comparative advertisements are (1) attitude toward the brand, and (2) purchase intention, based on Grewal and Kavanoor (1997). The main point of interest is brand attitude, because this is formed as a direct result of advertisements, and is directly applicable in the model. Purchase intentions, as predictor of purchasing behavior, are also influenced by prior formed brand attitudes, and can be less predictable upon the identified independent variables. However, first the influence of ad believability and the moderators on purchase intentions is investigated in the same model as brand attitudes.
2.5.1 Effect of believability on attitude toward the brand
Creating favorable brand attitudes is a purpose of many advertisements. If the believability of an ad is higher, counter arguing and negative statements are less likely being formed. Fostering believability in a comparative advertisement is based on creating less negative thoughts, diminishing counter arguing, and the ads tend to be less aggressive. Consumers form less source derogating thoughts, and trust the message of the ad more. When this is accomplished, consumers are open to the information in the ad, which is relatively persuasive. As a result, the probability of forming positive attitudes toward the sponsoring brand is increased. Hence, it is expected that with an ad with high believability, brand attitudes for the sponsoring brand will increase as well.
2.5.2 Effect of believability on purchase intentions
behavior or purchase intentions, is whether the persuasive information that is being presented, is actually believable or not. It is very likely that persuasive information that is totally unbelievable, is being discarded, counter argued, and no purchasing intentions are being made as a result of it. However, persuasive ad claims that are perceived to be believable, will increase the probability of consumers forming purchase intentions. Hence, it is expected that high advertisement believability will lead to increased purchase intentions. However, brand intentions are an important antecedent of purchase intentions (Grewal and Kavanoor, 1997), and therefore, purchase intentions can be hard to predict, solely based on ad believability and moderators.
2.6 Moderating effect of perceived product risk 2.6.1 Definition of product risk
2.6.2 Moderating effect of product risk on attitude formation and purchase behavior
When consumers perceive a product to comprise high risk, they are more careful in their considerations in buying a product or not. Also, an ad promoting a high risk product, is being judged more critically, because it is more important for customers to prevent them from making wrong decisions. In other words, the consumer is more involved with the ad, and judges the ad more critically and centrally (Petty & Cacioppo, 1983). The consumer is less influenced by peripheral cues, but on the other hand he is more critically assessing the arguments and truthfulness of the ad. A low risk product involves little risks to the consumer, so the result of bad choices is diminished, which leads to less ad involvement. In this case, the consumer is more sensitive to peripheral cues than the quality of arguments or believability of the ad (Petty & Cacioppo, 1983).
In an effort to convince consumers of a lower product risk (so to diminish the perceived risk for a consumer), advertising can be an effective method. Endorsers are often being used to lower the perceived risk. It is generally acclaimed that different types of risks need different types of endorsers. Friedman and Friedman (1979) argue that perceived psychosocial risks are best reduced by celebrity endorsers, and that financial, performance (and physical risks) should be reduced by using expert endorsers. Expert endorsers draw heavily on the underlying dimension source credibility. So, the goal of using expert endorsers is to increase the believability, when normally, the financial and performance risk to consumers is high. This conclusion suggests that products that contain great financial and performance risks to consumers, reducing the perceived risk is optimally attained by boosting the believability of advertising. Thus a high ad believability, that reduces consumers’ distrust and perceived risk, leads to increased liking and formation of favorable attitudes. On the other hand, when financial and performance risk is high, but the ad believability is low (so the advertiser apparently does not make an effort in lowering the perceived risk for consumers), consumers will most likely distrust the product, and not form favorable attitudes.
This line of reasoning is very acceptable when it is construed in examples of products that are advertised. For example, a (financially and performance) high risk television, that is advertised as being a better bargain in terms of price and performance without substantiating the claim (low ad believability), is unlikely to convince consumers of its superior value. And because of the high risks of the product, it is unlikely that consumers will consequently form favorable attitudes. On the other hand, if the advertisement takes high effort in substantiating the believability of the ad, to reduce perceived risks, it is likely that consumers’ beliefs and attitudes are influenced positively.
good in a supermarket), high ad believability is presumably not significantly influencing consumers’ attitudes and beliefs toward the brand. If the product is already perceived to contain little risk, it is not likely to be important to substantiate advertisement claims and foster believability.
