ABSTRACT
Gender is many things to many people. It is at once uniting and divisive. Gender in advertising presents a range of opportunities for marketers but is linked inextricably to concepts of masculinity–femininity and sexuality and has the potential to create strong positive or negative responses from different audiences. Using an experimental methodology, this study examines how depictions of gender and sexuality combine to influence consumer attitudes. The findings demonstrate how individuals' political ideology determines their emotional response to such advertisements and that the resulting emotional response and attitude to the advertisement act as causal mechanisms responsible for product-related attitudes.
MANAGEMENT SLANT
Depictions of gender in advertisements influence viewer emotions.
Male homosexuality creates more powerful, negative emotions than female homosexuality.
Such responses are more prevalent in viewers who subscribe to a conservative political ideology.
Managers could look to geopolitical segmentation and targeting as a valuable technique to avoid alienating specific segments, whereby managers can utilize election data as a proxy for consumers' political ideologies.
INTRODUCTION
Mainstream media focus on aggregating large, heterogeneous audiences, so they typically are designed to appeal to as many people as possible, alienate as few as possible, and follow a “supposedly nonideological middle ground” (Gross, 2012, p. 7). As a result, most mainstream advertising typically has been targeted at the heterosexual customer (Burnett, 2000), and heterosexual masculinity and the hypersexualized female are considered “normal” (Rubin, 1984).
Despite that, recent years have seen marketing practitioners and scholars increasingly consider the role of lesbian, gay, bisexual, transgender, and queer (LGBTQ) imagery in advertising. This has been driven somewhat by an emerging level of discontent within the LGBTQ community regarding their underrepresentation in mainstream media. Research (Nölke, 2018) has found that even when LGBTQ-explicit advertising is employed, non-LGBTQ consumers are the underlying target group. Advertisers and media organizations therefore have been encouraged to reconsider their advertising practices to become more inclusive of the LGBTQ community (Ginder and Byun, 2015).
Although LGBTQ consumers have been found to be receptive to such imagery, however, there are negative effects, particularly for straight audiences (Dotson, Hyatt, and Petty Thompson, 2009). In this respect, it has been suggested that some sections of the heterosexual community are not readily accepting of homosexuality in general (Kite, 1984) and gay men in particular (Gentry, 1987). In terms of heterosexual attitudes to homosexuality, there appears to be a form of “homo-gender bias,” whereby gay men are perceived in a more negative way than gay women. Yet, what differentiates those sections of the heterosexual community that are or are not accepting of homosexuality in advertising? The answer might lie in an individual's political ideology.
Political ideology is the set of ideas and values that shapes an individual's worldview and justifies his or her political stance (Dawson, 2001). When it is considered in a marketing context, political ideology has the ability to influence consumer attitudes (Kaikati, Torelli, Winterich, and Rodas, 2017), because it aligns an individual's political and moral identity (Winterich, Zhang, and Mittal, 2012). As such, a person's political ideology is central to his or her attitudes and behaviors involving consumption (Crockett and Wallendorf, 2004).
This is particularly so if any product or brand advertising happens to include LGBTQ imagery, because politically conservative individuals are less open than liberals to sexual activity that falls outside the traditional definition (Chan, 2019). Additionally, individuals' level of “right-wing authoritarianism” is correlated negatively with their attitudes toward homosexuals (Haddock, Zanna, and Esses, 1993). Given this, it is plausible that consumers who subscribe to a conservative political ideology will be less accepting than politically liberal consumers of advertising that includes LGBTQ imagery.
On the basis of this, the current research set out to examine the effects of gender representations in advertisements containing LGBTQ imagery. In doing this, the study looks at individual-level differences to explain why some politically conservative consumers have difficulty accepting LGBTQ imagery in advertising. The authors also examine the types of emotional responses individuals experience when viewing LGBTQ imagery.
What emotions are being generated, however, and what role do these play? One emotion that stands out is disgust. Disgust is related to an individual's aversion to contact with unsavory individuals or those who may “contaminate” in some way (Hodson, Choma, Boisvert, Hafer, et al., 2013). As such, it could be inferred that advertisements referring to or depicting male-to-male (versus female-to-female) homosexuality likely will elicit disgust (Maison and Pawłowska, 2017) in viewers who subscribe to a certain type of conservative political ideology (Hodson and Dhont, 2015).
The contributions of this study are threefold. First, according to current literature, this study appears to be the first to identify disgust as the primary causal mechanism influencing consumer evaluations of advertisements containing LGBTQ imagery. Second, the findings from this research demonstrate that individuals' conservative political ideology has a positive influence on their level of disgust. Third, results show that for politically conservative viewers, both implicit and explicit (hypersexualized) representations of male-to-male homosexuality generate disgust. By contrast, imagery depicting heterosexual couples or female homosexuality are acceptable across the population. The rest of the article begins with a literature review and conceptual development, sets out the methodology and analyses, and concludes with a discussion that outlines theoretical and managerial contributions.
