Queer ad for the straight eye : the extent of brand personality on the impact of out-group advertising on brand equity

(1)

Queer ad for the straight eye

T

HE EXTENT OF BRAND PERSONALITY ON THE

IMPACT OF OUT

-

GROUP ADVERTISING ON

BRAND EQUITY

MSC BUSINESS ADMINISTRATION THESIS

by Edgar Jacobson

(2)

Colophon

Author: Edgar Jacobson

Student number: 6063632

Email address: edgar.jacobson@student.uva.nl Phone number: +31 6 131 96 346

University of Amsterdam

Faculty of Economics and Business Amsterdam Business School Master Business Administration Marketing Management Track

Thesis subgroup: Brands and Consumers Course code: 6314M0220

Thesis supervisor: Dhr. Drs. Ing. A.C.J. Meulemans

Date: January 27th, 2017 Word count: 16 077

(3)

Statement of originality

This document is written by Student Edgar Jacobson who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(4)

Abstract

This study aims to examine a majority social group interacting with the ad imagery depicting a minority social group to affect customer-based brand equity in terms brand attitude and purchase intentions of the brand advertised. In addition, the extent of the moderating effect of brand personality based on gender dimensions on this relationship is researched. Hence the main objective is to find under which conditions such depictions are feasible for brands targeting the homosexual community without alienating the mainstream.

The research design experimentally manipulated advertising imagery. Results suggest that in-group ad imagery scores higher on brand attitude and purchase intentions than out-group ad imagery. We also found evidence that higher felt out-group belonging of out-out-group ad imagery to negatively affect brand attitude and purchase intentions. In addition, results suggest that masculine brands in combination with out-group ad imagery report lower scores of brand attitude, compared to combinations with the other brand personalities. Furthermore, brands with an undifferentiated brand personality exposing out-group ad imagery are significantly more likely to report higher scores of purchase intentions than other brand personalities. Implications of the findings are discussed.

Key words: Social group identity theory, ad imagery, customer-based brand equity, brand attitude,

(5)

List of Tables and Figures

Tables

Table 1: Means, Standard Deviations of Romance in Advertising Imagery. 38 Table 2: Means, Standard Deviations of Felt Group Belonging Advertising Imagery. 39 Table 3: Survey flow and treatment exposure within groups. 42

Table 4: Sample per respondent group. 46

Table 5: Descriptives Normal Distribution: Skewness & Kurtosis. 47

Table 6: Reliability Scores. 48

Table 7: Means, Standard Deviations, Correlations for In-group Ad Imagery. 49 Table 8: Means, Standard Deviations, Correlations for Out-group Ad Imagery. 50 Table 9: Means, Standard Deviations, Correlations for Control Ad Imagery. 50 Table 10: Descriptive Statistics for Brand Attitude with Statistics Test Scores 51

for Control Ad Imagery, In-group Ad Imagery and Out-group Ad Imagery

Table 11: Hierarchical Regression Model of Brand Attitude within In-group 53

Ad Imagery.

Table 12: Hierarchical Regression Model of Brand Attitude within Out-group 54

Ad Imagery.

Table 13: Descriptive Statistics for Purchase Intentions with Statistics Test Scores 55

for Control Ad Imagery, In-group Ad Imagery and Out-group Ad Imagery.

Table 14: Hierarchical Regression Model of Purchase Intentions within In-group 56

(8)

Table 15: Hierarchical Regression Model of Purchase Intentions within Out-group 57

Ad Imagery.

Table 16: Means, Stand Deviations of Brand Personality. 58 Table 17: Descriptive Statistics for Brand Attitude with Statistics Test Scores for 59

Out-group Ad Imagery * Brand Personality.

Table 18: Descriptive Statistics for Purchase Intentions with Statistics Test Scores for 61

Out-group Ad Imagery * Brand Personality.

Figures

Figure 1: Conceptual model. 30

Figure 2: In-group Advertisements. 34

Figure 3: Out-group Advertisements. 35

Figure 4: Control Advertisements. 36

Figure 5: Mean scores of Ad Imagery on Brand Attitude, moderated by 60

Brand Personality.

Figure 6: Mean scores of Ad Imagery on Purchase Intentions, moderated 62

(9)

1 Introduction

1.1 Introduction

Recently, Hallmark launched a Valentine's Day commercial featuring a gay couple. Hallmark is not the only company in doing so, as the amount of this ad imagery depicting a gay couple, of mainstream brands is increasing (Hester & Gibson, 2007). Brands such as Calvin Klein, Diesel, Banana Republic, Campbell's Soup, Amazon and Oreo have been demonstrated the use of gay depicted advertising over the past years. This is not without a good reason. In the United States alone, there are approximately 16 million gay and lesbian consumers over 18, having an expendable income of approximately $641 billion (Angelini & Bradley, 2010). Thereby, the gay and lesbian market represents the highest buying power per capita of any minority group in the United Stated (Oakenfull & Greenlee, 2005). Additionally, gay consumers have proven to be extremely brand loyal when it comes to being respectful to this population (Angelini & Bradley, 2010).

Yet despite this context, many mainstream companies remain suspicious to target gay consumers through mainstream advertisements that portray their sexual orientation (Hester & Gibson, 2007). There is much speculation about its effects on consumers, both straight and gay, which leaves marketers with a trade-off of valuing inclusion over any potential backlash from bigoted consumers. Moreover, research regarding this topic has been utterly limited and empirical proof is lacking (Oakenfull & Greenlee, 2005; Hester & Gibson, 2007; Tuten, 2005). Limited recent studies stress both positive and negative outcomes of the impact of gay themed advertisements on consumer attitudes, though it remains unclear under which conditions this relation can be altered (Bhat, Leigh & Wardlow, 1998; Tuten, 2005; Angelini & Bradley, 2010).

(10)

In an effort to fill this void, this study examines a majority social group interacting with the ad imagery depicting a minority social group to affect customer-based brand equity in terms brand attitude and purchase intentions of the brand advertised. In addition, the moderating effect of brand personality based on gender dimensions on this relationship is researched. Hence the main objective is to find under which conditions such depictions are feasible for brands targeting the homosexual community without alienating the mainstream.

1.2 Literature review

Empirical research on the topic of gay and lesbian depiction in advertising on consumer responses has been extremely absence in academic literature (Oakenfull & Greenlee, 2005; Hester & Gibson, 2007; Tuten, 2005). As a matter of fact, Oakenfull and Greenlee (2005) stated in 2005 that prior their study there have been only three published studies that empirically examined the effect of gay advertising on consumers' attitudes (Oakenfull & Greenlee, 2005). Moreover, these studies have predominantly concentrated on ad responses of the advertisement. Research regarding the impact of gay and lesbian portrayal in advertising on customer-based brand equity, and its influencing factors, remains scarce to date (Tuten, 2005; Ginder & Byun, 2015).

