Sexual Appeals in Advertising: The Role of Nudity, Model Gender, and Consumer Response

Abstract

This study examines whether sexual appeals, specifically nudity (body coverage) and model gender, shape consumer responses to advertising. Guided by the Elaboration Likelihood Model, we test whether these factors operate as peripheral cues when argument strength is minimal (fragrance advertising). In a 3 (model gender: male, female, both) × 2 (clothing: swimwear vs. outdoor wear) between-subjects experiment (n = 195), participants viewed one of six real advertisements from the same luxury fragrance brand featuring the same professional models. Nudity level did not affect attitudes toward the ad, brand attitude, or purchase intention. By contrast, ads depicting both a male and a female model produced small but reliable omnibus improvements in brand attitude and purchase intention relative to single-model ads; attitudes toward the ad were unaffected. Mediation tests indicated that these differences were not explained by attitude toward the ad or brand attitude, consistent with peripheral-cue processes rather than the classic ad → brand attitude sequence. Practically, the results challenge the assumption that “more nudity” is persuasive and suggest that, in low-argument contexts, gender-balanced model composition can be a modest, context-dependent cue. Advertisers should prioritize brand/category fit, pretest in the intended media environment (and locale), and expect incremental rather than large effects.

1. Introduction

Advertising plays a central role in introducing brands to consumers and shaping their perceptions and behavior (Vakratsas & Ambler, 1999; Singh & Cole, 1993). In addition to conveying basic product information such as price, availability, and benefits, advertisements aim to influence consumer outcomes such as brand image, awareness, loyalty, and ultimately, purchase intentions (Reichert et al., 2007). A key determinant of advertising effectiveness is the viewer’s attitude toward the ad itself, a construct grounded in consumers’ emotional, cognitive, and behavioral responses (MacKenzie & Lutz, 1989; Mehta, 2000; Prendergast et al., 2009; Vakratsas & Ambler, 1999; Fennis & Stroebe, 2020). Positive attitudes can enhance brand image, while negative reactions may undermine advertising goals (Spears & Singh, 2004; Khalid et al., 2023). Emotional responses, such as irritation, enjoyment, or offense, are especially impactful in shaping consumer reactions, particularly in digital and cross-cultural contexts (Mahmoud, 2013). Meta-analyses confirm that advertising inputs, including message content and source characteristics, significantly affect cognitive and affective responses, although the effects vary in strength (Eisend & Tarrahi, 2016). Building on this literature, the present study focuses on a distinct and increasingly prevalent form of advertising content: sexual appeals, specifically nudity, and how they affect consumers’ evaluations of advertisements and brands. This focus also aligns with recent calls in management and marketing research to better understand how emotional, perceptual, and identity-based responses influence consumer behavior (König, 2025).

1.1. Sexual Appeals in Advertising

Sexual appeals in advertising, brand messages that use sexualized imagery, narratives, or sounds, are not a new phenomenon, but their use has increased in recent decades, particularly in visual media such as magazines and television commercials (Reichert et al., 2001; Reichert & Carpenter, 2004; Reichert et al., 2012; Dianoux & Linhart, 2010). Since the late 1960s, the inclusion of sexual content in ads has been a subject of controversy and sustained scholarly interest (Pardun & Forde, 2006). While sexual appeals can be effective in some contexts, marketers must carefully consider when and where such content is appropriate. The effectiveness of sexualized imagery depends not only on the product category but also on the advertising platform and brand identity. For instance, luxury fashion brands like Gucci and Chanel have faced criticism for using sexualized content on social media, where such appeals may be perceived as incongruent with the brand’s image and can weaken consumers’ emotional connection to the brand (Choi et al., 2022). Despite the debates, advertisers often use sexual appeals to capture consumer attention and stimulate favorable evaluations, interest, and purchase intentions (Khalid et al., 2023). Moreover, advertising strategies that harmonize emotional content with brand identity, particularly in the luxury fashion sector, have been shown to enhance engagement and brand loyalty (Montanés-Sanchez et al., 2025).

Many advertisers appear to believe that sexual appeals increase sales (Fennis & Stroebe, 2020). Supporting this belief, Reichert et al. (2012) conducted a content analysis of magazine advertisements from 1983 to 2003 and found that the use of sexual content nearly doubled during that period, rising from 15% to 27% of ads featuring provocative clothing or suggestive poses. Notably, this increase occurred almost exclusively among female models, while the use of sexualized male models remained virtually unchanged at around 6%.