2.7 Moderating effect of brand preference for comparison brand 2.7.1 Advertising believability and brand preference for comparison brand
Grewal and Kavanoor explain that the believability is lower, because the claims in comparative ads challenge consumers’ prior beliefs (1997). This especially affects users of the brand that is being compared (Swinyard, 1981). Comparative ads normally compare a brand to the market leader with the largest share of consumers. If these ads are less believable than noncomparative ads, and especially for consumers who prefer the compared brand, then this is an important shortcoming to CA. A large proportion of the consumers that are reached (consumers of the market leader), might be addressed in a negative way. So, consumers who prefer the comparison brand are expected to resent a comparative ad stronger. This relation is expected in literature, but has not been tested (Grewal and Kavanoor, 1997; Swinyard, 1981).
A conclusion to these questions might be that a balance is needed in the relative market size of the comparison brand; a too large or too small market share may negatively affect advertising effects. Swinyard (1981) describes that consumers counterargue with advertising claims, especially if they feel that ‘their’ brand is being attacked, and if the ad seems hostile. Counterarguements can best be reduced by employing two-sided messages or substantiating advertising claims. So for the group of customers where heavy counterarguing is expected (consumers who prefer the comparison brand), the effect of enhancing ad credibility might be stronger. Furthermore, even though it is established that brand preference influences several processes induced by the ad, this study seeks if there is still a significant difference in brand attitudes and purchase intentions for consumers who prefer the comparison brand and consumers who do not prefer the comparison brand, given the perceived ad believability. It is expected that brand attitudes and purchase intentions are lower for people who prefer the comparison brand. Moreover, these consumers are stiffer in changing their existing views about their brand, and the advertised brand, due to brand loyalty and product knowledge. Therefore, even if ad believability is high for these consumers, the influence on the formation of favorable brand attitudes and purchase intentions is expected to be smaller, than for consumers not preferring the comparison brand. So in short, the effects of ad believability are greater for people not preferring the comparison brand, than for people preferring the comparison brand.
This means that many relations have to be measured, because it can be important in determining whether it is always sensible to compare a product with the market leader. Positive effects (more attention, more elaboration) may be offset, when believability of the ad is low, if a considerable large proportion of the consumers resents the ad. Identifying these relations might increase the knowledge of how and when to target the market leader in comparative advertising.
2.8 Literature summary and conceptual model 2.8.1 Literature summary
In this literature review it is described that comparative advertising can have several advantages compared to noncomparative advertising. Because comparative advertisements are generally processed centrally, more intense brand attitudes are being formed, and consumer behavior is more predictable. This is a goal for many advertisers, and therefore it is important to investigate optimal conditions for this type of advertising.
very low believability. Increasing ad believability should result in less source derogation thoughts of consumers, less counter arguing, and the forming of less negative thoughts. Consequently, it can be measured whether believability has a direct influence on affective and conative consumer responses. It is expected that brand attitudes are more positive when believability is greater. Also, purchase intentions are expected to increase with a higher believability. Product risk (specifically financial and performance risk) is expected to moderate these relationships. Under condition of high perceived product risk, the influence of believability is extra important in forming favorable attitudes and forming purchase intentions. Under condition of low perceived product risk, believability is less important in forming favorable attitudes and purchase intentions. Furthermore, brand preference is identified as a variable that can influence the relationship of ad believability and the dependent variables. Consumers who prefer the brand that is used as a comparison are expected to be more critical and rejecting. Believability is more important for this group of consumers, where it is expected that ads with low believability have increased negative ratings. The difference in brand attitudes and purchase intentions is expected to be smaller for people preferring the comparison brand or not preferring the comparison brand, with high ad believability.
2.8.2 Conceptual model explained
Figure 1: Conceptual model
Believability in CA
Attitude toward brand
Purchase intentions
Effects of advertising believability in comparative advertisement
3.
Methodology
In the methodology part, the causal method to test the hypotheses is described and the methods for calculating the results are explained. First, the hypotheses accompanied with the research question are described.