Findings of Bhat, Leigh and Wardlow (1996; 1999) indicate that heterosexuals have less favourable attitudes towards gay portrayal in advertising than gay consumers do. This is in line with the work of Grier and Brumbaugh (1999), that suggests that heterosexual consumers are likely to respond negatively towards gay and lesbian advertisements, viewed from an in-group perspective (Grier & Brumbaugh, 1999). Later work Oakenfull & Greenlee (2005) supports their views by suggesting that these depictions will result in an unfavourable response from mainstream consumers (Oakenfull and Greenlee, 2005). Our study draws upon research of Bhat et al. (1999), Grier & Brumbaugh (1999) and Oakenfull

(11)

and Greenlee (2005) by looking at the research gap from a social identity theory point of view, as defined by social psychologist Henri Tajfel (1974).

In addition, there is a need to look into the factors that influence the ad imagery brand equity relationship in order to draw general findings (Ginder & Byun, 2015). We propose that brand personality is an important factor that helps brands to differentiate and to build emotional aspects of a brand (Aaker & Fournier, 1995) and can be of influence on the impact of ad imagery on customer-based brand equity. Brand personality articulates the specific set of human characteristics associated with a brand. Aaker (1997) discusses brand personality based on five dimensions, namely, sincerity, excitement, competence, sophistication and ruggedness. In addition to Aaker (1997), Grohmann (2009) has articulated brand personality based on two dimensions, brand masculinity and brand femininity, resulting in four different types of brand personality. The latter seems appropriate of use for this study.

Researchers of brands and advertising often develop and test their models based on the underlying premise that the audience processing of ad content and imagery is homogeneous (McCracken, 1987; Stern, 1993). However, research suggests that individual consumers process ad information differently based on their backgrounds, experiences, motives and social group membership (Mick & Buhl, 1992; Stern, 1993; Bhat, Leigh & Wardlow., 1998). Moreover, it is proposed that the personality of brands play a crucial role in setting boundaries of what imagery a brand can depict, and therefore this urges the need for research that not only incorporates the impact on brand attitudes and purchase intentions, but also examines moderating factors on this relationship. Therefore, the aim of this study is to look into this gap and provide clarity on the topic of the gay depicted advertisements of mainstream brands, from a heterosexual social group perspective. In

(12)

other words, to what extent has brand personality an influence on out-group advertising imagery affecting brand attitude and purchase intentions of in-group consumers?

1.3 Relevance

This study answers to recent research by Um (2016) that indicated further study is needed to validate concerns over gay-themed ads appearing in mainstream media (Um, 2016). In addition, the academic relevance of this research is based on the added extension of the moderating effect of brand personality. Moreover, rather than the general demographic classification of race and gender, this study is based on self-identified sexual orientation categories. Branchik (2002) states the gay market segment will grow and develop in the next hundred years. The author states that either increasing societal acceptance of gays will result in an integration of the gay market segment into larger demographic or psychographic segments, or that the movement will suppress ability of businesses to reach out on this segment (Branchik, 2002). Therefore, it is necessary to stay up to date in conducting research that implies this acceptance on brand responses. In a similar fashion, this study contributes to managerial knowledge in a way that it provides practical clarity in how to target gay and lesbian consumers, without alienating heterosexual consumers.

1.4 Reading outline

The remainder of this study is structured as follows. Chapter two gives a theoretical overview of the framework that is used in our research. We draw upon the social identity theory of Henri Tajfel (1974), and mainly extent the research of Bhat, Leigh and Wardlow (1998), and Oakenfull Greenlee (2005), with the theory of Grohmann (2009). Our hypotheses follow directly from theory. In the third chapter, our methodology is discussed and our experimental treatment is pre-tested. Also methods of data collection and sampling are explained. The fourth chapter gives an overview of our statistical procedures, directly

(13)

followed by results. This study ends with the conclusions that provide implications for theory and practice, limitations and directions for further research.

(14)

2 Theoretical Framework

In order to understand audience differences of ad evaluation, we first discuss the social identity theory according to social psychologist Henri Tajfel (1974), and the particular definition of social groups based on sexual orientation. Further, we examine the manifestation of out-group imagery in advertising followed by its impact on customer-based brand equity in terms of brand attitude and purchase intentions. The chapter closes with the discussion of the brand personality concept, which is proposed to be an influential factor on the ad imagery brand equity relationship.

2.2 Social Groups

Membership of a social group may strongly affect the processing of the advertising content, and thereby the effect on customer-based brand equity (Bhat et al., 1998). Rather than a mere individual system of beliefs, actions, intentions and attitudes, it is being a member of a particular social group that contributes to the image one has of himself and his environment (Tajfel, 1974). In order to understand social group membership, we address four linked concepts: social categorization, social identity, social comparison, and psychological distinctiveness (Tajfel, 1974).

According to Tajfel (1974) the process of social categorization is used by individuals in order to systematise his environment, and can be understood as ordering the social environment into categories that are meaningful to the individual subject. In this respect, a social group designates a cognitive entity that is meaningful for the individual subject, rather than that it implies an objective connection between a sum of people (Tajfel, 1974). Social categorization is thus a process of bringing together social objects or events into

(15)

grouped entities which are similar based on one's system of beliefs, actions, intentions and attitudes. (Tajfel, 1974).

In this manner, social categorization forms the basis of the concept of social identity. One's individual self-concept is derived from his knowledge of his membership of the social group, and the emotional meaning attached to this membership. Thus, the social identity is constituted through the system of social categorization, by which the individual recognizes his identity in socially defined terms, and as he lives in society these defined terms become reality (Berger, 1966). Social categorization can therefore be seen as a system of orientation that defines the individual's place in society (Tajfel, 1974).

The recognition of identity into socially defined terms comes with consequences (Tajfel, 1974). According to Tajfel (1974), we can assume that individuals tend to remain a member of a group. They will look for becoming a member of new groups if these groups may add positive aspects to their social identity. If a group does not satisfy the requirement of adding positive aspects to their social identity, the individual will tend to leave the group. (Tajfel, 1974). However, leaving the group is not possible for some objective reasons, such as age, or when it conflicts with important values which are themselves part of the satisfactory social identity (Tajfel, 1974). In this case the individual can reinterpret the attributes of the group so that its unacceptable aspects become acceptable or justified, or one can undertake social action which would lead to desirable adjustments (Tajfel, 1974).

The positive attributes of social identity only gain meaning in relation to, or in comparison with, other groups. It is this social comparison that forms the link between social categorization and social identity. The meaning of one's group as a whole and its characteristics is formed in relation to the perceived differences from other groups. In this respect, a social group makes no sense unless there are other groups around (Tajfel, 1974).

(16)

Only because there are other groups around, a social group becomes a group in the way that it is perceived as having common characteristics or a common fate (Tajfel, 1974).