1.2. The Role of Nudity

A primary benefit of using sexual appeals in advertising lies in their ability to capture attention. According to Fennis and Stroebe (2020), sexual imagery functions as a visual cue that draws the viewer in, an essential first step in persuasion. If the message fails to attract notice, it is unlikely to have any persuasive effect, even at a subconscious level. From this perspective, nudity may sell simply because attractive models succeed in grabbing consumer attention. However, while nudity can increase arousal, engagement, and memorability (Bello et al., 1983; Dudley, 1999; Reichert et al., 2001; Reichert & Carpenter, 2004), it may also distract from the core message. Viewers may focus on the model or the sexual content rather than the advertised brand (Reichert & Carpenter, 2004; Choi et al., 2016), reducing their ability to recall product-related information (Terlutter et al., 2021). In addition, sexual imagery often communicates little about the brand or product itself (Reichert et al., 2007).

Advertisers may nevertheless use nudity to attract viewers or appeal to audiences who respond favorably to such content. While controversial, some studies suggest that nudity can enhance purchase behavior and brand usage, particularly when it resonates with the target audience (Reichert & Carpenter, 2004; Choi et al., 2016). Viewer reactions to nudity often hinge on the emotional arousal elicited by the advertisement. Positive arousal can enhance attitudes toward the ad, whereas negative arousal or emotional discomfort can provoke unfavorable responses (LaTour, 1990). When sexual content triggers feelings of offense, viewers may not only form negative evaluations of the advertisement but also reject the product it promotes (Terlutter et al., 2021). Such reactions raise important concerns for advertisers, as negative emotions can undermine the overall effectiveness of the campaign (Beales et al., 1981; Calfee & Ringold, 1988; Pollay & Mittal, 1993).

Despite the controversy, nudity has become common in certain product categories, such as cosmetics, fashion, and perfumes. For instance, Dove’s “Real Beauty” campaign featured real women of various ethnicities and body types confidently posing in white underwear, challenging conventional beauty standards while still employing nudity as an appeal (Strategy Online, 2023). Other brands, including Victoria’s Secret, Calvin Klein, and Abercrombie & Fitch, have built and sustained sexualized brand identities through consistent use of provocative imagery (Choi et al., 2022; Reichert et al., 2007).

To provide a theoretical framework for these mixed findings, the Elaboration Likelihood Model (ELM; Petty & Cacioppo, 1986) offers a useful perspective. The model has recently been applied in diverse marketing contexts such as social media personalization (Haq et al., 2025), viral marketing (Kraiwanit et al., 2025), and eco-labeling and recycling behaviors (Nguyen & D’Souza, 2025), cross-border e-commerce (Zhang et al., 2025b), and sustainability signaling in hospitality (Qin et al., 2025). Together, these studies highlight the continuing relevance of ELM in explaining how consumers process persuasive messages. When message arguments are weak or minimal, as is often the case in fragrance advertising, consumer responses rely on peripheral cues such as model attractiveness, inclusivity, and affective reactions. From an ELM perspective, nudity may function as one such peripheral cue, but its effects are likely to depend on how audiences interpret the broader social context of the advertisement. From a complementary perspective, Objectification Theory (Fredrickson & Roberts, 1997) suggests that gender-balanced portrayals, for example, featuring both a male and a female model, may attenuate perceptions of objectification (especially of women) that can elicit negative affect and reactance. Reducing such perceptions is consistent with more favorable consumer responses (Cervone et al., 2025; Liu et al., 2025; Zhang et al., 2025a).