CONCEPTUAL DEVELOPMENT AND HYPOTHESES
Gender and Sexuality
The difference between male and female is a common, recurring theme that transcends both plant and animal kingdoms (Gochfeld, 2007). In the human species, such differences are magnified, because not only does human gender involve the standard (biological) male–female dyad, it has the potential to influence an individual's self-identity. Consider the typical human as an example. In such a case, gender and masculinity–femininity not only define manhood–womanhood but also serve as “the primary building blocks of sexuality” (Fracher and Kimmel, 1995, p. 367). This is certainly evident in North America and Western Europe, where societies have organized their systems of sex, gender, and sexuality around “a set of intersecting dichotomous pairs: masculine/feminine and homosexual (forbidden)/heterosexual (permitted)” (Segal, 2003, p. 4).
Not only does such a paradigm limit and guide attitudes and behaviors, it also situates sexuality as a powerful expression of masculinity or femininity, and this can manifest itself in the form of acceptance or nonacceptance of groups that do or do not conform to an individual's idea of accepted norms. Prior research has shown that heterosexuals tend to have a dislike for homosexuals (Lim, 2002). When gender is included in such discussions, research (Gentry, 1987) suggests that heterosexual individuals experience higher levels of discomfort toward same-sex homosexuals than opposite-sex homosexuals. This has important implications for any brand looking to use gender or sexuality in its advertising.
Gender and Sexuality in Advertising
Given that the male role in Western society typically emphasizes the importance of heterosexuality to masculinity (Oakenfull and Greenlee, 2004), it is natural to suggest that homosexual imagery in advertising may elicit disapproval from segments of the population (Hester and Gibson, 2007). To some degree, this may be explained by the natural aversion male heterosexuals have toward homosexuality (Kite, 1984), particularly when the homosexuality is male to male versus female to female (Gentry, 1987). Such bias has been shown to influence viewer responses to advertisements (Oakenfull, McCarthy, and Greenlee, 2008; Read, van Driel, and Potter, 2018), whereby heterosexual viewers react more negatively to advertisements featuring male homosexuality than they would to ones featuring female homosexuality (Bhat, Leigh, and Wardlow, 1996). What one effectively sees is a form of homo-gender bias.
In such cases, male-to-male homosexuality is seen as unappealing and confronting, whereas lesbian imagery provides some form of erotic value to the viewer (Whitley, 1988). This may explain why straight-looking gay men are used to portray gay men to heterosexual audiences. Such an assimilationist appeal is designed to include everyone, albeit within the confines of mainstream heteronormative ideals (Tsai, 2012; Warner, 1993).
Does this mean an audience's sexuality (homosexual or heterosexual) is the most important concern for advertisers when developing marketing collateral and media strategy? Some research certainly supports this idea. In a study investigating responses to financial-services advertising (Jaffe, 1991), female characters were depicted in either modern (career woman making a decision about financial services) or traditional (nurturing woman focusing on her family's financial security) roles. Although the modern positioning resulted in more positive responses to the financial-services advertisements, these findings were dependent on the sex-role identity of the viewer.
Likewise, viewers' sexuality has been shown to influence their acceptance of implicit versus explicit LGBTQ imagery (Oakenfull and Greenlee, 2005). Similar results were found in a later study in which heterosexual audiences were shown to prefer less-overt LGBTQ imagery in advertisements (Dotson et al., 2009). What is more, heterosexuals invariably preferred overt depictions of female homosexuality before overt depictions of male homosexuality (Oakenfull and Greenlee, 2004). This may explain why the most dominant appeal for women in LGBTQ-themed advertisements is sexual (Van Esch, Geisler, Kleintop, Northey, et al., 2017).
Such differential effects in viewer segments come about because of dissimilar experience and information asymmetries among viewer groups. As a result, target and nontarget viewers construct meaning differently (Grier and Brumbaugh, 1999). In effect, target viewers may take a positive referential interpretive stance, whereby they generate meanings that positively relate the advertisement to themselves. Alternatively, nontarget viewers may take a negative referential interpretive stance if they feel the advertisement is targeted to an outgroup. In such instances, it could be assumed that advertisements containing homosexual imagery may be viewed negatively by all heterosexuals.
Such assumptions may be too broad, however. What is more likely is that certain segments of the heterosexual audience, such as those who are politically conservative, will have a negative referential stance. This is because political conservatism is correlated negatively with attitudes toward homosexuality (Haddock et al., 1993).
Red Effects: The Influence of Political Ideology
The relationship between homosexuality and political ideology has been shaped over the past 2,000 years. This period has seen the confluence of Greco-Roman and Judeo-Christian ideologies, which relied on the “patriarchal construction and regulation of sex, gender and sexuality as key tools for cultural organisation” (Valdes, 1996, p. 200). This evolved into what is now the Euro-American system, founded on androsexist and heterosexist ideologies. This is not surprising given that throughout history, gender often has been called on to support or legitimize social formations, including nations and nation-states (McElhinny, 2003).
