Boy oh boy! : show some skin! : a quantitative inquiry into women’s responses towards explicit sexual advertising : the effects of various attractive male portrayals in print advertisements

(1)

BOY OH BOY!

Show Some Skin!

A Quantitative Inquiry into Women’s Responses towards Explicit Sexual Advertising:

The Effects of Various Attractive Male Portrayals in Print Advertisements.

Master Thesis

June 2014

Graduate School of Communication Master’s programme Communication Science

Persuasive Communication By: Isabel Beijers, 10625666

(2)

ABSTRACT

This quantitative study explores women’s responses towards explicit sexualized portrayals of males in advertising by measuring the effects of three suggestive sex appeals in print

advertisements retrieved from previous studies: Body display, Sexual Behavior and Sexual Referents. Investigating different sex appeals appears to be important due to a relatively new trend in contemporary advertising of sexually portrayed men and because women’s responses towards those explicit various sex appeals have never been experimentally tested before. Advertisements of all categories were pre-tested among women and in the main experiment rated on purchase intentions (PI), attitude towards the brand (Ab) and attitude towards the advertisement (Aad). Underlying processes, namely evoked mood (for body display and sexual behavior appeals) and cognitive elaboration (for sexual referents appeals) were tested as well. Multivariate analyses of variance results showed that there were no significant differences between the body display and sexual referents conditions among PI, Ab and Aad. Yet, a relationship was found between respondents in the sexual behavior condition for Aad. Although there was no significant mediated effect found in the sexual behavior condition, there was an effect of the body display condition and sexual referents condition (compared to the control) on PI, Ab and Aad, which was fully mediated by mood. However, this mediated effect in the sexual referents condition contradicted the expected prediction. The study concludes by describing important information for the advertising and marketing industries, discusses limitations and future research propositions, and ends with a concluding remark.

Keywords: Sexual Appeals; Print Advertisements; Male Portrayals; Elaboration Likelihood

(3)

INTRODUCTION

Not only women but also men have become sex objects these days (Pardun, 2014). Many current online critics are discussing a “hunkvertising” trend in television commercials and argue if this rise is just a trend (Hess, 2013), to ‘shock value’ – letting people talk about the ad and subsequently about the brand (ABCnews, 2013), whether this trend possibly leads to more equality between women and men (Gianatasio, 2013) and whether the explicit sexual imaging of men could put women off (Morse, 2013).

Also in the academic field, recent studies began to focus on how men are displayed sexually in the media, while an extensive amount of the literature investigating sex appeal, mainly focused on the volume and the variety of sexual appealing portrayals of women. Reichert, Childers and Reid (2012) investigated the prevalence of visual sexual imaging in print advertising and found that the use of sex appeal significantly rose in the past years from 15% in 1987, to 27% in 2003. During the rise of sexual imaging in print advertising in 1964, 31% of women were sexually dressed compared to 6% of men. Yet, the percentage of

sexually portrayed males has grown significantly in the following twenty years. Women were sexually portrayed in almost 35% of the advertisements in 1984, whereas the percentage of sexually portrayed men increased tremendously to 14% (Reichert et al., 2012).

In line with sex appeal research, Reichert (2002) characterized five key sex appeal types in print advertising, namely Body display, Sexual behavior, Contextual factors, Sexual referents and Sexual symbolism. However no efforts have been made to examine or test the actual “appeals” in sexual advertisements (Scott & Batra, 2008). Since there are many various sex appeal types in advertising, this study aims to provide a more thorough understanding of which types of male portrayals are accepted, which types are seen as obscure or harmful and which types eventually could lead to higher product sales for brands, as this could provide very useful information for advertisers. This study therefore wants to fill a scientific gap in

(5)

sex appeal literature by being the first study to test those various male portrayals in advertising. It will introduce insights into whether these different sex appeal types can generate various attitudes in women and it will also explore two underlying processes that may provoke or stimulate these particular attitudes.

Since much prior empirical work focused primarily on the emotional processes affecting consumers’ responses that are elicited by using sex appeals (Taflinger, 1996; LaTour et al., 1990; Liu, Cheng & Li, 2009), this study will include an additional process of cognitive elaboration, as consumers’ involvement, attention and attitudes towards the advertisement may change due to particular wording (MacInnes, Moorman & Jaworski, 1991). With this study, it is thus expected that certain sex appeal types may stimulate a more cognitive process, whereas others may stimulate a more automatic, aroused process.

This research therefore goes beyond the provided information of the current

“hunkadvertising” trend, leaving the effects of those various advertising strategies using sex appeal unclear.The central research questions throughout this paper will be the following:

RQ1: To what extent are women persuaded to purchase a product by different

portrayals of attractive men in print advertisements?

RQ2: What are the roles of emotional arousal and cognitive elaboration in this

(6)

LITERATURE REVIEW

Sex Appeals in Advertising

Many advertising agencies and ad designers have used several explicit forms of sex appeal in their strategies for decades. The ultimate goal of using sex appeals in advertising is to

eventually sell more products, regardless of the product type that is being advertised

(Kilbourne, 1999; Bordo, 1999; Rohlinger, 2002). In Western countries such strategies have increased most prominently in the mid-to-late 1960s and it is argued that advertising agencies started to use these particular approaches to break through the extreme advertising clutter that had begun to rise when the ‘new’ media were introduced (LaTour & Henthorne, 1994; Liu et al., 2009). Due to the explicit visual nature of this type of advertising strategy, scholars have realized the importance of analyzing sex appeals in advertising and why sex appeals became so widely used by advertising practitioners (Scott, 1994; Nelson & Paek, 2005).

Sex appeals in advertisements are described as persuasive messages, where brand information in its context is associated with sexual information (Reichert, Heckler & Jackson, 2001). According to sexuality research, sexual information in a message is the information that could elicit sexual thoughts and/or feelings (Pardun, 2014) and is frequently used to draw attention to the product being advertised. Subsequently, ad designers have the choice to incorporate greater or lesser degrees of sexual elements as those may vary from visual images to verbal elements, or the combination of both. Sex appeal in advertising is therefore defined as “sexuality in the form of nudity, sexual imaging, innuendo, and double entendre ... employed as an advertising tool for a wide variety of products” (Courtney & Whipple, 1983, p.103).

Several elements of this definition can be categorized under different subgroups. For example, sexual content in television commercials has been coded by scholars as follows; sex objectification, the self-gazing, the presence or absence of alluring behavior, implying

(7)

provocative clothing and obvious beauty concerns (Coltrane & Adams, 1997). However, this study focused merely on television commercials instead of print magazines, which is different as print advertisements in magazines generally contain more nudity displays and sex appeals than television commercials. This is probably due to the stringent restrictions on broadcast television (Nelson & Paek, 2005). Moreover it is apparent that in the contemporary

advertising practices, the uses of overt sex appeals has grown considerably in print

advertising (LaTour 1990; Severn, Belch & Belch 1990; Soley and Reid 1988; Tinkham and Reid 1988; LaTour & Henthorne, 1994)

Other scholars, investigating print advertisements, identified four overlapping sex appeal types including; the physical features of models, behavior and movements, intimacy between more models, and contextual factors (Reichert & Ramirez, 2000). Reichert (2002) extended this study and identified five important and most common sex appeal types in print advertising; 1. Body display/nudity, 2. Sexual behavior, 3. Contextual factors, 4. Sexual referents and 5. Sexual embeds/symbolism. Based on Reichert’s (2002) study, three key sexual appeals are selected that appear most in the literature, are most visible and apparent in print advertising, and that can be investigated in an experiment (body display, sexual behavior and sexual referents).