1.3. Audience Characteristics and Gender Differences

The impact of nudity in advertising on viewer attitudes is influenced by individual characteristics, particularly age and gender (Dianoux & Linhart, 2010). In addition, the ELM framework suggests that the composition of models may act as an especially influential peripheral cue (Petty & Cacioppo, 1986). Featuring both a male and a female model can increase self-relevance for a broader audience, as more viewers see someone with whom they can identify (Hu et al., 2025). At the same time, such balanced portrayals may reduce perceptions of objectification or stereotyping, leading to more favorable affective responses (Segura-Nebot et al., 2025). Thus, model composition may be more consequential for brand attitudes and purchase intentions than nudity level per se, and these effects may also interact with audience characteristics (Trkulja et al., 2024). For instance, attitudes toward nudity tend to become more negative as individuals age (Reichert, 2002). This trend is reflected in advertising practices: younger characters are more frequently depicted nude than older ones, implying that advertisers may be responding to or reinforcing greater acceptance of nudity among younger audiences (Matthes & Prieler, 2020).

Audience evaluations often differ by gender, with both the viewer’s gender and the gender of the model featured in the advertisement playing a role. Importantly, sexualized portrayals in advertising have overwhelmingly featured women, reflecting a persistent gender imbalance (Choi et al., 2016; Matthes & Prieler, 2020). For instance, women were three times more likely than men to be depicted in a sexualized manner in 1993, despite similar overall representation in advertisements (Pardun & Forde, 2006). The proportion of sexualized female portrayals increased from one-third in 1964 to one-half by 2003 (Reichert et al., 2007). In contrast, male nudity was rare until brands like Calvin Klein and Coca-Cola began to feature it in the 1990s (Simpson et al., 1996).

Men and women tend to respond differently to advertisements that feature sexual appeals, particularly those involving female nudity. Numerous studies have documented gender-based differences in emotional and cognitive reactions (Ford & LaTour, 1993; LaTour, 1990; LaTour et al., 1990; Pradhan, 2016). For example, women generally experience greater arousal but also more negative emotions when exposed to explicit female nudity in print ads, whereas men tend to react more positively and energetically (LaTour, 1990; Ford & LaTour, 1993; Choi et al., 2016; Peterson & Kerin, 1977). Viewers also tend to respond more favorably to sexualized models of the opposite sex, and male viewers report more positive evaluations when viewing heterosexual couples in sexually suggestive ads than female viewers do (Reichert et al., 2007). However, not all studies confirm strong gender differences. For instance, Wirtz et al. (2018) found no significant difference in overall audience attitudes toward ads containing sexual content, even when the products being advertised were unrelated to sexuality, such as computers and laundry detergent.

Beyond consumer psychology, gender-balanced portrayals also connect to broader debates on diversity and inclusivity in marketing practice, where equitable representation has been linked to stronger brand trust and stakeholder engagement (Ferraro et al., 2023; Morgan et al., 2025; Trkulja et al., 2024).

1.4. The Present Study

Prior research has shown that nudity in advertising can elicit both positive and negative reactions, with responses often shaped by factors such as model gender, clothing, and audience expectations (Reichert et al., 2007; LaTour, 1990; Halldórsdóttir et al., 2020). Furthermore, studies have documented gender-based differences in audiences’ emotional and cognitive reactions (Ford & LaTour, 1993; LaTour, 1990; LaTour et al., 1990; Pradhan, 2016). However, many of these studies suffer from methodological inconsistencies, such as varying models or products across conditions, which compromise internal validity.

The present experiment addresses these gaps by examining the effects of model gender (male, female, both), model clothing (swimwear vs. outdoor wear), and participant gender (men vs. women) on ad evaluations. Importantly, we introduce a novel experimental condition in which both a male and a female model appear together. This condition allows us to test not only whether nudity level influences consumer responses (H1), but also whether gender-balanced portrayals yield more favorable brand attitudes and purchase intentions than single-gender portrayals (H2). Finally, we examine whether this potential advantage of balanced representation generalizes across male and female participants, thereby reducing gender-based asymmetries in responses (H3). By holding constant the brand, product, and professional models across conditions, our design achieves high internal validity while maintaining external realism. This enables a more rigorous test of whether nudity and model composition affect attitudes toward the advertisement, brand evaluations, and purchase intentions.

Based on the Elaboration Likelihood Model (Petty & Cacioppo, 1986) and prior findings on gender representation, we advance the following hypotheses:

• H1: Advertisements featuring models in swimwear will not yield more favorable attitudes toward the advertisement, brand, or purchase intentions than ads featuring models in outdoor clothing.
• H2: Advertisements featuring both a male and a female model will generate more favorable brand attitudes and purchase intentions than single-gender advertisements.
• H3: The advantage of the both-genders condition will be observed for both male and female participants, thereby reducing gender-based asymmetries in responses.