3.1 Research question and hypotheses
This study is designed to give an answer to the research question, formulated in the introduction: What is the influence of the believability of comparative advertising on consumers’ attitudes toward the promoted brand and purchase intentions, and what is the influence of perceived product risk and a brand preference for the comparative brand on this relation?
Research on believability in CA is essential in improving this advertising format. Based on literature, believability is associated positively with attitude toward the brand and purchase intentions. These two dependent variables do not form part of the same model in this research. Therefore, the hypotheses are measured for both DV’s (resulting in double hypotheses for H1, H2 and H3). All hypotheses used are explanatory hypotheses, which means that the existence of or a change in one variable (IV) causes or leads to a change in the other variable (DV) (Cooper and Schindler, 2006). The hypotheses concerning believability and DV’s are:
Hypothesis 1a: Believability in CA positively influences attitude toward the brand. Hypothesis 1b: Believability in CA positively influences purchase intentions.
The main effects of believability are expected, but moderating variables deliver interesting additions to current knowledge on believability in CA. A moderating variable affects the direction and/or strength of the relation between independent and dependent variables. In this study, it is proposed that perceived product risk strengthens the relationship of advertising believability and the dependent variables. On the basis of earlier research, it is expected that the believability of ads is a more important factor for the dependent variables, when product risk is high. On the other hand, when product risk is low, believability is expected to be of less crucial importance. The following hypotheses are formulated:
Hypothesis 2a: The influence of believability on attitude toward the brand is strengthened by perceived
Hypothesis 2b: The influence of believability on purchase intentions is strengthened by perceived product
risk.
Another moderator in this study is brand preference of consumers. In the current literature, the influence of brand preference on the relationship of believability and the dependent variables has not been determined. It is expected that believability is more important for consumers who do not prefer the comparison brand, for forming favorable brand attitudes and subsequently purchase intentions. The following hypotheses describe the relationships:
Hypothesis 3a: The influence of believability on attitude toward the brand is stronger for consumers who
do not prefer the comparison brand, and weaker for consumers who prefer the comparison brand.
Hypothesis 3b: The influence of believability purchase intentions is stronger for consumers who do not
prefer the comparison brand, and weaker for consumers who prefer the comparison brand.
As was described, it is likely that ad believability is generally lower for consumers who prefer the brand that is being compared. Because the comparison brand is often the market leader, it is of great importance that behavior of this large group of consumers is investigated thoroughly. A strong negative relation can prove to be an important finding in CA literature. The following hypothesis is formulated:
Hypothesis 4: Brand preference for the comparison brand has a negative influence on ad believability.
3.2 Pretest determining product categories
and Karson (2007). The expectations are that toothpaste and shampoo (as fast moving consumer goods) generally pose little performance and financial risk to customers, in contrast to new technological expensive products as tablets and cell phones. These products are suitable because the respondents in the sample are expected to use such a product, consider buying it, or know friends who use the product. For example, a product like cigarettes is less suitable, because a majority of the respondents does not want to buy it. Other reasons for products to be less suitable are that the products are intended for a certain gender or age.
A pretest deducted by 10 respondents can support the chosen product categories. Toothpaste and shampoo have been selected as low risk products. Involvement is expected to be of average level (everyone needs personal hygiene), and brand awareness is expected to be quite broad, due to advertising.
Mobile phones and tablet computers have been selected as high risk products, because these products pose potential financial and performance risks. People are expected to be involved, because they own a product or might be interested in buying such a product. Brand awareness is also expected to be high, due to heavy advertising in these product categories (this especially accounts mobile phones).
Product risk is assessed by four questions per product category, on a 5-point likert scale (1 means low risk, 5 means high risk). As can be seen in table 3, the Cronbach’s alpha is four all products is approximately 0,6-0,7, which means the constructs have been measured reasonable reliable. Cronbach’s alpha for the 2 involvement questions is good, except for mobile phones (0,449), which is too low. Toothpaste and shampoo are regarded as very low risk products (means of 1,63 and 1,78 respectively), mobile phones are regarded as moderate risky (3,25) and tablet as highly risky (4,28). Furthermore, the involvement for all products is of a moderate level (between 2,85-3,7), except for the involvement in mobile phones (4,55) which is very high. Brand awareness is of an acceptable level, although respondents mentioned the least tablet brands (2,7 on average), which is explained due to the fact that this market has only few well-known brands. Moreover, in every product category, one product was mentioned by most of the respondents, which is useful in selecting a comparison brand.