The major catalyst for the alignment of political ideologies with sexual preference was the 1969 Stonewall riot in New York City. At the time, the political left moved away from the Stalinist view that homosexuality is a form of bourgeois decadence alien to the working class, and instead adopted the idea that individuals should have the freedom to use their body as they wish (Thorstad, 1995). This was a defining moment for gay rights and drew a very clear line in the political sand.
The result was that in the ensuing decades, liberals and conservatives have had differing views on homosexuality and its place in American society. As a result, individuals' level of “right-wing authoritarianism” is correlated negatively with their attitudes toward homosexuals (Haddock et al., 1993).
When this is applied in an advertising environment, what occurs is that viewers seeing LGBTQ-themed advertisements find meaning through the lens of their specific subcultural identity. This way, the political meanings of such niche-targeted advertisements can be accentuated not just by the minority group, who see them as a depiction of their subcultural identity, but by the broader (heterosexual) community, who evaluate the messages in relation to their own cultural norms (Tsai, 2012). Essentially, people view and respond to advertisements—and associated products—on the basis of how these reflect their subcultural identity.
This also means, however, that consumption can be a way to make a political statement. As an earlier study stated, “contemporary consumption in the United States is a primary arena in which political ideology is expressed and constructed” (Crockett and Wallendorf, 2004, p. 511). In a marketing context, this means that what one consumes or wears also can be interpreted as a visible symbol of one's politics and values (Rohlinger, 2002).
In a similar vein, those who subscribe to right-wing authoritarianism likely will respond to advertisements and brands in a way that mirrors their values and political ideology. This is because politically conservative individuals are less open than liberals to sexual activity that falls outside the traditional definition (Chan, 2019). On the one hand, therefore, more positive approval and brand attitudes may result from advertisements containing LGBTQ imagery for those who are more accepting of homosexuality (Hester and Gibson, 2007). On the other hand, however, if an individual's level of right-wing authoritarianism is correlated negatively with his or her attitudes toward homosexuals (Haddock et al., 1993), it is plausible that advertisements containing LGBTQ imagery—and the products they promote—also will be received negatively.
H1: Male (versus female) homosexual imagery will have a negative influence on product evaluations for those consumers who are politically conservative.
Although the prevailing literature provides ample evidence for the hypothesized effects, there appears to be little, if any, literature that explains why these effects might occur. For this reason, the authors turned to research on human emotions to unpack and identify what might be causal mechanisms in this milieu of gender and political ideology.
That's Disgusting! The Influence of Emotion on Consumer Attitudes
Emotions are complex response tendencies that unfold over a short period of time and usually occur because of an individual's assessment of an antecedent event (Fredrickson, 2001). In relation to homosexuality, these emotional reactions are strong predictors of overall affect in relation to gay men and women (Miller, Smith, and Mackie, 2004). Interestingly, though, this research (Miller et al., 2004) measured 12 emotions - afraid, angry, disgusted, uneasy, hopeful, proud, sympathetic, resentful, respectful, grateful, admiring, irritated—but these were collapsed into two dimensions, positive or negative.
Similar findings were provided in a later study, although, again, the results were constrained by the use of general positive–negative dimensions (Puntoni, Vanhamme, and Visscher, 2011). Although the findings demonstrated that emotional responses mediated the effects of sexual orientation and advertisement type on attitudes toward an advertisement, they did not identify the discrete emotions responsible. On the basis of this, a review of available literature was undertaken, with the output showing that of all the negative emotions, disgust appeared to be a potential outlier.
Disgust is a basic, universal emotion (Rozin, Haidt, and McCauley, 2008) and is associated with the neurological areas that signal danger (Harris and Fiske, 2006). Despite that, there appears to be limited research examining disgust in relation to prejudice. This is surprising given that disgust concerns aversion to contact with unsavory individuals or those who may “contaminate” (Hodson et al., 2013) and thus is theoretically relevant to prejudice.
If a group feels that some form of contact with people outside their group will cause contamination, the emotional response is disgust. In turn, this leads the in-group actively to avoid or reject the out-group to remove the repulsive stimulus (Cottrell and Neuberg, 2005). This is particularly pertinent to the current study, given that sensitivity to disgust is also a key predictor for homophobia (Terrizzi, Shook, and Ventis, 2010).
In an advertising context, then, advertisements that refer to or depict homosexuality likely will elicit disgust in certain segments of the audience (Maison and Pawłowska, 2017). Those segments, however, likely will be ones that subscribe to right-wing authoritarianism, given that political conservatism is a significant predictor of out-group prejudice (Hodson and Dhont, 2015) and intergroup disgust sensitivity is associated with authoritarian, conservative ideologies (Hodson et al., 2013).
H2a: Male (versus female) homosexual imagery will have a positive influence on perceived level of disgust for those consumers who are politically conservative.