2. Materials and Methods

2.1. Participants and Procedure

A power analysis was conducted to determine the required sample size. Based on prior research (Halldórsdóttir et al., 2020), the expected effect size for the relationship between nudity in advertising and attitude toward advertising was estimated to be in the medium to medium-large range. To reliably detect an effect of this magnitude (Cohen’s ƒ = 0.25 or 0.35) with adequate statistical power (1 − β = 0.80), a minimum of 98 to 188 participants would be required for a two-sided test (Faul et al., 2007, 2009).

A total of 195 individuals participated in the experiment: 76 men, 112 women, 6 non-binary participants, and one participant who did not disclose their gender. Only those who self-identified as male or female (n = 188) were included in the ANOVA analyses, which treated participant gender as a binary variable (male vs. female). The six nonbinary participants and one undisclosed participant were respectfully excluded from analyses involving participant gender to maintain statistical clarity and interpretability. The average age of the final sample was 29.9 years (SD = 11.8).

Participants were recruited via an initial email invitation sent through a university-wide undergraduate email list at the University of Iceland to all enrolled undergraduates, followed by one reminder to the same list. Participation was voluntary, and informed consent was obtained online prior to viewing the stimulus. The invitation explained that the study examined attitudes toward advertising and that participants would view an advertisement and answer questions about it. The study used a nonprobability volunteer sample from a single institution. Because messages were distributed via a university-wide email list without tracking email opens, click-throughs, or undeliverable messages, a conventional response rate could not be computed.

2.2. Stimuli

The experimental design followed a 3 (model gender: male, female, both) × 2 (model clothing) structure. Model gender had three levels: male model, female model, and both male and female models. Model clothing had two levels: swimwear and outdoor wear.

Participants were randomly assigned to one of six groups, with each group viewing a real advertisement from the perfume brand Dolce & Gabbana. A real brand was used because research suggests that external validity increases when an actual brand is employed rather than an imaginary one (Choi et al., 2013). Using the same model, brand, and product across conditions was essential to ensure internal validity.

Six advertisements for Dolce & Gabbana’s Light Blue fragrance was used as experimental stimuli, comprising two seasonal themes, winter and summer, and varying by model gender composition (male, female, or both).

In the winter-themed set, one advertisement featured a male model in a snowy alpine environment. He was dressed in a partially unbuttoned thermal top with a sweater tied around his waist and held ski poles, emphasizing outdoor activity and cold-weather styling. A second advertisement showed a female model in a similar alpine setting, wearing black ski pants, a cropped black top, and a partially open sweater, with ski goggles positioned on her head. A third winter-themed advertisement included both the male and female models standing together, each dressed as in the individual versions, and both holding ski poles. The mountainous background and coordinated styling across all three ads reinforced the seasonal branding of the fragrance.

In the summer-themed set, one advertisement depicted a male model seated on a white boat near a rocky coastal formation, wearing white swim briefs. A second featured a female model reclining in the same boat, wearing a white bikini, with a similar Mediterranean coastal backdrop. The third summer-themed advertisement included both models seated closely together in the boat, dressed as in their respective solo ads. The bright lighting, setting, and wardrobe across these advertisements conveyed a sunlit, summer atmosphere aligned with the brand’s warm-weather fragrance campaign.

Following prior work on sexual appeals that manipulate body coverage via wardrobe (e.g., Dudley, 1999; LaTour, 1990; Matthes & Prieler, 2020; Reichert & Carpenter, 2004), we operationalized “nudity level” as the degree of body coverage conveyed by the model’s clothing: swimwear = greater exposure vs. outdoor wear = greater coverage. Throughout, “nudity level” is used as shorthand for this difference in body coverage rather than literal nudity. All other elements (brand, product line, professional models, seasonal theme within set) were held constant across conditions.