Currently, there are not many owners of tablets in the Netherlands, but the market is growing rapidly, and competition is high. The figures for toothpaste and shampoo are very similar. However, toothpaste is selected for two reasons: (1) toothpaste is the same for both genders, where shampoos often focus on men or women, and (2) toothpaste is more suitable for advertising on product characteristics, as is exemplified by the well-known ad of Oral-B (see appendix C-5).
Product Risk Involvement Brand awareness
Cronbach α Mean Cronbach α Mean # brands Most mentioned brand
Toothpaste 0,591 1,63 0,816 3,70 4,8 8 out of 10 (Oral-B)
Shampoo 0,604 1,78 0,845 3,20 4,3 9 out of 10 (Andrelon)
Mobile Phone 0,621 3,25 0,449 4,55 6 10 out of 10 (Apple)
Tablet 0,701 4,28 0,850 2,85 2,7 10 out of 10 (Apple) Table 3: Pretest results
3.3 Research design of main study 3.3.1 Research format
The hypotheses formulated in paragraph 3.1 are tested in a 2x2 between participants factorial design. This means that 2 (high believability and low believability) x 2 (high product risk and low product risk) advertisements are used. The high believability ads contain two-sided messages, substantiate the claim, and utilize credible sources. The low believability ads contain one-sided messages, do not substantiate the claim and do not utilize credible sources. In the survey, this is an independent variable. Its effect on the construct ‘ad believability’ is mainly used to determine the positive/negative influence the variable has on ad believability. The statistical models however, are based on how respondents rate the construct ad believability. And the dependent variables (brand attitude and purchase intentions) are explained by respondents’ ad believability ratings.
The other independent variable in the survey, product risk, is manipulated by using two different products, divided over the participants’ groups. On the basis of the results of the pretest, toothpaste is chosen as the low risk product and tablets are chosen as the high risk product category.
3.3.2 Data collection
are caused by manipulations in the four versions of the ads. And since differences in the ads are minimized, except for the deliberate manipulations, the variances of the results are directly attributed to the preselected IV’s. The division of groups and independent variables of the survey can be seen in table 4 below.
High product risk Low product risk High believability Group 1 Group 3
Low believability Group 2 Group 4
Table 4: Overview of manipulation across the four groups
The sample used is primarily based on students of the University of Groningen and Hanze University Groningen. The use of students for this type of study is very common. Students of Groningen are relatively easy to reach, their product category knowledge is of a fair level, they are generally interested, are familiar with this type of research, and are easily accessible through social media, university group mailings and other channels. Sudman justifies the use of students from an efficient use of limited resources point-of-view (1976). However, because students are a homogenous group in society, the external generality of the study might be reduced.
The data of 15 of the 186 respondents that filled out the survey have been deleted from the sample for various reasons:
- Respondents did not finish the survey, leaving several questions unanswered. - Age of the respondent does not fit the sample (for example, age -5).
3.3.3 Stimuli: the advertisements
The four ads that were used can be seen in appendix C. The ads contain several elements, as can be seen in table 5 and appendix A-2. There are two versions of an ad with the low risk product toothpaste, and two versions of an ad with the high risk product tablet. The two versions of the same product are identical, but differ on some elements, listed in table 5. Differences of variance between the two versions can be attributed to these elements. So within the product category, the ads are very similar. But also the ads of the other different product categories are very similar. The layout, use of colors and effects, amount of arguments, use of a main claim, use of an endorser, depiction of the price and brand name, subtext, and emphasis on believability or not is executed in the same manner, so differences are minimalized on purpose.
Small elements are different (but similar) to fit the product (background color corresponds with brand color, expert endorser is dentist or technology salesman, claim substantiation by ‘care organization’ or ‘technology forum’, and other small differences), as can be seen in table 5.