Along these lines psychological distinctiveness between groups is established. Because of the multi-group structure, a social group will be able to maintain its contribution to those positively valued attributes of an individual's social identity, exclusively if one manages to keep its positively valued distinctiveness from other groups (Tajfel, 1974). In that sense, social comparison between social groups is fixed on the establishment of the distinctiveness between the own and other groups (Tajfel, 1974). In this manner, we can speak of one's own in-group that has the assumed need for differentiation from the other groups as out-groups (Tajfel, 1974).

This need for distinctiveness provides the outcomes of related phenomena that exist in a big array of social situations, and on different levels (Tajfel, 1974). This means for the marketing practice that each group is likely to apply different reading strategies when seeing an advertisement, and thereby each group member develops its own shared evaluation for the brand advertised (Bhat et al., 1998). As such, research implies that there are patterns in meaning creation that is structured by social group membership (Scott, 1994; Stern, 1993). Empirical evidence suggests differences in groups based on ethnicity, gender, and sexual orientation (Faber, O'Guinn & Meyer, 1987; Leigh, Rethans & Whitney, 1987; Whittler, 1989; Stern, 1993; Bhat et al., 1998).

Regarding the context of ethnicity for example, studies have found evidence for differences in brand attitudes towards ethnic portrayal in advertisements, whereat African-Americans and Hispanic-Americans differ from white Americans in their brand evaluations (Faber, O'Guinn & Meyer, 1987). Outcomes of the study suggest that African-Americans and

(17)

white Americans, and vice versa. Moreover, both minority social groups shared the belief that the portrayal of minority characters was inadequate, while among white Americans the opposite was true (Faber et al., 1987).

Findings of another study by Whittler (1989) also indicate group differences in advertisement and product evaluation. In this study, black and white participants had to rate an advertisement depicting a white or black actor. As in line with the premise of social group membership, white participants had less favourable attitudes towards the ad imagery depicting a black actor, than to the ad depicting a white actor. His findings also suggest that viewers with strong social group beliefs were more likely to employ heuristics that focuses on group distinction, such as racially distinctive attitudes, than viewers with weaker social group beliefs (Whittler, 1989). These observations are consistent with the view of Qualls & Moore (1990) that presents evidence for the effect of race in consumers' evaluation of advertising. Their results posit that in-group members will evaluate other in-group members, depicted in ad imagery, more favourably than out-group member on the basis of race (Qualls & Moore, 1990).

These attitudinal differences among social groups can also be found with regard to gender (Mick & Politi, 1989; Stern, 1993; Wolin & Korgaonkar, 2003; Okazaki, 2007). According to Stern (1993) men and women have systematically different styles when evaluating gender related ad imagery (Stern, 1993). Men tend to evaluate the ad imagery from the "outside", whereas women as a group tend to feel and empathize with the content from the "inside" (Stern, 1993: 560). According to Wolin & Korgaonkar (2003) these differences in beliefs also results in different behaviour (2003). This is consistent with the view of Okazaki (2007), who claims that men behaving according to their sex role membership, tend to rely

(18)

on more functional aspects, as opposed to women who tend to rely more on feelingly aspects when evaluating advertisements (Okazaki, 2007).

But also within groups based on gender social groups can be defined and differences exist. Leigh, Rethans & Whitney (1987) have examined ad responses along the psychological distinctiveness of traditional versus modern women. In their study traditional and modern women were exposed to advertisements depicting verbal statements that implied traditional or modern role portrayals (Leigh et al., 1987). As in line with social group behaviour, as stated by Tajfel (1974), women evaluated the ads positively when the role portrayals matched their social in-group identity. The evaluations were negative if the depicted role portrayals were not congruent with their group membership identity (Leigh et al., 1987). Similar results are also found by Jaffe (1991), who defines social identity as women's role orientation rather than sex-role orientation (Jaffe, 1991).

So far these studies provide evidence that evaluations to the exposure of advertisements differ with social group membership. However, the social groups in these studies have been defined on broad demographic characteristics, such as ethnicity and gender. Less common is the definition of social groups in terms of sexual orientation (Bhat et al., 1998). Bhat et al. (1998) make a distinction between heterosexuals, a large majority social group, and homosexuals, a minority social group. The outcome of their study suggests that the majority social group has less favourable attitudes towards advertisements that portray imagery corresponding to the minority social group (Bhat et al., 1998). This is consistent with the premise of the social identity theory of Tajfel (1974). With regard to our study, we take on the same in-group/out-group approach as Bhat et al. (1998), and speak of heterosexuals as large mainstream social group, and of homosexuals as a small minority social group.

(19)

When looking at heterosexual consumers and homosexual consumers from an in-group/out-group perspective, a few things are important to discuss (Oakenfulll & Greenlee, 2005). Given that heterosexual imagery in advertising is mainstream, homosexuals as a minority out-group are used to these kinds of images and therefore it is unlikely their evaluations will be negative, as it simply accessed their belief system (Oakenfulll & Greenlee, 2005). Moreover, when homosexual consumers face homosexual ad imagery, thus that is consistent with the positive aspects of their social group identity, they are not only likely to evaluate the ad positively based on their group membership, but they consider their inclusion in marketplace activities as recognition of the progress of the homosexual social movement. Therefore it is possible that they may view this as a great endorsement of the support of gay social movement and are extra likely to have positive attitudes towards the brand (Oakenfull & Greenlee, 2005).

However, given the majority position of heterosexual consumers and the scarcity of homosexual imagery in mainstream advertising, the evaluations of heterosexuals as an out-group are likely to be the other way around. Seen from a social out-group perspective and in line with the findings of Bhat et al. (1998), several studies suggest that out-group heterosexual consumers are likely to take a negative stance when seeing an advertisement that is targeted towards a homosexual social group (Grier & Brumbaugh, 1999; Ricker & Witeck, 2003; Hester & Gibson, 2007).

2.3 Ad Imagery

Before going further, we first discuss what an effective advertisement looks like. Then we dive deeper into meaning of ad imagery in relation to social groups. In approaching the concept of ad imagery we continue on the basis of the social identity theory (Tajfel, 1974) and examine the meaning-based approach as stated by Grier & Brumbaugh (1999) and later

(20)

work by

Aaker, Brumbaugh & Grier (2000). Furthermore, we discuss two forms of

social group advertising imagery which are examined in this study. That is: heterosexual ad imagery, as means of imagery of a majority social group; and homosexual ad imagery, as means of imagery belonging to a minority social group.