2.3. Measures

Participants answered five questions regarding their attitude toward the advertisement (Muehling & McCann, 1993). The questions were rated on a seven-point scale, where 1 indicated a very negative evaluation and 7 a very positive evaluation. The items used to measure ad attitudes were as follows: How bad or good did you find the advertisement?, How much did you dislike or like the advertisement? How negative or positive are you toward the advertisement? How uninteresting or interesting did you find the advertisement? And how unattractive or attractive did you find the advertisement? These five items were averaged to create a single index of attitude toward the advertisement. The internal consistency of the scale was very high (Cronbach’s α = 0.93), consistent with Halldórsdóttir et al. (2020), who reported a nearly identical reliability (α = 0.926) using the same scale in Iceland.

Participants were then asked about their attitude toward the brand using the question: How much do you like or dislike the brand? This was also rated on a seven-point scale, with 1 indicating dislike very much and 7 like very much.

Finally, participants were asked about their purchase intention with the question: How likely or unlikely are you to purchase the fragrance? Again, responses were given on a seven-point scale, where 1 meant very unlikely and 7 meant very likely.

2.4. Statistical Analysis

We first conducted three three-way ANOVAs to examine the effects of model gender (male, female, both), model clothing (swimwear vs. outdoor wear), and participant gender (male vs. female) on participants’ attitudes toward the advertisement, brand attitude, and purchase intentions. Inspection of scatterplots of standardized residuals against the predicted values for each dependent variable indicated no major violations of regression assumptions (e.g., normality). Residuals were approximately symmetrically distributed around zero, with no evidence of heteroscedasticity or systematic patterns. In addition, Levene’s tests of equality of error variances were nonsignificant for all analyses (all ps > 0.05), indicating that the assumption of homogeneity of variances was satisfied.

To probe underlying psychological mechanisms, we used Hayes’s (2022) PROCESS macro. A serial mediation model (Model 6) tested whether the effect of model composition (multicategorical independent variable with three levels) on purchase intention operated indirectly through attitudes toward the advertisement (M1) and brand attitude (M2), consistent with advertising process models. Model composition was dummy-coded (female-only as the reference), with 5000 bias-corrected bootstrap samples and HC3 robust standard errors. Nudity level and participant gender were included as covariates. To assess robustness, we also estimated two simple mediation models (Model 4), treating either attitudes toward the advertisement or brand attitude as the sole mediator. We report total, direct, and indirect effects with 95% bootstrapped confidence intervals (CIs), along with pairwise contrasts of indirect effects between model-composition conditions.

3. Results

Table 1. Descriptive statistics and correlations among study variables.

					Correlation Coefficients
	Variables	M	Sf	N	2	3
1	Attitude toward the advertisement	3.17	1.42	195
2	Brand attitude	4.04	1.33	192	0.52 ***
3	Purchase intentions	2.15	1.55	194	0.63 ***	0.60 ***

*** p < 0.001.

presents the means, standard deviations, and bivariate correlations for all three dependent variables. As shown, attitudes toward the advertisement, brand attitude, and purchase intention were positively correlated, indicating that more favorable ad evaluations were associated with more favorable brand evaluations and stronger purchase intentions.

We first examined the effects of model gender, nudity level, and participant gender on participants’ attitudes toward the advertisement, brand attitude, and purchase intentions using three three-way ANOVAs. We then conducted mediation analyses with Hayes’s PROCESS macro to test whether attitudes toward the advertisement and brand attitude explained the effects of model composition on purchase intentions.

3.1. Attitude Toward the Advertisement

A three-way ANOVA was conducted to examine the effects of model gender (male, female, both), nudity level (swimwear vs. outdoor wear), and participant gender (male vs. female) on participants’ attitudes toward the advertisement. There were no statistically significant main effects of model gender, F(2, 176) = 0.36, p = 0.701, η² = 0.004; nudity level, F(1, 176) = 0.01, p = 0.932, η² < 0.001; or participant gender, F(1, 176) = 1.44, p = 0.232, η² = 0.008. Additionally, none of the two-way interactions, model gender × nudity level, F(2, 176) = 0.91, p = 0.406; model gender × participant gender, F(2, 176) = 0.05, p = 0.951; or nudity level × participant gender, F(1, 176) = 0.22, p = 0.643 were significant. The three-way interaction was also non-significant, F(2, 176) = 0.05, p = 0.948, η² = 0.001. These results suggest that none of the experimental manipulations or participant gender influenced ad attitudes (Figure 1).