Contextual factors are excluded since these are not specifically addressed as a component of sex appeal advertising (Reichert, 2002). They are contributing elements, enhancing and framing the sexual content that is already present in an advertisement and therefore often occur in conjunction with the other three defined sex appeals. Furthermore, sexual embeds and symbolisms are excluded as those fall within ‘subliminal advertising’, meaning that consumers are often not consciously aware of their presence. Also, this sex appeal is excluded because scholars failed to demonstrate the effects of this sex appeal type in

(8)

Different Sex Appeal Types

Numerous empirical studies indicated that the use of sex appeals in advertisements stimulate more attention, a higher favorability and enjoyment when compared to advertisements with no sex appeal(Dudley, 1999; Grazer & Keesling, 1995; Reichert et al., 2001; Wyllie, Carlson & Rosenberger, 2014). However, the study by Reichert (2002) showed that there are various sex appeal types. Due to the differences in sex appeal types, it is important to understand which sex appeal types influence attitudes.

For example, on the one hand, when nudity is used as a sex appeal, it clearly involves visual influences to a higher degree, yet on the other hand, there are sex appeal types that appear to have a lesser degree of visual sexiness, which can be described as types including suggestive sexual behavior or the use of sexual referents. Based on Reichert’s (2002) study, the three different sex appeal types that will be used in this study are described as follows:

1) Body Display. This sex appeal type describes primarily the revealing parts of the

human body. Previous studies mostly evaluated advertisements where explicitly the amount of clothes that were worn by the models, was coded (Reichert, 2002). The evaluations were based on how much of the human body was revealed in the advertisement (e.g. women: short skirts, innerwear, low-cut blouses, and men: tight jeans, underwear, or shirtless). Other scholars refer to body display in their research as ‘nudity’, and subsequently quantified the nudity of the model with regard to the amount and style of clothing and the amount of skin that was shown (Nelson & Peak, 2005). Body display is an important sex appeal type, as another analysis showed a clear trend in print advertisements where men were dressed

provocatively in revealing clothes in 18% of the advertisements in 1993, compared to 11% in 1983 (Reichert, Lambiase, Morgan, Carstarphen & Zavoina, 1999). Briefly, body display in this study means the naked revealing of the upper body part of the male model.

(9)

2) Sexual Behavior. Sexual behavior in advertisements is considered to be different

from body display as it implies the sexually provocative actions of men or women, or between them in advertisements. Reichert and Ramirez (2000) found that 39% of the respondents indicated that the acted behavior was the reason why they perceived an

advertisement as sexually appealing. This included for example the flirtatious behavior of the models based on their moving, talking, making eye contact, winking, sexual teasing, smiling, exposing neck, emphasizing cleavage, erotically dancing, which evoked sexual interest of the viewers (Givens, 1983; Fullerton & Kendrick, 2001a; Coltrane & Adams, 1997; Reichert, 2002). Although Reichert (2002) also included interaction to be part of this sex appeal type, this will not be taken into account as this research focuses primarily on the emotions aroused in women being exposed to a print advertisement with solely the projection of a male model. In short, sexual behavior in this study means the explicit suggestive behavior of the male model by looking seductively into the camera, emphasizing its flirtatious behavior.

3) Sexual Referents. Very often, advertisers make use of sexual innuendo and double

entendre in a message. Some examples of sexual innuendo used by brands are: Campari: “the first time is never the best”, M&M’s: “melts in your mouth, not in your hands” and Toyota: “oh what a feeling!” (Wroblewski, 2009). Advertisements including this sex appeal mainly use a combination of both verbal and visual message elements. Besides the setting, the location, and the model, they use sexually supporting comments, phrases or rhetorical questions with inherent multiple interpretations that can often be understood sexually by the viewer. However, less attention has been given to explore or test the interaction of visual sex appeals with verbal content (Severn, Belch & Belch, 1990). In this study, the sexual referents in advertisements specifically promote the products being advertised by using sexual

(10)

Attitudes towards Sex Appeals in Advertising

Several empirical studies investigated the attitudinal reactions of men and women towards the gratuitous use of explicit sexual images in advertising (Sengupta & Dahl, 2008). Most studies found that men generally hold a positive attitude towards sexually themed advertisements, whereas women generally hold a negative attitude towards explicit sex appeal use in advertising. It appears that men tend to find physical gratification in a sexual advertisement (Cohen & Shotland, 1996; Hill, 2002; Dahl, Sengupta & Vohs, 2009), as opposed to women who tend to focus more on the impact of intimacy and commitment in a sexual relationship (Birnbaum et al., 2006; Hill 2002; Malamuth, 1996; Schachner & Shaver 2004; Dahl et al., 2009).

Other scholars found similar results implying that women indeed show a spontaneous negative attitude towards explicit sexual images in advertising (Griffitt & Kaiser, 1978; Sengupta & Dahl, 2008; Vohs et al., 2013). This is mainly due to their different perception of what they perceive to be sexually attractive (Garrett, 1993; Liu et al., 2009) and how they have their own implicit, personal sexual motivations and desires (LaTour, 1990; Price, 2002; Liu et al., 2009). Vohs et al. (2013) relate this to the fact that the use of sexual imaging for most women is inimical, meaning that they do want to regard sex as something that is highly valued and of great worth instead of something unloved.

Yet, this study also found that women generate positive attitudinal reactions when the advertisement including sex appeal promotes an expensive product. However, the

advertisements in the study by Vohs et al. (2013) used advertisements in which there was an interaction (intimacy) between a male and female model, which is excluded as a sex appeal in this study. Nevertheless, due to the supporting findings that product status has a positive effect on women’s attitudinal reactions, it expects that women would generate positive

(11)

attitudinal reactions after being exposed to the three sex appeal types, as this study will use products that fall within a high expensive product class.

Four Purposes of Sex Appeal Advertising

The use of sex appeals in advertising tends to have various purposes (Figure 1) and various scholars studied these purposes of sex appeals independent from one another. Firstly, many previous studies have investigated the effect of sexual content by focusing on the attention that it could draw from its consumers as this is seen as the main motivation why advertising agencies include sex appeal in advertisements (Reid & Soley, 1983; Liu et al., 2009).

Secondly, it has been discussed how sexual content in advertisements can stimulate a certain

elaboration with the consumers and whether consumers are able to recall and recognize the

advertisements (Alexander & Judd 1978; Grazer & Keesling 1995; Reid & Soley 1981, 1983; Severn et al., 1990; Steadman 1969; Reichert et al., 2001). Thirdly, sexual content is used to

elicit emotions, e.g. arousal, excitement or lust. Many supporting findings suggest that sexual

content in advertisements helps to stimulate a desire for products or brands and that this type of content evokes positive arousals, evaluations, sexual thoughts and feelings (LaTour et al. 1990; Liu et al., 2009). Lastly, it has been argued that sex appeals in advertisement can result into having a favorable attitude towards the ad and attitude towards the brand and possibly lead to greater purchase intentions (Simpson, 1996; Severn et al., 1990). Also, Grazer and Keesling (1995) found that the use of sex appeal in print advertisements leads to higher purchase intentions than non-sexual advertisements. However, advertisements with moderate appeals seem to generate higher purchase intentions than other high-, low-, or no- sex

(12)

Figure 1: Various purposes of sex appeal in stages.