According to Shimp (2010: 208) we can identify some generalizations about the creation of advertisements in order to become effective. Effective advertisements are usually creative, and may sufficiently motivate the audience to pay attention and engage in the processing of ad meaning on a higher order (Shimp, 2010: 210). Creative ads share three common features: connectedness, appropriateness and novelty. Connectedness addresses whether an ad provides information or creates feeling that resonate with the group members' motivations. Appropriateness means that the ad must provide information that is pertinent to the brand advertised. Novelty is the element in the ad that makes the ad unique, fresh and unexpected (Shimp, 2010: 209-210). It is the latter that provides space for the use of unconventional imagery.

Shimp (2010: 212) also stresses the use of emotional advertising that aims to reach the consumer through the use of emotional strategy, for example, depicting a couple in a romantic way. This style of advertising seems to work especially in product categories that are associated with emotions, such as the fashion apparel category, which forms the context of this thesis (Shimp, 2010: 212). Moreover, according to Shimp (2010: 245), advertisements contain certain cues that attract involuntary attention. Such cues may hold the use of novel stimuli, which can be described as unusual, unpredictable or distinctive. The use of endorsers in ad imagery allows consumers to see themselves in the ad, thus enabling cues for group identification (Grier & Brumbaugh, 1999).

(21)

We can assume that certain cues in advertisements determine to which social group the ad belongs. According to Grier & Brumbaugh (1999) the meanings consumers derive from ad imagery are dependent on the way consumers interpret the advertisement, which is as we have seen earlier, social group specific. What this means is that advertisements are created for certain groups of whom to expect they share certain in-group beliefs. The advertising message is thereby encoded to be interpreted with this group-specific system of beliefs. If group members are indeed able to decode the intended meaning of the ad with their shared belief system, which is likely, the tendency to recreate a common (the intended) meaning of the ad is more feasible. In this way, group members will search for cues contained in the ad that are consistent with the member's identity (Grier & Brumbaugh, 1999). It is thus the higher likeliness of the in-group members to recreate the intended meaning that shapes the link between the product and the consumer's identity (Grier & Brumbaugh, 1999).

In contrast, out-group members might not exhibit this system of beliefs. When exposed to an advertisement that is not created for their social group, they may not read the ad any further and may not evaluate it (Grier & Brumbaugh, 1999). In case they might consider reading the ad further, the meaning of the ad may not break through, because they lack the belief system to decode the intended meaning of the ad. Hence, the out-group consumers may derive meaning from an advertisement that is not intended by the advertiser (Grier & Brumbaugh, 1999). Furthermore, out-group members may realize the ad is not indented for them, which may make them criticize the ad because it excludes them. The ad content is rejected and the out-group members may take a negative stance towards the brand advertised. Consequently, the ad becomes ineffective (Aaker et al., 2000).

With regard to the aim of this study, we can distinguish two forms of ad imagery used in advertisements. One that belongs to heterosexual consumers, a large majority social group,

(22)

and the other belonging to homosexual consumers, a small minority group. According to Oakenfulll & Greenlee (2005), explicit heterosexual imagery in advertising can at best be represented by the depiction of a heterosexual couple. This can be seen as the emotional style of advertising as stressed by Shimp (2010), which aims to reach the consumer through affection rather than cognition (Shimp, 2010).

In a similar fashion, the most explicit type of homosexual imagery in advertising is by endorsing a homosexual couple (Oakenfulll & Greenlee, 2005; Bhat et al., 1998). Additionally, implicit cues can be provided by the use homosexual symbolism, such as a pink triangle. However, the use of only implicit cues could be considered to be unethical as it may represent prejudices (Oakenfulll & Greenlee, 2005). In this fashion, explicit cues are considered more favourable in providing social group identification grips. Hence, we can speak of heterosexual and homosexual ad imagery belonging to the respective social group (Oakenfulll & Greenlee, 2005).

2.4 Ad Imagery on Brand Attitude and Purchase Intentions

So far we have focused on ad imagery belonging to social groups within the framework of advertising and its effect on consumer evaluations in broad speaking terms. In this section we define these evaluations in terms of brand attitude and purchase intentions, which form the dependent variables in this study. Hence, we constitute the important relationships that form the basis of this research. In order to do so, we first define brand attitude as a parameter of brand equity. Second, we define purchase intentions. Lastly, the influence of advertising, within the practice of integrated marketing communications, on the dependent variables is being discussed.

(23)

34). It contains elements of a brand identity, such as a name, sign, symbol, term, design or a combination of them, which enables to identify the goods and services of one seller or a group of sellers. It also enables them to differentiate themselves from its competitors (Kotler, 1991). From a consumer perspective a brand possesses equity to the extent that people are familiar with the brand. In their memory they have stored favourable, strong and unique associations with the brand. To this we can refer to as brand knowledge (Keller, 1993). It is the differential effect of brand knowledge on customer response to the marketing of the brand that defines customer-based brand equity (Keller, 1993).

In order to build strong brands, customer-based brand equity must be established (Keller, 2001). Keller (2001) has developed a model of brand building which maps out what brand equity is and how it should be built, measured and managed. He states that the power of the brand is what resides in the mind of customer (Keller, 2001). We can define six brand building blocks on four different levels, and in order to build customer-based brand equity, all levels need to be reached by the customer. Though, a higher level can not be reached without the establishment of the lower levels. The strongest brands excel in all six of the brand building blocks and together they for the CBBE pyramid (Keller, 2001).

On the base level, brand identity, identification of the brand is ensured. Associations of the brand in the customer's mind with a certain product class or consumer need are established. Hence, salience, or brand awareness can be created (Keller, 2001). On the second level, meaning to the brand in the mind of the customer is established. We can divide brand meaning in functional terms of brand performance and more abstract terms of brand image (Keller, 2001). On the third level proper customer responses to this brand identity and brand meaning should be elicited. These responses consist of judgements and feelings. The aim is to convert these responses to create an intense, active loyalty relationship between

(24)

the brand and the customer on the fourth level of the brand pyramid: brand resonance (Keller, 2001).

Brand attitude is formed on the third level of the CBBE pyramid in terms of response. We can distinguish these responses according to personal judgements and the feelings that are elicited towards the brand, both based on how different performance and imagery associations are put together (Keller, 2001). Brand attitude can be defined as consumers' overall evaluation of the brand (Wilkie, 1986), and thus as the function of the attributes and benefits that are associated with the brand (Keller, 1993). Brand attitudes positively influence consumer behaviour if consumers internalize or think of positive responses when confronted with the brand. In this way, the top level of resonance can be reached that has to do with brand loyalty, attitudinal attachment, a sense of community and active engagement (Keller, 2001). Thus, it is an important step for firms to generate positive brand attitudes (Keller, 2001).