Although viewers will experience an emotional response following exposure to an advertisement, the advertising content also likely will influence the viewers' evaluation of the advertisement. As such, when LGBTQ (versus heterosexual) imagery is present, viewers likely will have a less-positive response to the advertisement (Read et al., 2018; Septianto, Northey, and Dolan, 2019). What is more, attitudes toward the advertisement will be less positive for advertisements containing gay male imagery when compared with advertisements containing lesbian imagery (Ivory, 2017). So, while viewers might even prefer overt lesbian imagery over overt gay male imagery (Oakenfull and Greenlee, 2004), positive evaluations of an ad will generally come about when the LGBT imagery present fits with the viewers' existing gender schemas (Pounders and Mabry-Flynn, 2016). Because of this, individuals' attitudes to homosexuality are a strong moderator in their evaluation of advertisements containing LGBTQ imagery (Ginder and Byun, 2015).
H2b: Male (versus female) homosexual imagery will have a negative influence on attitude toward the advertisement for those consumers who are politically conservative.
The hypothesized influence of male (versus female) homosexual imagery on consumer perceived disgust and attitude toward the advertisement is well supported by existing literature. Each variable (disgust, attitude to advertisement), however, unlikely will occur in isolation. This is particularly the case if one considers the influence of male (versus female) homosexual imagery on consumer attitudes toward the advertised product. In such instances, one might expect LGBTQ imagery to influence consumer disgust, which acts as a determinant for attitude to the advertisement, thereby influencing product-level affect. Previous research (Edell and Burke, 1987; Holbrook and Batra, 1987) provides evidence that emotional reactions to advertisements mediate the effect of advertisement content on various consumer attitudes. More relevant to the current study, previous research demonstrates that emotional responses fully mediate the effects of sexual orientation and advertisement type on attitudes toward an advertisement (Puntoni et al., 2011), whilst another group extended this by demonstrating that the effects are dependent on the viewers' general attitudes toward homosexuality (Bhat, Leigh, and Wardlow, 1998). As a result the following hyposthesis is proposed:
H3: For politically conservative consumers, the influence of male (versus female) homosexual imagery on product attitudes will be mediated by an individual's level of disgust and attitude toward the advertisement, in a moderated serial mediation.
A full conceptual model outlining key constructs and hypotheses is included below (See Figure 1).
STUDY 1A
Methodology
Participants and Design. A total of 859 participants (47 percent female; MAge = 37.69, SD = 11.34) were recruited through a research agency in the United States. The study employed a 3 (gender: homosexual men, homosexual women, heterosexuals) × 2 (sexuality: implicit, explicit) × 2 (race: same, interracial) between-subjects design. The study was conducted online, using only U.S. residents. Participants received a unique URL to complete the survey using the Qualtrics (www.qualtrics.com) survey instrument.
Although the influence of gender of the model was the focus of the study, the authors included race and level of sexuality to test the effects of gender in the presence of other factors, thereby adding to the robustness of the findings. In particular, they included race to determine whether it helped explain the relationship of attitudes to the product, given earlier research that found viewers of advertising reacted differently to white and black models in an advertisement (White and Harkins, 1994).
Additionally, the level of sexuality was manipulated according to guidelines established in previous work (Tharp, 2001), whereby markers of gay identity would denote in a subtle manner the LGBTQ theme of the advertisement. For the implicit sexuality, this was achieved with the tagline “Love Is Love,” with pretesting confirming the LGBTQ connotation. Conversely, explicit LGBTQ-themed advertising uses erotic or sexual encounters between two members of the same sex (Oakenfull et al., 2008). As such, the authors achieved the explicit sexuality by posing models in intimate encounters. (Advertisements are included in Appendix A.)
Procedure. The authors created separate advertisements for a fictitious product and brand of luxury watches by manipulating the images of the models. Luxury watches were used because they are relatively gender neutral compared with other fashion products (Ehrnberger, Räsänen, and Ilstedt, 2012). A fictitious watch brand, Tempus (Latin for “time”), was created. The use of a fictitious brand is an accepted technique in consumer research (Stewart, 1992), because it avoids triggering any preexisting emotions or attitudes participants may feel toward a familiar brand (Money, Shimp, and Sakano, 2006).
The tagline “Love Is Love” was included adjacent to the product and brand name, because it is synonymous with the LGBTQ movement (Nichols, 2014). This enabled the authors simply to manipulate gender of the models in the advertisement while leaving all else constant. After evaluating the advertisement, participants completed several measures and provided demographic information.
Measures. To measure attitude toward the brand, the authors used the focal dependent variable, an existing 5-item scale (Aggarwal and McGill, 2011). All items involved a 9-point Likert-type scale with anchors “completely disagree” (1) and “completely agree” (9). The items were “I like the watch in the advertisement,” “I admire the watch in the advertisement,” “The watch would fit with my life,” “It would be embarrassing to be seen wearing that watch” (reverse-coded), and “I would avoid wearing that watch” (reverse-coded; α = 0.79).
To measure political ideology (the proposed moderator), participants completed three questions on a 7-point scale from prior research (Kaikati et al., 2017). These questions were
“Please indicate the political label with which you most identify” (1 = “extremely liberal,” 7 = “extremely conservative”),
“I think of myself as a…” (1 = “strong Democrat,” 7 = “strong Republican”), and
“Politically, I would describe myself as” (1 = “extremely liberal,” 7 = “extremely conservative”; α = 0.94).