While it has been acknowledged that sex appeals in advertisements have the ability to easily attract attention to the ad (Reid & Soley, 1983; Liu et al., 2009), this study aims to investigate which sex appeal type has a direct influence on the purchase intentions, attitudes towards the advertisement and attitudes towards the brand of women. Moreover, it is expected that all three sex appeal types positively influence these attitudes of women as the

advertisements promote a highly expensive product (Vohs et al., 2013). Therefore, the following hypotheses are defined:

H1a: The use of explicit sexual referents of men in print advertising positively influences

the PI, Ab and Aad of women compared to the control advertisement with no sex appeal.

H2a: The use of explicit sexual behavior of men in print advertising positively influences

the PI, Ab and Aad of women compared to the control advertisement with no sex appeal.

H3a: The use of explicit body display of men in print advertising positively influences the

PI, Ab and Aad of women compared to the control advertisement with no sex appeal.

Two Underlying Processes: Elaborative and Affective

Next to the investigation of particular attitudes towards the various sex appeal types, this study aims to investigate the effects of two elicited purposes of sex appeals in advertisements and will seek for possible mediating effects of two underlying processes that could possibly influence PI, Aad and Ab. The underlying processes discussed in this research are a) a high involvement process that follows a more central route and focuses on the consumers’

(13)

cognitive elaborative processes, and b) a low involvement process that focuses more on the aroused and affective state of consumers.

Petty and Cacioppio (1981, 1983) developed the Elaboration Likelihood Model (ELM) that presents two distinct routes of how a certain attitude can be established. Basically, the ELM is useful in predicting the likelihood of attitude change that is based on the amount of elaboration evoked by that message. If people are motivated and able to process the merits of issue-related arguments in a sex appeal message, then they are more likely to take a “central

route”. Mood will have no direct effect on the attitudes of women, but the message itself will

influence their thoughts and eventually have an impact on their attitude (Petty, Gleicher & Baker, 1991). The arguments in these messages are considered to be imperative while consumers are evaluating a brand or product advertised and the information presented in an advertised message is thus central in the evaluation and comprehension of the consumer.

Also the probability of a consumer to take the central route in the ELM depends highly on several factors (MacInnes et al., 1991). The two most important factors are the motivation and the ability of the consumer to attend and therefore to comprehend a message (Scholten, 1996). With the use of sexual referents in an advertisement, for example the use of a

rhetorical question or slogan in the framing of the message arguments, it is expected that the motivation and ability of women to think about a persuasive message will be stimulated (Petty, Cacioppo & Heesacker, 1981; Petty & Cacioppio, 1983). Moreover, the complexity of the innuendo and the comprehension of this may as well add to the ability to think about the message (Regan and Cheng, 1973; Petty & Cacioppio, 1983). The sexual referents in a message are thus used to connect the consumer to the advertisement and therefore to enhance the relevance of the brand to the self, thereby motivating consumers to process the message more elaborately. An example given by MacInnes et al. (1991) is by asking a rhetorical

(14)

brand or the product. Other research also has shown supporting evidence that an

advertisement argument shaped in a rhetorical question generates more cognitive responses than a statement posed declaratively (Bumkrant & Howard, 1984; Petty et al., 1986; Swasy & Munch, 1985, MacInnes et al., 1991). With this information, the following hypothesis is expected:

H1b: The effect of explicit sexual referents on PI, Ab and Aad will be mediated by

elaboration.

If people have relatively low motivation and ability to process a message, they are more likely to take a “peripheral route”, where affect serves as a simple peripheral cue. As a consequence, mood will have a direct effect on attitudes and will not influence issue-related thoughts or arguments (Petty et al., 1991). These messages are mostly dependent on heuristic cues, associations of various message elements with the brand or product and mere exposure to the brand or the product.

In this study, the message uses a particular heuristic cue, ‘sex appeal’, in order to stimulate arousal. However, due to the low involved engagement of consumers it could either stimulate positive or negative arousal, and it may also reduce the rational thoughts that mediate the persuasiveness of the message (Reichert, 2002). Due to these distractions that are inherent in these messages, various cognitions, that could both support or counter argue, can be inhibited. Therefore, the consumer is expected to rely heavier on peripheral cues (e.g. images, sexiness, positive feeling and heuristics) (Reichert, 2002), which are in this study the explicitly used sexual appeals. Severn et al. (1990) found that respondents focused their attention more on the sexual executional elements of the advertisements (nudity and suggestive behavior), through which they detracted the processing and retention of the message arguments. It is thus expected that women are becoming emotionally aroused when

(15)

being exposed to nudity and suggestive behavior, which in turn positively influences their purchase intentions and attitudes towards the advertisement and brand. Therefore, the following hypotheses are as follows:

H2b: The effect of explicit sexual behavior on PI, Ab and Aad will be mediated by mood.

H3b: The effect of explicit body display on PI, Ab and Aad will be mediated by mood.

Figure 2 shows a concise overview of the ELM including the three sex appeal types.

Figure 2: ELM including various Sex Appeal Types.

To summarize, because of the used sexual innuendo in the sexual referent

advertisements, it is expected that consumers will follow a more highly involved/elaborative process. Consumers that are not shown these particular texts will follow a more low

involved/aroused process which is driven by the inherent heuristic cue, the used sex appeal. A

High motivation and ability to think about the message Elaboration Focused on the amount of thoughts/argum ents - Attitude towards Brand - Attitude towards the Ad - Purchase intentions Sexual referents Low motivation and ability to think about the message Mood Focused on executional cues and heuristics - Attitude towards Brand - Attitude towards the Ad - Purchase intentions Advertisement Motivation and Ability Processing Approach Attitude Change Sex Appeal Type

Body display Sexual behavior

(16)

H1a, H2a, H3a

H2b, H3b

to the three different sex appeal types. Figure 3 shows the conceptual model including all hypotheses.

Figure 3: Conceptual Framework including the Hypotheses

METHODOLOGY

Sample and Research Design

In total 223 (M = 22.53, SD = 2.78, 18–35) women were randomly allocated to one of the four sex appeal conditions in a between subjects factorial design (control, body display, sexual

behaviour, sexual referents). Only women, between the ages of 18 and 35 years, were allowed

to participate in this study. This age range was chosen since it was suggested to focus at this particular group whilst analyzing sex appeal in advertisements (Cheng, 2014). This is due to the current globalization, assuming that young people are becoming similar in terms of their media attitudes and behaviors (Maciejewski, 2004; Cheng, 2014) mainly because they possess different characteristics that may distinguish them from other demographic groups; they have reported a greater liking of advertisements compared to the other (older) groups and they are less often insulted, misled and offended by explicit advertising, which also includes sex

H1b Attractive Male Representations - Body display - Sexual Behavior - Sexual Referents - Controlled Ad

Attitude towards Ad (Aad) Attitude towards Brand (Ab)

Purchase Intentions (PI) Mood

(Arousal & Pleasure)

Elaboration

(17)

appeal advertising (Shavitt, Lowrey & Haefner, 1998; Maciejewski, 2004). Focusing on this group, Generation Y, will thus exclude the dissimilarities in women’s preferences for sex appeal advertisements that might have occurred when analyzing other segment groups as well and additionally exclude the potential errors in the measurement model.