Brand attitudes indicate how consumers respond to the brand, its marketing activities, and other sources of information (Keller, 2001). Mass media advertising can be seen as the primary form of integrated marketing communications that puts together performance and imagery associations (Shimp, 2010: 39). This happens by borrowing associations from secondary sources of brand knowledge, such as ad imagery (Keller, 2001). Specifically through the use of endorsers advertising imagery may add associations to the brand, thus creating meaning on the second level of the brand pyramid (Keller, 2001). Therefore, in order to enhance brand equity, marketing communications programs should forge strong, favourable and unique associations with the brand in the mind of the consumer that lead to positive responses and thus positive brand attitudes on the third CBBE level (Shimp, 2010:

(25)

42). Hence, ad imagery plays an important role in enhancing positive brand attitudes (Shimp, 2010: 42).

Regarding the social group dynamics as we have seen earlier, and the response towards social group imagery in terms of brand attitudes, we can now propose that:

H1a: Differences exist between in-group and out-group ad imagery towards brand attitude: in-group ad imagery scores higher on brand attitude and out-group ad imagery scores lower on brand attitude, compared to the control ad imagery.

H1b: There is a positive relationship between the felt group belonging of the in-group ad imagery and brand attitude: higher levels of the felt group belonging of the in-group ad imagery will lead to higher levels of brand attitude.

H1c: There is a negative relationship between the felt group belonging of the out-group ad imagery and brand attitude: higher levels of the felt group belonging of the out-group ad imagery will lead to higher levels of brand attitude.

However, the eventual challenge of marketing communications is to influence behaviour, rather than merely enhancing equity (Shimp, 2010: 50). According to Shimp (2010), creating awareness and boosting brand meaning serve little positive effect, unless consumers ultimately make purchases or engage in any other form of desired behaviour. Therefore, it is important to not only examine brand attitudes, but also investigate consumers' intentions to take action (Shimp, 2010). Within the context of business, behaviour basically refers to purchasing behaviour and is a customer-based concept. We refer to purchase intentions as a consumer's conscious plan to make an effort to purchase a brand (Eagly & Chaiken, 1993: 168). In a similar way ad imagery has an effect on brand attitudes it may have an influence on purchase intentions (Shimp, 2010: 50).

(26)

Therefore, it is hypothesized that:

H2a: Differences exist between in-group and out-group ad imagery towards purchase intentions: in-group ad imagery scores higher on purchase intentions and out-group ad imagery scores lower purchase intentions, compared to the control ad imagery.

H2b: There is a positive relationship between the felt group belonging of the in-group ad imagery and purchase intentions: higher levels of the felt group belonging of the in-group ad imagery will lead to higher levels of purchase intentions.

H2c: There is a negative relationship between the felt group belonging of the out-group ad imagery and purchase intentions: higher levels of the felt group belonging of the out-group ad imagery will lead to higher levels of purchase intentions.

2.5 Brand Personality

In contrast to what we examined in the previous sections, recent work of Tuten (2005) states that, while noticing gay-friendly cues in brand communications, heterosexual consumers do not develop negative attitudes when exposed to homosexual ad imagery (Tuten, 2005). In the author's explanation this has to do with the gay-vague positioning of the brand, which left the consumers indecisive in how to perceive the brand (Tuten, 2005). However, the author failed to investigate the exact antecedents of why heterosexuals may not develop negative attitudes when exposed to homosexual ad imagery. In this study, we propose that it is the brand personality of the brand advertised that has a moderating effect on the ad imagery brand attitude and purchase intentions relationship.

Similar to ad imagery, brand personality is created on the second level of the CBBE pyramid (Shimp, 2010). On this level, brand image is based on a variety of associations that

(27)

having their own personality. In this regard we can speak of a brand's brand personality (Shimp, 2010). What is important here is that brand personality too serves as a cue for group identification. Thus, on the second level of the CBBE pyramid, perceptions of brand personality serve as meaningful cues that foster group identification (Grier & Brumbaugh, 1999).

Aaker (1997) has developed a theoretical framework of the brand personality construct in which she distinguishes five dimensions of brand personality. According to Aaker (1997) brand personality can be defined as the set of human characteristics associated with the brand (Aaker, 1997). Brand personality is formed and influenced over time by any direct or indirect touch point the consumer has with the brand and resides in the mind of the consumer (Aaker, 1997). According to the author, brands have five distinct personality dimensions: sincerity, excitement, competence, sophistication, and ruggedness that can be measured on a brand personality scale (Aaker, 1997).

Grohmann (2009) has extended the research of Aaker (1997) by defining brand personality in terms of masculine brand personality (MBP) and feminine brand personality (FMP). According to her framework brand personality can be defined along the dimensions of the gender construct. That is, masculine personality treats and feminine personality traits (Grohmann, 2009). She states that these dimensions are orthogonal, which means brands can have both masculine and feminine personality traits. As such, we can define four brand personality categories along these gender dimensions: masculine brands, feminine brands, undifferentiated brand, and androgynous brands (Grohmann, 2009).

Grohmann (2009) has created a measurement scale to define MBP and FBP. If a brand scores high on FBP and low on MBP, we can speak of a feminine brand. If a brand scores low on FBP and high on MBP, we can speak of a masculine brand. When scores are low

(28)

on both FBP and MBP, we can speak of an undifferentiated brand. When scores are high for both FBP and MBP, we can define this brand as an androgynous brand (Grohmann, 2009). For example, research of Grohmann (2009) has shown that the brand Chanel is perceived as a feminine brand. Hugo Boss tends to be seen as a masculine brand (Lieven, Grohmann,, Herrmann, Landwehr, & Tilburg, 2014). According to Lieven et al. (2014) H&M is likely to be perceived as an undifferentiated brand, as it scores low on both FBP and MBP (Lieven et al., 2014). According to Grohmann (2009) Calvin Klein scores high on both MBP and FBP, and can therefore be seen as an androgynous brand (Grohmann, 2009).

In her study, Grohmann (2009) articulates that consumers tend to have more positive responses to brands with a gender personality that is congruent with their own gender personality, and with the gender personality depicted in the ad. For example, women have more favourite responses towards FBP depictions. However, with regard to homosexuals these gender assumptions may not apply. Evidence has shown that homosexual males are more likely to have less masculine and more feminine personality traits than heterosexual men (Lippa, 2000). As such, depictions of homosexual men may not be congruent with a masculine brand personality or a feminine brand personality, resulting in less favourable evaluations. Conversely, we propose that because of this congruency aspect, brands with an undifferentiated brand personality or an androgynous brand personality may alter the ad imagery brand evaluations relationship in a positive manner. Note that in this study we refer to brand personality as a priori to ad imagery treatment.

Therefore, we propose that:

(29)

undifferentiated or androgynous brand personalities score higher on brand attitude than ad imagery combined with masculine or feminine brand personalities.

H4: Brand personality has a moderating effect on the relationship between out-group ad imagery and purchase intentions: ad imagery combined with undifferentiated or androgynous brand personalities score higher on purchase intentions than out-group ad imagery combined with masculine or feminine brand personalities.