A higher political-ideology score thus indicated a more conservative participant, whereas a lower political-ideology score indicated a more liberal participant.
To measure disgust (the proposed first mediator), participants rated their emotional response to the advertisement on a 9-point Likert-type scale (1 = “not at all,” 9 = “extremely”). The items were “disgusted” and “repulsed” (collapsed to form an index of disgust; r = 0.92, p < 0.001). To ensure a robust dataset and to avoid overlooking other factors that might be active determinants, the authors measured other emotional responses. These included “angry,” and “mad” (collapsed to form an index of anger; r = 0.84, p < 0.001), “amused” and “happy” (collapsed to form an index of happiness; r = 0.51, p < 0.001), and “affectionate” and “love” (collapsed to form an index of love; r = 0.91, p < 0.001). These items have been validated by prior research (Cavanaugh, Bettman, and Luce, 2015; Garg, Wansink, and Inman, 2007; Lerner and Small, 2004).
To measure individual attitudes toward the brand (the proposed second mediator), the authors used a 4-item scale (developed by Lawrence, Fournier, and Brunel, 2013). The items, measured on 9-point Likert-type scales (1 = “completely disagree,” 9 = “completely agree”) were “The advertisement was entertaining,” “The advertisement made me happy,” “The advertisement made me pleased,” and “The advertisement was a pleasure to look at” (α = 0.97).
Finally, the authors also measured prior attitudes toward homosexuality as covariates. Participants responded to four items on a 9-point scale (1 = “completely disagree,” 9 = “completely agree”) from previous research (Puntoni et al., 2011). The items were “Sex between two people of the same sex is just plain wrong” (reversed-coded), “I think homosexuals are disgusting” (reversed-coded), “Homosexuality is a natural expression of sexuality,” and “Homosexuals are just as likely to be good people as anyone else” (α = 0.90).
Results and Discussion
Common Method Bias. To ensure robust data, the authors used Harmon's one-factor test to check for common method bias, wherein all items of the latent variables are loaded onto one factor. The factor combined explained less than 50 percent (29.41 percent) of variance, indicating that common method variance did not pose a problem to the data (Podsakoff and Organ, 1986).
The authors also examined the correlations among the variables (attitude to the advertisement, attitude to the brand, consumer disgust, and political ideology) and calculated their corresponding variance inflation factor values to ensure that there was no multicollinearity problem. In models with fewer variables involving one or two interactions, multicollinearity typically is not a concern (Hayes, 2013). The authors consequently found no evidence of multicollinearity, because the variance inflation factor scores ranged between 1.08 and 1.22, far below the critical value of 5 (Hair, Sarstedt, Hopkins, and Kuppelwieser, 2014; Hayes, 2013).
Hypothesis 1: Attitudes toward the Product. Because political ideology is a continuous variable, data were analyzed according to guidelines set out by previous researchers (Aiken and West, 1991). A moderated regression analysis was run with attitudes toward the product as the dependent variable. For the independent variables, the authors included political ideology scores; gender of models (−1 = homosexual men, 0 = heterosexuals, 1 = homosexual women); race of models (0 = same race, 1 = mixed race); sexuality (0 = implicit, 1 = explicit); and their two-way, three-way, and four-way interactions. As expected, the only significant effect in the model that emerged was the interaction between political ideology and gender of models (B = −0.45, SE = 0.19), t(827) = −2.37, p = 0.018 (See Table 1).
The authors conducted spotlight analyses to examine this interaction effect further by examining how liberals (defined as low on the political ideology scale [−1 standard deviation]) and conservatives (defined as high on the political ideology scale [+1 standard deviation]) reported their positive attitudes toward the product. Results showed that liberals had similar attitudes toward the product when viewing an advertisement involving male-to-male homosexuality versus heterosexuality (B = −0.10, SE = 0.22), t(827) = −0.43, p = 0.665, and an advertisement involving male-to-male (versus female-to-female) homosexuality (B = −0.32, SE = 0.22), t(827) = −1.49, p = 0.137. Conservatives, however, showed more negative attitudes toward the product when viewing the advertisement of male-to-male homosexuality versus heterosexuality (B = 0.66, SE = 0.22), t(827) = 3.01, p = 0.003, and the advertisement of male-to-male (versus female-to-female) homosexuality (B = 0.44, SE = 0.22), t(827) = 1.95, p = 0.051 (See Figure 2). These findings provide evidence in support of H1.
Hypothesis 2a: Disgust. To test the hypothesized effect, the authors ran a moderated regression analysis with disgust as the dependent variable, and political ideology scores, models' gender (−1 = homosexual men, 0 = heterosexuals, 1 = homosexual women), and their interaction as independent variables. Regression analysis revealed a main effect of political ideology (B = 0.50, SE = 0.05), t(852) = 10.97, p < 0.001. The analysis also showed a main effect of gender (B = 0.53, SE = 0.22), t(852) = 2.38, p = 0.017. It is important to note that, as predicted, a significant interaction was shown between political ideology and gender (B = −0.21, SE = 0.06), t(852) = −3.69, p < 0.001.