Two respondents were deliberately left out as those were male. Additionally, two statistical tests were conducted to check for potential outliers. An univariate analysis

identified several outliers. Out of the many outliers, two respondents were removed from the dataset that appeared to be very extreme outliers. After checking their responses it was observed that they rated each construct considerably low, contrary to the other respondents. Moreover, after observing their comments, thoughts and ideas, it seemed that the two respondents were not seriously partaking in this research. Another multivariate analysis suggested leaving out two respondents who had very strange combinations of scores on the various dependent variables. The Mahalanobis distance recorded of these were 30.43 and 23.36 and therefore clearly exceeded the critical value of 20.52 (Tabachnick & Fidell, 2007).

Also, it was suggested to exclude the homosexual and bisexual respondents as in contemporary sexology ‘sexual orientation’ is referred to the stronger preference that people may have towards various sexes. This means that people are more strongly sexually aroused by members of their own sex, the opposite sex, or both sexes (Bailey & Zucker, 1995). In total, three homosexual and six bisexual respondents were excluded, in order to prevent the results from being influenced by their responses. However, the results of the analyses including the homosexual and bisexual respondents were not very different from the

MANOVA’s excluding those (Appendix A, Tables 1-5).Consequently, out of 238 responses, 223 were to be useful for this research.

(18)

Prior to the experiment, a pre-test was conducted on 95 women (M = 22, SD = 1.95, 18-30) in order to select the most suitable set of advertisements. Four different models were selected from a model agency website (Ford models, 2014) and displayed with one of the four selected expensive products (champagne bottle, fragrance, car and wristwatch) in each advertisement set (Appendix B). Vohs et al. (2013) studied the effects of sex appeal on product status and gender attitudes and found that women changed their attitude to a more favorable attitude towards sex appeal when the advertisement promoted a product that is rather expensive versus inexpensive. This pre-test therefore extends the study by Vohs et al. (2013) by representing each set with a different product that falls within a high expensive product class.

Based on the portfolio of the models, three different photographs were selected

representing; Body display (N = 27), Sexual behavior (N = 24) and Sexual referents (N = 21) / Control (N = 23) to which the female respondents were randomly allocated. The control advertisements were similar to the sexual referents advertisements, however the only differences were the attached sexual slogans or rhetorical questions in the sexual referents advertisements. All photographs across the four different sets displayed models that had similar expressions and were slightly dressed in the same manner wearing causal and

fashionable clothing (except for the models in the body display advertisements), so that each advertisement was only different for its manipulation. Each set consisted of four

advertisements per condition, meaning that respondents assigned to the body display conditions were required to evaluate the four body display advertisements, etc.

The respondents rated the advertisements on four affective items

“appealing/unappealing, impressive/unimpressive, attractive/unattractive, eye-catching/not eye-catching” (α = .819) and four cognitive items “believable/unbelievable,

informative/uninformative, clear/unclear” (α = .888) retrieved from Baker & Churchill (1997). Moreover, they rated three ad perception items “professional/unprofessional,

(19)

credible/non-credible, trustworthy/untrustworthy” (α = .911) (Yang & Oliver, 2004), and one congruency item “fitting/non-fitting” (Kamins & Gupta, 1994). All items were measured on bipolar 7-point Likert scales.

Furthermore, three items retrieved from Hebbani & Athanassiou’s (2008) content analysis, were included measuring the manipulations of each advertisement. Body display was measured on a 7-point Likert scale ranging from 1 (entirely neutral/normal) to 7 (completely nude), asking the respondents to “rate the amount of clothing in the

advertisement”. The other two items measuring sexual behavior (“rate the body language in the advertisement”) and sexual referents (“rate the word-play in the advertisement”) were measured on 7-point Likert scales ranging from 1 (neutral) to 7 (very sexual).

Results Pre-test

Moet & Chandon Set. A MANOVA with independent variables (body display, sexual

behavior, sexual referents and control) and dependent variables (PI, Aad and Ab) revealed a significant effect, F (3,89) 28.48, p < .001, Wilks’ Lambda = .15, η = .47 (Table 6).

An inspection of the mean scores indicated that respondents in the body display condition (M = 6.33, SD = 1.11) reported higher levels of nudity than respondents in the control (M = 2.39, SD = 1.23), referents (M = 2.62, SD = 1.02) and the behavior (M = 1.58;

SD = .83) conditions. Also, respondents in the referents condition reported slightly higher

nudity levels (M = 2.62, SD = 1.02) than respondents in the behavior condition (M = 1.58, SD = .83).

Respondents in the body display condition had reported higher scores of body language (M = 5.78, SD = .93) than respondents in the control (M = 4.43, SD = 1.41), referents (M = 4.29, SD = 1.74) and behavior (M = 4.71, SD = 1.46) conditions.

(20)

Lastly, in the referents condition, respondents had reported higher scores of wordplay (M = 4.52, SD = 1.33) than respondents in the control (M = 2.57, SD = 1.56), behavior (M = 1.96, SD = 1.30) and body display (M = 2.93, SD = 1.44) conditions.

Dolce and Gabbana Set. A MANOVA for this set with similar independent variables

and dependent variables revealed a significant effect, F (3,89) 24.61, p < .001, Wilks’ Lambda = .15, η = .44. Yet, there were no significant statistical differences measured among the four groups rating wordplay (Table 7).

Respondents in the body display condition had reported higher levels of nudity (M = 6.15, SD = 1.13) than respondents in the control (M = 1.52, SD = 1.08), referents (M = 1.71,

SD = .96) and behavior (M = 1.54, SD = .93) conditions.

Similar to the previous item, respondents in body display condition had reported higher body language scores (M = 5.44, SD = 1.42) than respondents in the control (M = 3.70,

SD = 1.77), referents (M = 3.62, SD = 1.53) and behavior (M = 3.58, SD = 1.61) conditions.

FIAT Set. The MANOVA for this set with similar independent variables and

dependent variables revealed a significant effect, F (3,89) 29.02, p < .001, Wilks’ Lambda = .15, η = .47 (Table 8).

Respondents in the body display condition had reported higher rates of nudity (M = 5.30, SD = 1.07) than respondents in the control (M = 1.30, SD = .88), referents (M = 1.48, SD = .98) and behavior (M = 1.29; SD = .62) conditions.

Also, respondents in the body display condition had reported higher scores of body language (M = 5.11, SD = 1.19) than respondents in the control (M = 3.78, SD = 1.65), and the behavior (M = 3.04, SD = 1.57) conditions. Also, respondents in the referents condition had reported higher body language scores (M = 4.29, SD = 1.23) than respondents in the behavior condition (M = 3.04, SD = 1.57).

(21)

Lastly, respondents in the referents condition had reported higher scores of used wordplay (M = 3.86, SD = 1.46) than respondents in the control (M = 2.17, SD = 1.64) and the behavior (M = 1.71, SD = 1.16) conditions. Also, respondents in the body display condition had reported higher amounts of used wordplay (M = 2.89, SD = 1.67) than respondents in the behavior condition (M = 1.71, SD = 1.16).