2.6 Conceptual Model and Hypotheses

The aim of this study is to examine how brand personality alters the relationship of a majority social group interacting with the imagery depicting a minority social group to affect brand attitude and purchase intentions of the brand advertised. Hence, the research question is: "To what extent has brand personality an influence on out-group advertising imagery affecting brand attitude and purchase intentions of in-group consumers?"

(30)

Figure 1: Conceptual model.

H1a: Differences exist between in-group and out-group ad imagery towards brand attitude: in-group ad imagery scores higher on brand attitude and out-group ad imagery scores lower on brand attitude, compared to the control ad imagery.

H1b: There is a positive relationship between the felt group belonging of the in-group ad imagery and brand attitude: higher levels of the felt group belonging of the in-group ad imagery will lead to higher levels of brand attitude.

H1c: There is a negative relationship between the felt group belonging of the out-group ad imagery and brand attitude: higher levels of the felt group belonging of the out-group ad imagery will lead to higher levels of brand attitude.

H2a: Differences exist between in-group and out-group ad imagery towards purchase intentions: in-group ad imagery scores higher on purchase intentions and out-group ad imagery scores lower purchase intentions, compared to the control ad imagery.

Ad Imagery • In-group • Out-group • None (Control) Brand Personality • Masculine • Androgynous • Undifferentiated • Feminine Brand Attitude H1 H2 H3 H4 Purchase Intentions

(31)

H2b: There is a positive relationship between the felt group belonging of the in-group ad imagery and purchase intentions: higher levels of the felt group belonging of the in-group ad imagery will lead to higher levels of purchase intentions.

H2c: There is a negative relationship between the felt group belonging of the out-group ad imagery and purchase intentions: higher levels of the felt group belonging of the out-group ad imagery will lead to higher levels of purchase intentions.

H3: Brand personality has a moderating effect on the relationship between out-group ad imagery and brand attitude: out-group ad imagery combined with undifferentiated or androgynous brand personalities score higher on brand attitude than ad imagery combined with masculine or feminine brand personalities.

H4: Brand personality has a moderating effect on the relationship between out-group ad imagery and purchase intentions: ad imagery combined with undifferentiated or androgynous brand personalities score higher on purchase intentions than out-group ad imagery combined with masculine or feminine brand personalities.

(32)

3 Methodology

In order to test the hypotheses, the research design experimentally manipulated advertising imagery. In this chapter the research methodology will be described, including the undertaken steps, the design of stimulus, measurement and data collection techniques.

3.2 Stimulus Design

The starting point of the research was to find appropriate advertising imagery that could be used to represent out-group and in-group advertising imagery within the same brands. To ensure maximum realism of the potential stimuli, and thus to ensure maximum external validity, real advertising campaigns were used. Advertisements of the apparel industry were chosen for three reasons. First of all, advertisers of fashion items, such as of brands as Diesel and Calvin Klein, have a long history of targeting the homosexual community with ads depicting images in gay media (Bhat et al., 1996). Therefore, advertisements with same-sex couples depicted are more available within this category. Secondly, to control for category differences, only one product category is used: the apparel industry. Thirdly, fashion brands have been pre-tested and embedded in the literature of Grohmann (2009) and that of Lieven, Grohmann, Herrmann, Landwehr and van Tilburg (2014) in terms of brand personality based on gender constructs. The brand personality of these brands represent the moderating variable in this research, thus the research of Grohmann (2009) and Lieven et al. (2014) enabled us to pick a maximum variation in the moderating variable.

Drawing upon the ad requirements of the stimuli used in the research of Bhat et al. (1996), images depicting a couple, either male/female or male/male, in physical proximity to each other, and with gestures that imply a romantic involvement were considered. Images

(33)

depicting nudity or sexual contact were excluded from consideration. In order to eliminate confusion from respondents with unfamiliar brand names, established brands were selected from the research of Grohmann (2009) and Lieven et al. (2014). According to their studies these brands represent the maximum variance in terms of brand personality based on gender constructs. These brands include H&M that should represent an undifferentiated Brand, Calvin Klein as an androgynous brand, Chanel as a feminine brand, and Hugo Boss should represent a masculine brand. To ensure these categories to be a true fit, these brands are tested in this research on brand personality too.

In order to ensure a maximum fit in brand identity, real advertising campaigns of these brands were selected. For H&M, the holiday campaign of 2010, was used. This campaign already included an ad depicting a heterosexual couple and an ad depicting a same sex couple. The same goes for Calvin Klein. The Calvin Klein (Jeans) autumn campaign of 2015 was selected because it already included advertisements of male/female and male/male couples which meet our requirements. For both Chanel and Hugo Boss there were no same-sex couple ads available. For this, advertising imagery of other (unfamiliar) brands was used to complete the stimuli. A professional art director edited all ads to guarantee consistency of the brand identity of the brands. For example, advertising text copy was removed and logos for all ads were placed at the same place on the ad, and lightning was adjusted. This was done to enhance realism of the advertising campaign within the four brands and to control for possible visual differences.

Consequently, twelve advertisements were created. Four in-group advertisements of the different brands, depicting a male/female couple, which serve as treatment, were created. Four out-group advertisements of different brands were made, depicting a male/male couple, which also serve as treatment. Furthermore, four neutral advertisements were

(34)

created, showing only the brand logo in front of a white background, which serve as non-treatment (control). The advertisements are shown in Figure 2, Figure 3 and Figure 4.

(35)

(36)

(37)

3.3 Pre-test

When the advertising design was complete, the eight treatment advertisements were pre-tested to meet two conditions. First, it was necessary to determine romance between the depicted couples in the ads of the two treatment ad imagery groups. In other words, if the couples were perceived to have a romantic involvement depicted, we can clearly state that we are talking about a group of heterosexual ad imagery (in-group) and homosexual ad imagery (out-group). Second, in order to assess the level of social group identification with the ad imagery depicted, a second measure was used. With this measure we can numerical rank the ad imagery within the two groups in terms of felt group belonging. Put differently, if an ad is rated higher on the measurement scale, the ad has a higher belonging to the respective social group (Aaker, Brumbaugh & Grier, 2000). What is important is that the ad imagery in the pre-test was shown without the brands logos. In this way, we could prevent any possible interfering of attitudes towards the brands.

Through snowball sampling a group of 31 respondents were firstly asked with what social group they identified. They were asked to identify as "heterosexual" or "LGBTQ" (Lesbian Gay Bisexual Transgender Queer), or "prefer not to say". None of the respondents chose the option "prefer not to say". The group that identified as heterosexuals (n = 16) then were shown the ad imagery with male/female couples. The group that identified as LGBTQ (n = 15) were shown the ad imagery with male/male couples. Both groups were asked to complete a 3-item scale measuring romantic involvement of the couples seen in the ad, adapted from Bhat et al. (1998), yielding a Cronbach's α = 0.768. Measurement ranged from 1 (strongly disagree) to 7 (strongly agree) to check whether the couples in the ad were perceived romantically involved. The mean results are shown in Table 1. In all eight ads the couples were perceived to be romantically involved, as all scores reported a

(38)

mean average > 4.00. Thus, the first condition was met. From now on we can speak of the four ads with male/female couples as heterosexual ad imagery, and the four ads with male/male couples as homosexual ad imagery.