The authors conducted additional spotlight analyses to examine this interaction effect further. Results revealed that liberals reported similar levels of disgust after evaluating an advertisement featuring male-to-male homosexuality versus heterosexuality (B = 0.24, SE = 0.26), t(852) = 0.93, p = 0.354, and an advertisement involving male-to-male (versus female-to-female) homosexuality (B = 0.24, SE = 0.26), t(852) = 0.91, p = 0.362. In contrast, conservatives reported higher levels of disgust when viewing the advertisement with male-to-male homosexuality versus heterosexuality (B = −0.83, SE = 0.26), t(852) = −3.15, p = 0.002, and the advertisement involving male-to-male (versus female-to-female) homosexuality (B = −0.44, SE = 0.19), t(852) = −2.35, p = 0.019 (See Figure 3). These findings support H2a.
Hypothesis 2b: Attitudes toward the Advertisement. A moderated regression analysis was run with attitude toward the advertisement as the dependent variable and political ideology scores, models' gender (−1 = homosexual men, 0 = heterosexuals, 1 = homosexual women), and their interaction as independent variables. Regression analysis revealed a significant interaction between political ideology and models' gender (B = 0.18, SE = 0.06), t(851) = 2.86 p = 0.004 (other main effects were nonsignificant).
Spotlight analyses showed that liberals had similar attitudes toward the advertisement when viewing male-to-male homosexuality versus heterosexuality (B = −0.03, SE = 0.29), t(851) = −0.08, p = 0.938, and when viewing male-to-male (versus female-to-female) homosexuality (B = 0.18, SE = 0.29), t(851) = 0.62, p = 0.535. Conservatives, however, showed more negative attitudes toward the advertisement when viewing male-to-male homosexuality versus heterosexuality (B = 1.11, SE = 0.21), t(851) = 5.39, p < 0.001, and male-to-male (versus female-to-female) homosexuality (B = 0.80, SE = 0.21), t(851) = 3.79, p < 0.001 (See Figure 4). These results support H2b.
Hypothesis 3: Moderated Serial Mediation. Consistent with the proposed conceptual model, to test for a moderated serial mediation, the authors used PROCESS Version 3.1 (Hayes, Montoya, and Rockwood, 2017), Model 85 (see Hayes, 2013), with 5,000 bootstrap resamples and a 95 percent confidence interval. The analysis examined the indirect effects of advertisement (−1 = homosexual men, 0 = heterosexuals, 1 = homosexual women) on attitudes toward the product (the dependent variable), serially mediated by disgust and attitudes toward the advertisement and moderated by political ideology, where higher scores reflect a conservative ideology. Because the independent variable was a multicategorical variable, PROCESS automatically created dummy variables to examine two separate analyses:
male-to-male homosexuality versus heterosexuality, and
male-to-male (versus female-to-female) homosexuality.
Research suggests that existing attitudes toward homosexuality can influence consumers' attitudes toward the advertisement and the product (Hester and Gibson, 2007). Similarly, other emotions also might explain the authors' predictions, instead of disgust. For instance, consumers have reported anger when viewing LGBTQ-themed advertisements (Oakenfull and Greenlee, 2004). Hence, to account for plausible alternative explanations for the predictions, the authors included the emotions of anger, happiness, love, and attitudes toward homosexuality as covariates (See Appendix B for correlations between constructs used in the mediation analysis).
When the authors examined advertisements depicting male-to-male homosexuality versus heterosexuality, the indirect effects of models' gender on attitudes toward the product, through the two serial mediators of disgust and attitudes toward the advertisement, were significant depending on the political ideology. As predicted, the index of moderated mediation was significant (B = 0.011, SE = 0.005, 95 percent confidence interval [CI]: .002 to .023). Among liberals, the indirect effect was nonsignificant (B = 0.010, SE = 0.008, 95 percent CI: −0.005 to 0.028). As predicted, however, the indirect effect among conservatives was significant (B = 0.045, SE = 0.016, 95 percent CI: 0.019 to 0.081).
When the authors examined advertisements depicting male-to-male (versus female-to-female) homosexuality, the indirect effects of models' gender on attitudes toward the product, through the two serial mediators of disgust and attitudes toward the advertisement, were significant, depending on the political ideology. As predicted, the index of moderated mediation was significant (B = 0.011, SE = 0.005, 95 percent CI: 0.002 to 0.023). Among liberals, the indirect effect was nonsignificant (B = 0.008, SE = 0.007, 95 percent CI: −0.007 to 0.023). As predicted, however, the indirect effect among conservatives was significant (B = 0.043, SE = 0.016, 95 percent CI: 0.017 to 0.079; See Appendix C for full mediation results). Taken together, these results provide support for H3.