GUESS set. A MANOVA for this set with similar independent variables and

dependent variables revealed a significant effect, F (3,89) 23.50, p < .001, Wilks’ Lambda = .15, η = .42 (Table 9).

Respondents in the body display condition had reported higher rates of body display (M = 5.89, SD = 1.40) than respondents in the control (M = 1.22, SD = .52), referents (M = 1.62, SD = 1.12) and behavior (M = 1.92; SD = 1.18) conditions.

Furthermore, respondents in the body display condition had reported higher body language rates (M = 4.85, SD = 1.54) than respondents in the control condition (M = 2.96, SD = 1.67). Also, respondents in the behavior condition had reported higher body language rates (M = 4.29, SD = 1. 42) than respondents in the control condition (M = 2.96, SD = 1.67).

Lastly, respondents in the referents condition had reported higher wordplay rates (M = 4.14, SD = 1.91) than respondents in the control (M = 2.57, SD = 1.56) and behavior (M = 2.50, SD = 1.62) conditions.

There were no significant statistical differences measured among the four conditions for the Moet & Chandon, D&G and Guess sets for affective attitude towards the ad, cognitive attitude towards the ad and ad perception (Tables 10-12). Yet, differences were found in the Fiat set among all variables. There were also no significant statistical differences measured among the four groups for the Moet & Chandon and Guess sets whilst analysing perceived fit between the product and model (Table 13) but there were statistical differences in the Fiat and

(22)

set for the three manipulation items compared to the Guess set and the consistent outcomes of the other variables, the advertisements of the Moet & Chandon set were used for the main study.

Main Experiment Procedure and Measures

The survey of the main experiment was distributed online and once the respondents opened the survey, they were presented with an introduction text informing them briefly about the requirements to participate and their exposure to particular advertisements (based on the pre-test the Moet & Chandon set) and several questions. The rationale of this research was intentionally excluded as this could have affected their responses. The survey consisted of five parts (Appendix D) measuring the following item constructs:

Part I: Purchase intentions (PI). This construct was measured using three items

retrieved from Spears and Singh (2004). The items were measured on a 7-point Likert scale ranging from 1 (very unlikely) to 7 (very likely), asking the respondents to indicate how likely they are to behave in the following situations: “I will probably buy this Moet Chandon ICE Imperial bottle”, “I intend to buy this Moet Chandon ICE Imperial bottle” and “I am definitely buying Moet Chandon ICE Imperial bottle” (α = .934).

Attitude towards the Brand (Ab). Three items were retrieved from MacKenzie and

Lutz (1989) for measuring respondents’ attitude towards the brand. Respondents were required to fill out on 7-point bipolar Likert scales the likelihood of them drinking a glass of Moet Chandon ICE Imperial. This construct included the items: “bad/good,

unpleasant/pleasant and unwise/wise”. Also, one additional item was added, namely “disadvantageous/advantageous” (α = .787).

Attitude towards the Advertisement (Aad). Seven items retrieved from Tuten (2005)

(23)

following items: “bad/good, unpleasant/pleasant, unfavorable/favorable, unconvincing/convincing, unbelievable/believable, biased/unbiased and

unprofessional/professional” (α = .888).Again, all items were measured on bipolar 7-point Likert scales.

Part II: Cognitive Elaboration. The thought-listing technique is a common measure of

elaboration in advertising research (Aaker & Sengupta, 2000; Maheswaran & Chaiken, 1991; Mothersbaugh, Huhmann & Franke, 2002; Severn et al., 1990; Huhmann & Mott-Stenerson, 2008). It was used in this study to measure respondents’ cognitive elaboration, as it is assumed that the psychological significance of an individual’s thoughts and feelings can be examined by analyzing how they have reported their own ideas and feelings and thoughts (Cacioppo, Hippel & Ernst, 1997). After being exposed to the advertisement and filling out several questions regarding PI, Ab and Aad, respondents were presented with a short text in which they were asked to list any thoughts or ideas that came to mind while they viewed the advertisement. Each thought was recorded on a separate line. Besides this, they themselves were responsible for moving each idea into either a ‘positive’, ‘neutral’ or ‘negative’ box. In order to process these thoughts, the total amount of thoughts were important and the amount of thoughts listed in these various boxes, as the number of cognitive responses listed indicate the elaborative processing of the respondent (Aaker & Sengupta, 2000; Miniard et al., 1990; Mothersbaugh et al., 2002; Huhmann & Mott-Stenerson, 2008).

Part III: Mood. According to Olney, Holbrook and Batra (1991), two dimensions

(arousal and pleasure) can construct ‘mood’. Originally, the assessing of mood comes from the Affect Grid, which is a scale designed for assessing affect (Russell, Weiss & Mendelsohn, 1989). The Affect Grid (Figure 4) assesses the two emotions in a clockwise direction (from the upper left to the right) in which the two major dimensions are bipolar opposites (Coan &

(24)

Figure 4: Affect Grid (Russell et al., 1989)

Based on the arguments by Russell et al. (1989) there are grounds for constructing and using this instrument measuring arousal and pleasure simultaneously. This excludes the risk of confounding one dimension with one another (Russell et al., 1989). Items for this study are retrieved from previous sex appeal studies (Olney et al., 1991; Simpson, Horton & Brown, 1996) as they suggested that the two dimensions appear to be relevant to studies investigating sex appeal and more specifically studies containing ‘nudity’ (LaTour et al., 1990; Severn et al., 1990; Simpson, Brown & Widing, 1998). Moreover, this assessment technique of mood appears to be very useful in detecting frequent short-term change in mood states (short exposure to the ad) and for studies that require manipulation checks (Coan & Allen, 2007).

Arousal. The following six items were used to measure arousal on bipolar 7-point

Likert scales, asking respondents to rate their emotions according to the way they made them feel: “sluggish/frenzied, calm/excited, dull/jittery, relaxed/stimulated, sleepy/wide awake and non-aroused/aroused” (α = .887) (Olney et al., 1991; Simpson et al., 1996).

Pleasure. For measuring pleasure, respondents were asked to rate their emotions on

six items indicating how the advertisement evoked the following feelings: “unhappy/happy, annoyed/pleased, unsatisfied/satisfied, despairing/hopeful, melancholic/contented and

(25)

bored/relaxed” (α = .874) (Olney et al., 1991; Simpson et al., 1996). Again, these items were measured on bipolar 7-point Likert scales.

A factor analysis revealed that three items of the arousal dimensions did not load on the same factor and therefore the items “non-aroused/aroused”, “relaxed/stimulated” and “dull/jittery” were excluded from further analysis. The reliability score of the new variable ‘mood’ showed a high score for the remaining nine items, namely α = .906.

Part IV: Manipulation checks. The body display manipulation check was retrieved

from the pre-test. However, the items measuring sexual behavior and sexual referents were changed, as the wording in the pre-test appeared to be slightly unclear for the respondents. The item measuring sexual behavior was changed into “How would you describe

the flirtatiousness of the model in the advertisement….”, and the item measuring sexual referents was changed into “How would you describe the used words and sentences (double entendre) that are used in the advertisement…”. The scales remained the same, ranging from 1 (very neutral) to 7 (very sexual).