Table 1: Means, Standard Deviations of Romance in Advertising Imagery.

In-group Ad Imagery Out-group Ad Imagery

Mean SD Mean SD

H&M 5.67 .75 5.00 1.23

Calvin Klein 5.23 .97 6.29 .56

Chanel 5.40 .78 4.91 1.11

Hugo Boss 4.79 1.07 6.13 .65

Note. Values are the mean of reported scores of Romance on a 7-point scale (1 = strongly disagree, 7 = strongly agree) adapted from Bhat et al. (1998).

In order to assess each advertisement on a scale of social group belonging, respondents of each social group were then asked to complete scales of the felt group belonging of each ad. For this, an adapted 3-item measure of Aaker et al. (2000) was used (Cronbach's α = 0.90). Examples of questions included "I feel the advertisement was intended for people like me". Respondents rated the ads on items ranging from 1 (strongly disagree) to 7 (strongly agree). Again, the respondents that identified as heterosexual (n = 16) rated the ads with male/female couples depicted. The respondents who identified as LGBTQ (n = 15) rated the ads with male/male couples. As Table 2 shows, all ads can be assigned to the respective social groups, as the mean average for all ads yielded > 4.00. Thus, also the second condition is met. The images with heterosexual couples can from now on be seen as in-group ad imagery. Similarly, the advertising images depicting same-sex couples can from now on be seen as out-group ad imagery (from a mainstream heterosexual perspective).

A first observation derived from the table is that the mean average scores of the out-group ad imagery yielded higher scores than that of the in-group imagery. This is consistent with

(39)

the theory of Peñaloza (1996), who states that homosexuals tend to rate ads higher when homosexual group cues are given. More importantly, from Table 2 we can draw a ranked scale of felt group belonging of each eight ads. In other words, the ad imagery that yields the highest score of social group belonging to the heterosexual mainstream is that of H&M. Respectively followed by Calvin Klein, Chanel and Hugo Boss. The ad imagery that scores the highest in terms of out-group belonging, thus belonging to a homosexual minority social group, is Calvin Klein, respectively followed by Hugo Boss and Chanel. The advertising imagery of H&M can be assessed to having the lowest score of felt out-group belonging of all out-group ad imagery.

Table 2: Means, Standard Deviations of Felt Group Belonging Advertising Imagery.

In-group Ad Imagery Out-group Ad Imagery

Mean SD Mean SD

H&M 4.80 .77 5.02 .86

Calvin Klein 4.72 .90 5.89 .64

Chanel 4.52 .96 5.24 1.24

Hugo Boss 4.19 1.06 5.87 .64

Note. Values are the mean of reported scores of Felt Ad Group Belonging on a 7-point scale (1 = strongly disagree, 7 = strongly agree) adapted from Aaker, Brumbaugh & Grier (2000).

3.4 Measurement

Ad imagery functions as the independent variable in this research. It will be manipulated into in-group ad imagery and out-group ad imagery. The non-manipulated Ad imagery serves as a control group. Within each treatment ad imagery group, the imagery is pre-tested on levels of felt group belonging, as discussed in the previous section. In order to analyze the effect of ad imagery, dummy variables are used within the dataset.

Brand attitude serves as the first dependent variable within this study. Brand attitude is measured on a six item scale adapted from Braner (1998). Respondents rate the brand

(40)

attitude of the brand exposed on a bipolar 7-point Likert. An example of a question is: "Please indicate your attitude towards the brand Hugo Boss after seeing this ad". Items range from 1 to 7 (bad - good; unattractive - attractive; boring - interesting; unappealing - appealing; dislike - like; unpleasant - pleasant). The scale reported previously a Cronbach's α = 0.89, thus scoring high on the reliability of the scale.

The second dependent variable that is used within this study is purchase intentions. To measure purchase intentions an adapted four item scale of Bearden, Lichtenstein and Teel (1984) is used. Respondents are asked, for example, "How likely is it that you would consider purchasing a product of the brand Chanel after seeing this ad?" The four items are measured on 7-point bipolar Likert scale, including the items: unlikely - likely; improbable - probable; uncertain - certain; definitely not - definitely; respectively ranging from 1 to 7. The measurement scale yielded in previous research Cronbach's α = 0.90, thus making it a suitable measure for this study.

Brand personality, based on gender dimensions, functions as the moderating variable in this research. As brand personality, as defined by Grohmann (2009), is an orthogonal concept, it is measured along the dimensions of masculine brand personality (MBP) and feminine brand personality (FBP). Masculine brand personality is measured on a six item 7-point Likert scale, ranging from 1 (not at all descriptive) to 7 (very descriptive). Respondents are asked, for example: "Please indicate to what extent the characteristics are descriptive for the brand H&M." Items include: adventurous; aggressive; brave; daring; dominant; sturdy. The scale yielded in previous research a Cronbach's α = 0.89. Feminine brand personality is measured in a similar fashion. Items include: express tender feelings; fragile; graceful; sensitive; sweet; tender. The scale previously reported Cronbach's α = 0.90, thus making it a reliable measure to use.

(41)

After measurement the brands can be assessed into four different categories based on the Brand personality scores, according to research of Grohmann (2009). Masculine brands score high on MBP (> 4) and low on FBP (< 4) and feminine brands score vice versa. Undifferentiated brands score low on both measures (< 4), whereas androgynous brands score high on both MBP and FBP (> 4) (Grohmann, 2009). After assessing each brand, dummy variables are used within the analysis. Lastly, respondents are asked questions regarding their gender, age, country of residence and sexual orientation. The former three serve as control variables. Sexual orientation is used as a sampling tool, as only those respondents who identity as heterosexual can be used for the purpose of this research.

3.5 Sampling and Data Collection

This quantitative study is of exploratory nature and follows an experimental research design. Cross-sectional data is collected through four different surveys, which are distributed online with the use online survey software Qualtrics. In each survey, subjects are exposed to manipulated branded advertising imagery showing romantic heterosexual couples (treatment), homosexual couples (treatment), or no couples (non-treatment) of four different brands in the apparel industry. Before exposure to the advertisements (stimuli), brand personality, based on gender dimensions, of the brands exposed is measured with a questionnaire. After each exposure to the ad, subjects are given a questionnaire measuring brand attitude and purchase intentions of the brand advertised. Hence, each subject is assigned to rate one brand on masculine and feminine brand personality scales, and three randomized ads of all treatments. Subjects are asked to imagine they would come across the ads in real life. In this regard, the experiment can be seen as vignette experiment. The study can be viewed as deductive.