STUDY 1B: REPLICATION STUDY
Study 1b was conducted to replicate the findings of Study 1a using a different product category. The aim was to demonstrate that the predicted effects occurred as a result of the gender (male-to-male versus female-to-female homosexuality) presented in the advertisements and were not unique to a single product. To replicate Study 1a, the authors replaced the original luxury watch with bottled water, because bottled water is considered to be a neutral food product (Winterich and Barone, 2011). For brevity, three advertisements were used to test the effects of gender (male homosexuality, female homosexuality, heterosexuality). (Advertisements used in Study 1b are included in Appendix A.)
Methodology
Study 1b (N = 158; 47 percent female; MAge = 35.93, SD = 10.64) was a one-factor, three-level (gender: homosexual men, homosexual women, heterosexuals) between-subjects design. Similar procedure and scales were used to those in Study 1a, with the exception of the advertised product. As before, scores for political ideology were included, as well as gender of models (−1 = homosexual men, 0 = heterosexuals, 1 = homosexual women), their two-way interaction as independent variables, and attitudes toward the product as the dependent variable.
Results and Discussion
Consistent with predictions, there was a significant interaction between political ideology and gender of models (B = 0.30, SE = 0.12), t(152) = 2.45, p = 0.015. Spotlight analyses revealed that liberals had similar attitudes toward the product when viewing an advertisement involving male-to-male (versus heterosexual) sexuality (B = −0.46, SE = 0.44), t(152) = −1.05, p = 0.295, and an advertisement involving male-to-male (versus female-to-female) homosexuality (B = −0.18, SE = 0.45), t(152) = −0.40, p = 0.688. Conservatives, however, showed more negative attitudes toward the product when viewing advertisements containing male-to-male (versus heterosexual) sexuality (B = 1.57, SE = 0.47), t(152) = 3.35, p = 0.001. Similarly, conservatives showed more negative attitudes when viewing male-to-male (versus female-to-female) homosexuality (B = 1.66, SE = 0.54), t(152) = 3.08, p = 0.002 (See Figure 5). Using a different product category, these findings provide further support for H1.
GENERAL DISCUSSION
There are many reasons why brands may include LGBTQ imagery in their advertising. It might be to demonstrate their support of the LGBTQ community (Peñaloza, 1996) or to generate arousal and attention in the target audience (Angelini and Bradley, 2010; Read et al., 2018). Prior research (Oakenfull and Greenlee, 2004), however, has suggested that if a brand is to include LGBTQ imagery in its advertising, the focus needs to be “lesbian, lesbian, lesbian.”
There appears to be a lack of clarity as to why this is the case, however. For example, is lesbian imagery simply more visually appealing, or does it depend on individual differences in the viewer cohort? On the basis of this, the objectives of the current research were to test the influence of gender in LGBTQ-themed advertising and examine the influence individuals' political ideology has on their emotional response to advertising containing depictions of gender and sexuality.
The study employed an experimental design to see how consumer responses differed depending on the representations of gender and sexuality (gay, lesbian, heterosexual) in the advertisement. The findings extend prior research and demonstrate that lesbian and heterosexual imagery generated comparable responses in consumers. Consumer attitudes toward the product and brand, however, were dependent on their political ideology. In other words, for those consumers who were politically conservative (versus liberal), the brand or product was less appealing when advertisements incorporated male-to-male homosexual imagery. This effect occurred whether the male-to-male imagery included explicit (hypersexualized) or implicit representations of sexuality.
To extend existing theory, further analysis examined the role consumers' emotional response played in their product evaluations. The authors drew on discrete emotions theory (Izard, 1977; Tomkins, 1963) to guide which emotions might be applicable. Of all the (negative) emotions, however, disgust appeared to be a potential outlier. Disgust is a basic, universal emotion (Rozin et al., 2008) and is associated with the neurological areas that signal danger (Harris and Fiske, 2006). Findings demonstrate that conservatives (versus liberals) experienced a high level of disgust when viewing advertisements containing male-to-male homosexual imagery. This effect was not observed, however, when advertisements contained female-to-female homosexuality or heterosexual couples.
Similar to the consumer's emotional response, when advertisements contained male-to-male homosexual imagery, conservatives (versus liberals) reported a more negative attitude toward the advertisement. Once again, these effects were not observed when the advertisements contained female-to-female homosexuality or heterosexual couples.
The findings from this research are consistent with theorizations in previous literature that suggest conservatives may feel more pressure to conform to group norms (Cavazza and Mucchi-Faina, 2008), a personal characteristic that fosters group identification (Jugert and Duckitt, 2009). Part of this may stem from the fact that conservatives more strongly are motivated to adhere to social norms than liberals (Fernandes and Mandel, 2014) and have more “old-fashioned” attitudes than less-conservative consumers (Van Lange, Bekkers, Chirumbolo, and Leone, 2012).
Moreover, the results indicate that the more conservative participants claimed to be, the more disgust they felt when looking at advertisements containing images of male-to-male (versus female-to-female) homosexuality. No doubt, prior literature provides some evidence that disgust might be an observed response to LGBTQ imagery. One previous study found that those who were more disgust-sensitive evaluated LGBTQ-themed advertisements less favorably than standard advertisements (Inbar, Pizarro, Knobe, and Bloom, 2009). Other research also demonstrated that LGBTQ imagery caused disgust for those who had an existing negative attitude toward homosexuality (Dasgupta, Desteno, Williams, and Hunsinger, 2009).