Furthermore, an extra question was included measuring the overall sexiness of the advertisement (Putrevu, 2008). They were required to answer the following question on a 7-point Likert scale: “I would describe the ad as... 1 (not very sexual) to 7 (very sexual)”.

Part V: Demographic questions. This part included several general demographic

questions, such as; age, sexual orientation, nationality, level of education and proficiency in English. Also, the respondents had an opportunity to fill out any comments related to the questionnaire or the study.

Willingness to pay. As an additional question, respondents were asked to select either

a Moet & Chandon ICE Imperial bottle or a L’Oreal makeup set (both having a value of €49,95) by answering the following question: “If it would be your birthday (or another

(26)

required to fill out for both gifts the following question: “Please indicate the importance for you to get those gifts…” on a bipolar 7-point Likert scale ranging from 1 (not at all

important) to 7 (extremely important).

FINDINGS & ANALYSES

The female respondents (N = 223, M = 22.53, SD = 2.78, 18-35) were randomly allocated to one of the four conditions (body display, N = 64; sexual behavior, N = 58; sexual referents, N = 53; control, N = 48) and most respondents appeared to be Western Europeans (95.5%), 23 years old (20.2%), had a proficiency in English that was fluent (40.8%) and obtained a Bachelor’s degree (45.7%) as their highest education level.

Manipulation Checks

A MANOVA was performed to investigate the manipulation checks and extra question variable differences in the four conditions. A statistically significant difference between the four conditions was found on the combined dependent variables, F (4, 216) = 37.93, p < .001; Wilks’ Lambda = .21; η = .40. When the results for the dependent variables were considered separately, there were differences among all groups (Table 14).

Amount of Clothing. A closer look at these mean scores F (3, 219) = 150.81, p < .001,

η = .67, indicated that respondents in the body display condition reported slightly higher levels of nudity (M = 5.86, SD = 1.36) than respondents in the sexual behavior (M = 1.81, SD = .10), the sexual referents (M = 2.08, SD = 1.33) and the control (M = 2.13, SD = 1.20) conditions.

Flirtatiousness of the Model. For this manipulation check, F (3, 219) = 17.00, p <

(27)

1.28) and sexual behavior (M = 5.36, SD = .83) conditions reported slightly higher levels of flirtatiousness of the model than respondents in the control (M = 4.02, SD = 1.55) and the sexual referents (M = 4.40, SD = 1.71) conditions.

Sexual Wordplay. The scores for this manipulation check, F (3, 219) = 15.12, p <

.001, η = .17, indicated that respondents in the sexual referents condition reported slightly higher levels of sexual wordplay (M = 4.55, SD = 1.45) than respondents in the control (M = 2.58, SD = 1.67), the sexual behavior (M = 2.72, SD = 1.58) and the body display (M = 3.30,

SD = 1.90) conditions.

Extra Question. The mean scores for this question, F (3, 219) = 14.05, p < .001, η =

.16, showed that respondents in the body display condition reported higher levels of ad sexiness (M = 5.58, SD = 1.11) than respondents in the control (M = 4.31, SD = 1.40), sexual referents (M = 4.36, SD = 1.40) and sexual behavior (M = 4.90, SD = .87) conditions.

Dependent Variables

Another MANOVA was performed to investigate various item variable differences (time-clicking, willingness to pay, importance of getting gifts) in the four conditions. There was no statistically significant difference between the four conditions on the combined dependent variables, F (5, 210) = 1.37, p = .155; Wilks’ Lambda = .91; η = .03 (Table 15).

Willingness to Pay. However, with a marginal alpha (p < .01), there was a statistical

significant difference found for willingness to pay, F (3, 214) = 2.15, p = .095, η = .03. The inspection of the mean scores indicated that respondents in the body display condition

reported slightly higher levels of willingness to pay (M = 29.62, SD = 15.13) than respondents in the control condition (M = 23.88, SD = 12.37).

(28)

gifts (Moet & Chandon and L’Oreal) in the four conditions, χ² (3, n = 223) = 6.16, p = .10, phi = .16). However, the item which measured the importance of getting the Moet & Chandon gift was significant: F (3, 214) = 2.77, p = .043, η = .04. The mean scores indicated that

respondents in the body display condition reported a higher importance of receiving a Moet bottle (M = 3.98, SD = 1.31) than respondents in the control condition (M = 3.33, SD = 1.34).

Furthermore, a MANOVA was performed to investigate the dependent variable differences (PI, Aad and Ab) in the four conditions. There was a statistically significant difference between the four conditions on the combined dependent variables, F (3, 217) = 2.10, p = .028; Wilks’ Lambda = .92; η = .03 (Table 16).

Aad. With a marginal alpha (p < .01) there was a statistical significance difference for

Aad, F (3, 219) = 2.31, p = .077, η = .03. The inspection of the mean scores indicated that respondents in the sexual behavior condition reported slightly higher levels of Aad (M = 4.91,

SD = .87) than respondents in the control condition (M = 4.46, SD = .89).

Mediators

Mood. An ANOVA was performed to investigate the differences in mood in the four

conditions, F (3, 219) = 3.60, p = .014, η = .05 (Table 17). An inspection of the mean scores indicated that respondents in the body display condition reported higher levels of mood (M = 4.75, SD = .96) than respondents in the control condition (M = 4.40, SD = .89).

Elaboration. A MANOVA was performed to investigate the differences in the

amounts of thoughts elicited by the respondents in the four conditions. Four dependent variables were tested (total amount of thoughts, positive thoughts, neutral thoughts and negative thoughts). There were no statistically significant differences found between the four conditions on the dependent variables, F (4, 175) = .89, p = .558; Wilks’ Lambda = .94; η = .02 (Table 18).

(29)

Body Display. A positive direct effect in the body display condition was found on

mood (coded control condition = 0; and body display condition = 1; b = 0.60, t = 3.05, p = .003). The effect of the body display condition on PI was fully mediated by mood as confirmed in a bootstrapping analysis (10000 samples) with a significant indirect effect of 0.5579, with a 95% confidence interval ranging from .1924 to 1.0634 (Table 19). Also, the effect of the body display condition on Ab was fully mediated by mood as confirmed in a bootstrapping analysis (10000 samples) with a significant indirect effect of 0.2692, with a 95% confidence interval ranging from .0969 to .5177 (Table 20). The effect of the body display condition on Aad was also fully mediated by mood as confirmed in a bootstrapping analysis (10000 samples) with a significant indirect effect of 0.3979, with a 95% confidence interval ranging from .1442 to .7074 (Table 21).

However, no indirect mediated effects were found for elaboration, these indirect effects <| .02 | and all confidence intervals contained zero. Though, negative direct effects were found of respondents in the body display condition on elicited negative thoughts (b = -.49, t = -1.85, p = .067) and directs effects on PI and Aad, meaning that there were for these two dependent variables significantly more negative thoughts elicited in the control condition than in the body display condition. However, no indirect effects were found, therefore the effect of the body display condition (compared to control) was not mediated by negative elaboration.