High believability (Group 1 & 3) Low believability (Group 2 & 4) Claim substantiation & source
credibility (2.4.2, 2.4.3)
1 main claim, substantiated & with credible source
1 main claim
Claim substantiation (2.4.3) 2 ad arguments, substantiated 2 advertising arguments
Message sidedness (2.4.1) 1 negative arguments 0 negative arguments
Source credibility (2.4.2) Trustworthy and expert endorser None expert endorser
Table 5: Elements manipulating believability
Below in figure 2, the believable tablet ad, used for group 1, is compared to the other three ads, and important manipulations have been circled in red. As can be seen in the top right ad, the circled items have been replaced or removed from the ad. The text on the bottom of the ad is a short summary of the arguments. The toothpaste ads are very similar, and the manipulations have been carried out in the same way.
to be a credible source for toothpaste affairs. In the survey, respondents are asked to rate the believability of the endorsers, to affirm whether or not they have been chosen appropriately.
Figure 2: Depiction of the four ads used in the surveys
3.5 Procedure
On the internet, the respondents filled out a survey that is completely depicted in the appendix (appendix D-2). The questions of the group 1 (version of the believable tablet ad) are included; the other versions are the same, with a different ad, and tablet questions have been replaced by toothpaste questions.
biased strongly, because respondents mainly choose the first or last brand on the list, the primacy and recency effect (Cooper and Schindler, 2006).
After these questions, the respondents opened the next page, and saw either a believable or unbelievable advertisement, for either toothpaste of Parodontax or a tablet of Samsung, depending on the group that the respondent is randomly assigned to. After this, respondents were asked if they have seen the advertising before (question 5), and indicated how involved they are with the product (question 6). The fifth question is asked to increase the realism of the ad, the sixth question is asked to measure the degree of involvement in the product category. Subsequently, the last six questions were asked about the sponsoring brand, concerning the following:
- Attitude toward the brand - Purchase intentions - Ad believability - Argument credibility - Endorser credibility
These constructs are measured by three or four statements, or by bipolar adjectives, on a Likert scale of 1 – 5. The Likert scale is a frequently used summated rating scale, that consists of statements that express a favorable or unfavorable attitude toward the object of interest (Cooper and Schindler, 2006). The questions used to measure these constructs can be seen in appendix D-2. As can be seen, the questionnaire used was relatively very short, which diminished the chance that respondents were confronted with fatigue, disinterest or time pressure. This contributes to the reliability of the results. The questions about argument credibility and endorser credibility are not incorporated in the conceptual model (figure 1). These constructs are measured to check the manipulations (endorser credibility and argument credibility should differ due to the made manipulations).
3.4 Measurement
3.4.1 Measurement of constructs
the survey have all been based on previous research, and are well-accepted questions for assessing people’ attitudes for these particular constructs. In table 6 can be seen on what research the questions are based.
Construct Questions derived from article
Perceived risk Grewal et al, 1994;
DelVecchio and Smith, 2005; Harmon and Coney, 1982
Involvement Laroche et al, 2010
Brand attitude Zhang, 1996
Miniard et al, 1993 Purchase intentions Miniard et al, 1993
Believability Okazaki et al, 2010
Argument believability Okazaki et al, 2010 Endorser believability Okazaki et al, 2010 Counterarguments Harmon and Coney, 1982
Table 6: Origins of the used questions in the questionnaire
Furthermore, the statistical test reliability analysis assesses the internal consistency of several questions that form one construct. The statistical test checks whether the response of several questions correlates, and gives a measure of internal consistency with a Cronbach’s Alpha. The Cronbach’s alpha is widely used in social sciences, business and other disciplines, to measure if the construct (e.g. perceived risk) is measured using the appropriate questions (Devellis, 1991). The Cronbach’s alpha is an indicator of the internal reliability, but not a hard measure that is used to reject certain items. Generally, constructs are well measured with an alpha higher than 0,8, acceptable with an alpha higher than 0,6, and doubtful with an alpha lower than 0,6. Conclusions in a study based on constructs with doubtful Cronbach’s alpha’s are not well-grounded (Malhotra, 2010). In this study, Cronbach’s alpha of higher than 0,8 is good, higher than 0,6 is accepted, and lower than 0,6 is questionable, an explanation should be found for the low Cronbach’s alpha.