(42)

To minimize carry-over effects, the order of the advertisements are also randomized within each group. Moreover, in order to prevent carry-over effects and confusion from respondents, subjects are assigned to only one ad of each brand. In this way, it is made impossible for respondents to compare the ads on their in-group and out-group situation, which might lead to biased ratings. The respondent are all asked all the same questions. In this regard, the study takes on a repeated measures design. An overview of the survey flow is given in Table 3. The full survey can be found in the Appendix.

Table 3: Survey flow and treatment exposure within groups.

Group 1 Group 2 Group 3 Group 4

Survey Part Content/ Measurement Part 1 Introduction v v v v randomization

Part 2 Brand personality MBP/FBP

Hugo Boss MBP/FBP Chanel MBP/FBP H&M MBP/FBP Calvin Klein randomization

Part 3 Brand attitude and Purchase intentions Exposure to in-group ad of Calvin Klein Exposure to in-group ad of Hugo Boss Exposure to in-group ad of Chanel Exposure to in-group ad of H&M

Part 4 Brand attitude and Purchase intentions Exposure to control ad of Hugo Boss Exposure to control ad of Chanel Exposure to control ad of H&M Exposure to control ad of Calvin Klein

Part 5 Brand attitude and Purchase intentions Exposure to out-group ad of Chanel Exposure to out-group ad of H&M Exposure to out-group ad of Calvin Klein Exposure to out-group ad of Hugo Boss

Part 6 Control questions v v v v

Note. The order of ads exposed within each group (part 3, part 4 and part 5) is also randomized.

Part 1 includes an introduction to respondents with short instructions. Participants are asked to agree on taking part on the research. If they agree, they are randomly assigned to each of the groups of Part 2. In this part, respondents are given a brand, which they have

(43)

to rate on masculine and feminine brand personality scales. After this, another randomization takes part and participants continue with Part 4. In this part, the first advertisement is shown and respondents are asked to pretend they encounter the ad in an everyday situation, such as on a bill board or in a magazine. Questions regarding brand attitudes and purchase intentions are asked. Part 5 and Part 6 continue in a similar way. Moreover, Part 4, 5 and 6 are randomized within the groups, so that carry-over effects are minimized. Put differently, each participant gets to see an in-group ad, an out-group ad and a control ad of different brands, and in randomized order. The survey ends with Part 6, where control questions and demographics are asked to all respondents.

Prior to the main research, a pilot test was carried out among four respondents in order to account for any inconveniences or mistakes in the survey. These subjects are not included in the main study. After the pilot test, some adjustments were made in explaining the different parts of the survey. Also, some misspellings were taken out. The biggest adjustment was the re-sizing of each ad, because these were perceived as being too large by the pilot test respondents. After the adjustments the ads were also easier to view on mobile devices.

Following the main study, respondents are selected via non-probability convenience sampling and snowball sampling, using primarily e-mail and social media (Facebook and Instagram), with regard to the researcher's personal network. Respondents are asked to distribute the survey among their network as well. Due to this way of distribution, it is unclear to predict how many surveys are spread. Subjects are then automatically randomly assigned to one of four surveys to ensure a rough equivalent spread of attributes across all groups. In order to have sufficient data for all four surveys, a minimum of 160 respondents is required, nevertheless, more is aimed. Consumers in the ages 18-65 are considered the

(44)

population for this research, as they can be seen as the target group for the brands advertised. However, in order to answer the research question, only the data of consumers that identify as heterosexual is examined. Consequently, respondents that identify as homosexual or indicate "prefer not to say" are excluded from the analysis.

(45)

4 Results

In the following chapter the hypotheses are tested. We begin with an overview of the subjects, followed by a description of the undertaken preliminary step. After model testing, the chapter ends with a complete overview of the accepted or rejected hypotheses.

4.2 Subjects

On Wednesday 30 November 2016, the survey was sent out to respondents. Exactly two weeks later, on 14 December 2016, the survey was closed. Of the 238 responses that were recorded, 219 respondents fully completed the survey. The data set yielded 46.1% male (n = 101) and 53.9% female (n = 118) respondents. Of the total respondents, 9.1% (n = 20) indicated they identify as LGBTQ. This is in line with findings of Bath et al. (1996) who state that estimations of the population that identifies as LGBTQ ranges from 2.8% to 10%. However, for the purpose of this research and in order to draw conclusions from a heterosexual mainstream perspective, LGBTQ respondents are excluded from analysis. Moreover, as we can only guess the sexual orientation of the respondents that "prefer not to say," these subjects are excluded from analysis as well.

Consequently, after exclusion the valid data set yielded a response of 195 respondents, of which 45.1% is male (n = 88) and 54.9% female (n = 107). Respondents have an average age of 28.6 years (SDage = 7.9, age range: 18-61). So, the respondents are fairly young. Out of these respondents, the majority (88.2%) resides in the Netherlands. Followed by the United States of America (2.1%), Sweden and Germany (both 1.5%). The other respondents reside in other, mostly Western, countries.

(46)

4.2 Preliminary Analysis

When taking the survey, respondents were forced to answer in order to continue. This had the desirable effect of no missing cases. As respondents were randomly assigned to one of the four respondent groups, as outlined in the previous chapter (Table 3), it is important to see whether or not there are significant differences between the respondent groups in terms of gender and age. Table 4 depicts an overview of the sample per respondent group.

For gender, this was checked by means of a Pearson Chi-Square test. A contingency table was created to see whether the respondent groups significantly differed on the distribution of gender. All cells met the minimum expected cell frequency, so no assumptions of the Chi-Square were violated. The test yielded a score of p = .278, which implies that there are no significant differences between the groups in terms of gender, as p > .05. To check for any significant differences between the groups in terms of the distribution of age, a Kruskal Wallis was carried out. Again, no significant differences were found between the groups in terms of age. The test reported a significance of .326, so it can be stated that the distribution of age is the same across the four respondent groups (p > .05). Thus, we can statistically conclude that the randomly assigning of respondents to each of the groups was successful and equally spread, and did not lead to a response bias.

Table 4: Sample per respondent group.

Exposed to Ad Imagery of Brand Size Gender Age

Respondent

group In-group Control Out-group n Man Woman

Mean Age

Group 1 Calvin Klein Hugo Boss Chanel 46 18 28 27.5

Group 2 Hugo Boss Chanel H&M 52 29 23 27.8

Group 3 Chanel H&M Calvin Klein 49 19 30 28.4

Group 4 H&M Calvin Klein Hugo Boss 48 22 26 30.6

Total 195 88 107 28.6

Queer ad for the straight eye : the extent of brand personality on the impact of out-group advertising on brand equity