Sexual Behavior. No positive direct effect in the sexual behavior condition was found

on mood (coded control = 0 and sexual behavior condition = 2; b = 0.17, t = 0.93, p = .355). Yet, direct effects of mood on PI (b = 0.76, t = 5.34, p < .001, Table 22), Ab (b = .40, t = 4.17,

p < .001, Table 23) and Aad (b = 0.71, t = 9.13, p < .001, Table 24) were found, but the

(30)

elaboration; these indirect effects <| .00 | and all confidence intervals contained zero, which also accounted for the positive <| .02 | and negative <| .04 | elicited thoughts.

Sexual Referents. There was found a direct effect of the sexual referents condition on

mood (coded control = 0 and sexual referents condition = 3; b = 0.46, t = 2.28, p = .025). More importantly, the effect of the sexual referents condition on PI was fully mediated by mood as confirmed in a bootstrapping analysis (10000 samples) with a significant indirect effect of 0.4213, with a 95% confidence interval ranging from .0617 to .8603 (Table 25). The effect of the sexual referents condition on Ab was also fully mediated by mood as confirmed in a bootstrapping analysis (10000 samples) with a significant indirect effect of 0.1949, and a 95% confidence interval ranging from .0393 to .4546 (Table 26). Lastly, the effect of the sexual referents condition on Aad was fully mediated by mood as confirmed in a

bootstrapping analysis (10000 samples) with a significant indirect effect of 0.3011, with a 95% confidence interval ranging from .0477 to .5821 (Table 27). Furthermore, no mediated indirect effects were found for elaboration, these indirect effects <| .00 | and all confidence intervals contained zero, also for positive <| .03 | and negative <| .00 | elicited thoughts.

DISCUSSION AND CONCLUSION

The rising trend of male portrayals with explicit suggestive posing in advertising in the

contemporary media was the reason to test various sex appeal types. The purpose of this study was to investigate whether women responded differently to different portrayals of men

including various types of sex appeal (body display, sexual behavior and sexual referents) and whether this could influence their Aad, Ab and PI. Also, it was investigated which of these sex appeal types possibly influence underlying processes, either affective or cognitive.

The female respondents in the body display (nudity) and sexual referents (sexual innuendo) conditions appeared not to be affected by the used explicit sex appeal, when

(31)

compared to the no-sex appeal advertisement, influencing PI, Aad and Ab. Therefore, both H1a and H3a are rejected. Yet, both the body display and sexual referents conditions

(compared to the control condition) appeared to be fully mediated by mood on influencing PI, Aad and Ab. H3b is thus accepted, but H1b rejected. Although it was hypothesized that respondents in the sexual referents condition would generate more cognitive elaboration due to the attached rhetorical question in the advertisement, a mediated effect of mood was found instead of cognitive elaboration and thus contradicts the prediction. Lastly, for the female respondents in the sexual behavior (flirtatious) condition, the explicit sex appeal appeared to have an influence on Aad, but not on their PI and Ab. It can be concluded that H2a is partially accepted, yet, the effect of the sexual behavior condition (compared to the control condition) on influencing PI, Aad and Ab was not mediated by mood. Therefore, H2b is rejected. Besides the findings relating to the research questions, it was found that although respondents had no special interest in receiving either a Moet & Chandon bottle or L’Oreal set, that respondents in the body display condition did found it more important to receive a Moet & Chandon bottle compared to the respondents in the control condition with no sex appeal. Moreover, the respondents in the body display condition (€29,62) were willing to pay nearly 6 euros more for a Moet & Chandon bottle compared to the respondents in the control condition (€23,88). This finding provides highly valuable information for economists as it is shown that nudity in advertisements can influence the buying behavior of women. Even though this is not a significant finding, it seems to indicate that the initial aim of sex appeal to ‘sell more products’ (Rohlinger, 2002) is confirmed to some degree. Also, the significant outcome of the extra question measuring overall sexiness in the body display condition emphasizes that the fact that nudity in advertising sells. Although women were well aware of the sexual intention of the ad, it positively affected their mood and influenced their purchase

(32)

Limitations and Future Implications

Although, it was speculated in prior empirical work that the use of sentences, slogans, headlines and rhetorical questions would activate cognitive processing (MacInnes et al., 1991), a contradicting result was found in this study. Due to a rhetorical question, “Dare to

break the ICE?”, it was expected that this advertisement would offer promises, both explicitly

and implicitly through the use of wording for some kind of sexual fulfillment (Scott & Batra, 2003) for the female respondents.

A reason for the nonappearance of cognitive elaboration in the sexual referents condition could be related to the fact that the use of a rhetorical question in a sex appeal advertisement does not stimulate cognitive elaboration. A possible explanation could be that the male model in the advertisement, due to its physical attractiveness, may over-excite the evoked pleasure and arousal of respondents and therefore neglect the cognitive processing (Bower & Landreth, 2001). Meaning that the presence and portrayal of the male model in the advertisement acted as a ‘sex appeal’ on its own and therefore intensified the effect that sexual appeals have on mood. Although the male model in the advertisement in the sexual referents condition did not explicitly suggest any sexual act, this explanation could relate to existing findings suggesting that, when motivation is reduced through a possible used sex appeal (in this case physical attractiveness), respondents are more influenced by peripheral cues influencing affective reactions and not by extensive message elaboration, preventing respondents from thinking deeply about the advertisement (Reichert et al., 2001).

Also, the study by Severn et al. (1990) was one of the first to combine and test these two elements. Their results showed that when the advertisement is highly sexual in its nature, the processing of people tends to focus more on the execution of the message, drawing the cognitive processing away from the message included in the advertisement. Since the sex appeal in tis study was not highly sexual in its nature, more additional research is needed to

(33)

test whether sexual referents have any impact on cognitive processing, as there are no supporting results found in this study.

Another reason for not finding cognitive elaboration could be related to the used rhetorical question in this advertisement. Although the manipulation check, measuring sexual wordplay/double entendre, in the pre-test and main experiment appeared to be significantly different from the other conditions, it could be argued that the rhetorical question in the advertisement was not inviting enough to let respondents think critically about the question. This particular chosen rhetorical question could therefore have been the reason why there was no mediated effect of cognitive elaboration found.

Together with the previous explanation, it is suggested for future investigations to test the effectiveness of different sexual marketing messages before analyzing possible

elaboration effects on consumers. Including more follow-up questions relating to the

rhetorical question could reveal more insights on how respondents respond to sexual referents as a sex appeal in advertisements.

While many previous studies found supporting findings for the mediation of mood and arousal related to sex appeal (Olney et al., 1991; Simpson et al., 1996), this study only found significant findings for the body display and sexual referents conditions. It could be argued that the two dimensions, pleasure and arousal, did not accurately measure what was expected. Therefore, the constructed mediator in this study could have affected the outcomes as opposed to the assumption of Russell et al. (1989). Two sex appeal studies that included ‘mood’ and found significant results were Simpson et al., (1996) – which used mood as a dependent variable, and Olney et al. (1991) – which analyzed pleasure and arousal as two separate mediators. Thus, although sexual behavior advertisement appeared to be significantly manipulated, a possible clarification could be that pleasure and arousal should be measured

(34)

It could also be argued that the suggested “hunkvertising” trend is solely applicable to television commercials instead of print advertisements. Buchholz and Smith (1991) found that respondents’ elaboration might be very different when another medium is used due to

differences in peoples’ media involvement (Cheng, 2014). For example, a television commercial involves more stimulation of various sensory modes (e.g. hearing, viewing, reading) and is found to be highly superior in evoking cognitive responses, but not

particularly in evoking affective responses (Dijkstra, Buijtels & van Raaij, 2005) compared to print advertisements. They also found that product involvement seems to determine both advertisement affect and brand affect. It is therefore recommended for future studies to incorporate product involvement as an additional variable, as this may have influenced the mediated results of this study. Moreover, future studies could test the effectiveness of the sex appeal types for television commercials, examining, for example, whether a television

commercial could elicit more cognitive stimulation in the sexual referents condition compared to the other sex appeal types as this is suggested by Buchholz and Smith (1991).