The current research, however, has extended existing knowledge by identifying political ideology as the key determinant of an individual's emotional response to LGBTQ imagery in advertisements. What is more, it has identified gender (male-to-male versus female-to-female homosexuality) as the source of this emotionally driven “homo-gender bias.”
Given the existing bias against male homosexuality, it is not surprising that the findings of the current research show no difference between conservatives and liberals when viewing advertisements depicting either heterosexuality or female homosexuality. It is interesting that the bias toward male homosexuality appears to be congruent with social role theory, which suggests that prior inferences about presumed homosexuality are influenced by gender-associated beliefs (Bem, 1981). This theory suggests that a violation of gender roles in homosexuality exists and that gay men will be viewed in a more extreme manner than gay women.
Managerial Implications
Given the changing nature of LGBTQ attitudes in society and the prevalence of the movement in media, the findings of this study provide a number of important implications to guide marketing and advertising strategies. At the most basic level, findings from this research suggest that brand managers should avoid the use of male-to-male homosexuality in their advertising if they want to avoid alienating politically conservative viewers. That said, if one looks at past campaigns across a range of industries, including alcohol (Miller, Coors, Skyy), eyewear (Ray-Ban), soft drinks (Coca-Cola), and fashion (Benetton), it is clear that some brands are open to using representations of male-to-male homosexuality in their advertising. The current study, however, provides evidence that such creative executions likely will appeal to some yet alienate other (politically conservative) consumers. It is important to note that the findings from this research demonstrate that these effects are not unique to a single product category.
Of course, such effects have the potential to influence a brand's position in the marketplace. Because of this, managers must have a clear rationale for the use of (male-to-male) homosexual imagery in their campaigns. If this results in a dilution of the existing customer base, however, there likely will be a negative effect on sales and market share.
To combat such events, marketers could consider geopolitical segmentation to guide marketing and media strategy. This is in line with previous research (Jung, Garbarino, Briley, and Wynhausen, 2017) that suggests that election data affords managers an objective means to develop tailored campaigns for different geographical regions. This form of geopolitical segmentation and targeting is a valuable technique for marketers, given that effective and reliable location-based predictors of consumers' behavior have been difficult to obtain.
Managers therefore can utilize election data as a proxy for consumers' political ideologies. By doing this, they can develop and run different creative content and messages on the basis of geopolitical segments. As a result, they have the ability to move past typical stereotypes of homosexuality and communicate to a larger, geopolitically segmented audience, thereby increasing the potential for diversity and inclusion.
Limitations and Future Research
Like most research, this study is not without its limitations. The first is that this study was designed using a fictitious luxury watch brand. Although the results were replicated with a bottled water product, future research could examine whether the findings translate to other product categories. In tourism advertising, for example, would male-to-male (versus female-to-female) homosexuality elicit similar emotions and have a negative influence on destination identified or travel intent? Likewise, would these effects occur when companies are advertising services?
A second limitation is the use of monochrome advertisements. Color has a profound influence on individuals, including emotional and psychophysiological effects (Chylinski, Northey, and Ngo, 2015). Future research could explore whether color influences the emotional effects generated by representations of gender and sexuality and whether there are differential effects for those at alternate ends of the political spectrum.
ABOUT THE AUTHORS
Gavin Northey is a senior lecturer in marketing at Griffith University, Queensland, Australia. His research examines the effects of advertising on consumer decision making and has been published in a range of journals, including International Journal of Research in Marketing, Journal of Business Research, and Psychology & Marketing.
Rebecca Dolan is a senior lecturer in marketing at the University of Adelaide School of Business. Dolan's research focuses on contemporary issues in marketing, such as digital disruption, social media, and customer engagement. Her work has been published in various journals, including European Journal of Marketing, Journal of Strategic Marketing, Internet Research, and Journal of Business Research.
Jane Etheridge is a planning manager at Shopper Media Group Ltd, London. She coauthored this article when she was a strategist at Dentsu Aegis Network in Auckland, New Zealand.
Felix Septianto is a lecturer in marketing at the University of Auckland Business School. His research, which broadly explores the influences of feelings and emotions on consumer behavior, has been published in International Journal of Research in Marketing, European Journal of Marketing, Journal of Business Ethics, and Journal of Business Research, among others.
Patrick van Esch is a senior lecturer in the Department of Marketing, Auckland University of Technology. His main research interests are in artificial intelligence and consumer behavior. Van Esch's research has been published in Business Horizons, International Journal of Consumer Studies, Journal of Retailing and Consumer Services, and Marketing Intelligence and Planning.
Appendix A Individual Advertisements for Each Condition
Appendix B Correlations among Constructs Used in the Mediation Analysis
Appendix C Full Mediation Results (Hypothesis 3)
- Received January 31, 2019.
- Received (in revised form) September 12, 2019.
- Accepted October 24, 2019.
- Copyright© 2020 ARF. All rights reserved.