objective variables. In research, constructs are generally measured by several carefully selected questions, drafted by respected researchers. In this study, the constructs are measured the same, based on questions of previous researches. The constructs involvement, ad believability, attitude toward the ad and purchase intentions, are measured on a 5 point Likert scale. In these questions, ‘5’ is coded as the positive value, and ‘1’ is coded as the negative value. All constructs also contain a question that is inversely coded (1 is the positive value and 5 is the negative value), to check for abnormalities. In processing these constructs, the negative formulated questions are recoded to be similar as the positive formulated questions. The values of the constructs are calculated by using the average of the responses to the questions (Zhang, 1996). So for example, the value of a respondent for the construct perceived risk is the average of his score on the three questions. In the statistical models, product risk and ad believability are not the nominal variables by which the respondents are influenced (different products for product risk, and high or low effort to achieve ad believability), but the interval variables, as outcome of the questions on perceived product risk and ad believability. Because perceived product risk and ad believability are subjective constructs, it is more useful to utilize the ratings of the respondents than the arbitrary nominal division of the researcher. However, it will be checked whether the nominal variables collide or agree with the interval variables. Interval scales have the power of nominal and ordinal data (for instance, allowing the researcher to check for differences in chi-square), plus the additional strength that it incorporates the concept of equality of interval (scaled distance between 1 and 2 equals the distance between 2 and 3), which allows the researcher to execute statistically more power tests (like comparing means, standard deviation, correlation, regression or analysis of variance). It is generally accepted that researchers treat any attitude scale (like brand attitudes) as interval (Cooper and Schindler, 2006).
Furthermore, an almost infinite number of extraneous variables exist, that might conceivably affect the relations of IV’s and DV’s. Most of these variables have very little effect, or the impact occurs randomly. For example, changes of peoples’ mood, or short-term changes in weather can have some effect, but occur too randomly to take into account. If all possible variables are taken into account, behavior can be predicted perfectly. However, for consumer behavior, this is impossible and unnecessary. The r square of a regression model, which is explained later on, is an indication of the percentage of explained brand attitude (as a percentage of explained variance of the DV in the statistical model, due to the IV’s). 3.4.2 Statistical tests
(Malhotra, 2010). In this study, the commonly used Pearson’s chi-square statistic is utilized.
It has been stated that companies are interested in predicted how consumers react to advertisements. Predicting how dependent variables (like brand attitude, purchase behavior) are being influenced by individual variables is important for explaining consumer behavior. However, consumer behavior is complex, and variables can interact or be influenced by other (latent) variables. Integrating numerous variables in a regression analysis enables a research to assess (a) whether a variable contributes to the current variables in explaining a dependent variable, (b) the extent to which the variable contributes in explaining a dependent variable, and (c) the extent to which the variance of the dependent variable is explained by all the variables incorporated in the model (r-square value).
Also, in comparison with simpler tests like ANOVA and T-tests, a regression analysis is superior in that it also explains whether a relation is positive or negative, and the Beta values can be used to assess the individual relative contribution of the variables. The ANOVA and T-tests can be used to assess differences between groups. ANOVA is used to test means of two or more groups, comparing the variances. The null hypothesis means that means are equal; a significance of 5% or lower means this hypothesis can be rejected. Significant relations between variables are marked with a *.
Moreover, interactions between variables can really improve the insight in consumer behavior. For example, two individual variables significantly influence a dependent variable. However, incorporating both variables into one model shows that only one of the variables is really responsible for changes In the dependent variable, diminishing the influence of the other IV. An even more complex situation arises when two IV’s influence a DV, but by combining the influence of the IV’s, the direction and/or strength of the relation with the DV is altered. In statistics, this is called an interaction effect.
In this study, the main research question is to assess the influence of ad believability on (1) brand attitude and (2) purchase intentions, considering perceived risk and brand preference. To statistically identify the relations, the variables are investigated in a linear model explaining brand attitude, and a model explaining purchase intentions. So two regressions are carried out, because a regression is limited to 1 DV.