Another explanation for not establishing significant effects in this study can be related to the image manipulation of the advertisements as previous scholars also failed to code the elements of the used various sex appeals in a way that it would discern the nuances. Reichert et al. (2012) mentioned that the types of sexual imaging may not always be mutually

exclusive and that some of the various forms of sexual content outlined in this research could overlap as these can rarely appear in isolation. Since no previous studies ever examined the effects of different sex appeal types, manipulation items were retrieved from a coding scheme of a content analysis to test the sex appeal types (Hebbani & Athanassiou, 2008).

There is also a possibility that the respondents did not found the advertisements credible or believable enough, which could have influenced their PI, Aad and Ab. Another additional analysis was therefore conducted in order to check whether the effect of the

(35)

conditions on PI, Aad and Ab remained significant after controlling for perceived credibility and believability of the respondentsa. The results indicate different significant effects after controlling for these two items, meaning that the credibility and believability of respondents had an impact on PI, Aad and Ab. Although the material showed successful image

manipulations (and remains interesting for follow up studies), it is suggested for future studies to enhance the wording of the manipulation checks and the trustworthiness and presentation of the existing material in order to obtain more significant effects.

Furthermore, Liu et al. (2009) questioned whether Aad, Ab could be related to predicting PI as they found no significant differences for their groups on PI. A correlation analysis of the dependent variables PI, Aad and Ab showed a moderate correlation (Pallant, 2011)b. It is therefore suggested to carefully select the dependent variables in future

investigations as it would probably result into more significant findings and insights into the effects of different sexualized portrayals of men on women for the marketing and advertising fields. Nevertheless, this study was one of the first to test the effectiveness of various sex appeal types on PI, Aad and Ab.

Moreover, for future researchers it would be interesting to test the effects on PI, Aad and Ab among distinct age groups. As discussed, people in the Generation Y segment appear to be different from other demographic groups, considering their personal values, preferences and their tolerant attitudes towards sexual appeals in advertisements (Maciejewski, 2004). Therefore, it would be interesting to check for possible differences among different

a

The MANCOVA table can be found in Appendix G (Table 28). In the main experiment, only a marginal effect was found for respondents in the sexual behavior condition on Aad in the MANOVA. After controlling for the covariates ‘believability’ and ‘credibility’ (holding them constant), PI (marginally) and Ab appeared to be significant.

(36)

demographic groups. Moreover, this could enhance the external validity of the experiment by generalizing this research to other groups consisting of younger or older respondents.

Another opportunity to enhance the external validity of this study is to apply this research to a real life setting. Although respondents were asked to look at the advertisement as if they were flipping through a magazine, a real life study would be more effective as

respondents would report more valid responses. Nevertheless, using an experimental setting for this particular study was suggested, as it is a frequently used method for discovering the impact of sex appeals on consumers (Chan, Diehl & Terlutter, 2007; Liu et al., 2009; Cheng, 2014) and to capture various dimensions of sex appeals (Cheng, 2014), which was thus highly applicable to the purpose of this study.

Lastly, it is suggested for future researchers to incorporate other mediators, such as product involvement (Dijkstra et al., 2005) or negative affect/disgust (Vohs et al., 2013). Such variables could have possibly intervened with the variables of this study and potentially affected the outcomes. As already discussed, Dijkstra et al. (2005) found that product involvement determines peoples’ affective reactions towards the advertisement and brand. Including product involvement will provide more insights on how people personally perceive a print advertisement including sex appeal. Moreover, Vohs et al. (2013) included ‘negative affect’ as a mediator (e.g. measuring how upset, disgusted, unpleasantly surprised, and angry participants felt). Since this study focused only on ‘positive’ affective reactions, respondents may have rated their emotions differently on negative affective items. It is therefore

interesting to analyze whether the effect of the three sex appeal types on PI, Aad and Ab, could have been mediated by negative affect, which is opposed to the research aim of this study.

(37)

Conclusion

This study stresses the importance of more research on the effects of different sexual portrayals of males on advertisement and brand attitudes. Nevertheless, it is interesting that these findings, although some predictions were rejected, strongly support existing literature suggesting that print advertisements including male nudity congruent with high class products enhances positive attitudes and buying behaviors of women.

Most importantly, this study showed that the explicit use of various sex appeals can have various outcomes. Advertisers, media planners and advertising agencies should thus understand the strengths, weaknesses and differences of each sex appeal type measured in this study in order to effectively target their sex appeal advertisements on consumers. Yet, more research is required on this topic to make more conclusive claims and actual implications to relate this to involved advertising and marketing sectors. Nevertheless, Morse’s (2013) argument whether “hunkvertising” in advertising is ‘putting women off’ is clearly refused as this study showed that male nudity in advertisements has a positive effect on women’s purchase intentions and its antecedents, at least for Generation Y.

(38)

REFERENCES

Abcnews (2013). Sexy sells: hunk heats up new kraft ad. Retrieved from:

http://abcnews.go.com/blogs/entertainment/2013/04/sexy-sells-hunk-heats-up-new-kraft-ad/

Bailey, J. M., & Zucker, K. J. (1995). Childhood sex-typed behavior and sexual orientation: A conceptual analysis and quantitative review. Developmental Psychology, 31(1), 43-55. doi:10.1037//0012-1649.31.1.43

Baker, M. J., & Churchill, G. A. (1997). The Impact of Physically Attractive Models on Advertising Evaluations. Journal of Marketing Research, 14(4), 538-555.

doi:10.2307/3151194

Bower, A. B., & Landreth, S. (2001). Is beauty best? Highly versus normally attractive models in advertising. Journal of Advertising, 30(1), 1-12.

doi:10.1080/00913367.2001.10673627

Buchholz, L. M., & Smith, R. E. (1991). The role of consumer involvement in determining cognitive response to broadcast advertising. Journal of Advertising, 20(1), 4-17. Retrieved from http://www.jstor.org/stable/4188783

Cacioppo, J. T., Hippel, W. V., & Ernst, J. M. (1997). Mapping cognitive structures and processes through verbal content: The thought-listing technique. Journal of Consulting

and Clinical Psychology, 65(6), 928-940. doi:10.1037//0022-006X.65.6.928

Coan, J. A., & Allen, J. J. (2007). Handbook of emotion elicitation and assessment. Oxford, New York: Oxford University Press.

Cheng, H. (2014). The handbook of international advertising research. Bognor Regis, Sussex: John Wiley & Sons.

Collins, R. L. (2011). Content Analysis of Gender Roles in Media: Where Are We Now and Where Should We Go? Sex Roles, 64, 290-298. doi:10.1007/s11199-010-9929-5