3.2. Attitude Toward the Brand

A similar three-way ANOVA was conducted with brand attitude as the dependent variable. There was a significant main effect of model gender, F(2, 174) = 3.66, p = 0.028, η² = 0.040, and a significant interaction between nudity level and participant gender, F(1, 174) = 4.11, p = 0.044, η² = 0.023. No other main effects or interactions were statistically significant: nudity level, F(1, 174) = 0.02, p = 0.894; participant gender, F(1, 174) = 0.43, p = 0.513; model gender × nudity level, F(2, 174) = 0.01, p = 0.993; model gender × participant gender, F(2, 174) = 1.20, p = 0.305; and the three-way interaction, F(2, 174) = 1.48, p = 0.230 (Figure 2).

Post hoc comparisons (Tukey HSD) revealed that participants rated the brand more positively when both a male and female model were shown (M = 4.34, SD = 1.41) compared to when a female model alone was shown (M = 3.80, SD = 1.33), although this difference did not reach statistical significance (p = 0.058).

3.3. Purchase Intention

The third ANOVA tested effects on participants’ purchase intention. There was a statistically significant main effect of model gender, F(2, 176) = 3.25, p = 0.041, η² = 0.036, and a marginally significant interaction between model gender and participant gender, F(2, 176) = 2.61, p = 0.076, η² = 0.029. No other effects were significant: nudity level, F(1, 176) = 0.25, p = 0.619; participant gender, F(1, 176) = 0.18, p = 0.673; model gender × nudity level, F(2, 176) = 0.11, p = 0.896; nudity level × participant gender, F(1, 176) = 1.96, p = 0.164; and the three-way interaction, F(2, 176) = 0.92, p = 0.402 (Figure 3).

Tukey post hoc comparisons showed that participants reported higher purchase intention when both a male and a female model were featured (M = 2.60, SD = 1.65) compared to when only a female model was shown (M = 2.00, SD = 1.47), although this difference did not reach statistical significance (p = 0.075).

To evaluate whether the non-significant effects reflected a true absence of impact rather than insufficient power, we conducted equivalence tests using η² = 0.04 as the smallest effect size of interest (Lakens, 2017). For ad attitudes, all non-significant effects (η² range = 0.001–0.008) were statistically equivalent to zero, suggesting that model gender, nudity level, participant gender, and their interactions did not produce even small-to-medium effects. Similarly, for brand attitudes, the non-significant effects of clothing, participant gender, and their higher-order interactions (η² range = 0.000–0.023) were also statistically equivalent to zero. For purchase intentions, most non-significant effects (η² range = 0.001–0.011) were likewise statistically equivalent to zero. Only the marginal interaction between model gender and participant gender (η² = 0.029, p = 0.076) did not meet the equivalence criteria, indicating that this effect may warrant further investigation.

3.4. Mediation Analysis

To examine whether the effect of model composition on purchase intention operated through attitudes toward the ad and brand attitude, we estimated a serial mediation model (PROCESS Model 6). Model composition was dummy coded with female-only ads as the reference condition, attitudes toward the ad as M1, brand attitude as M2, and purchase intention as the dependent variable. Nudity level and participant gender were included as covariates.

The overall effect of model composition on purchase intention was marginally significant, F(2, 181) = 2.91, p = 0.057 (

Table 2. Indirect and direct effects of model composition on purchase intention through attitudes toward the advertisement and brand attitude.

Pathway	Effect (b)	BootSE	95% CI	Sig.
Both vs. Female → Attitudes toward the ad → Purchase intention	0.06	0.13	[−0.19, 0.34]	ns
Both vs. Female → Brand attitude → Purchase intention	0.05	0.10	[−0.14, 0.26]	ns
Both vs. Female → Attitudes toward the ad → Brand attitude → Purchase intention	0.03	0.06	[−0.10, 0.15]	ns
Male vs. Female → Attitudes toward the ad → Purchase intention	−0.03	0.11	[−0.27, 0.17]	ns
Male vs. Female → Brand attitude → Purchase intention	−0.15	0.08	[−0.33, 0.01]	ns
Male vs. Female → Attitudes toward the ad → Brand attitude → Purchase intention	−0.02	0.05	[−0.13, 0.09]	ns

ns = not significant.

). A significant contrast indicated that ads featuring both male and female models increased purchase intention relative to female-only ads, b = 0.63, p = 0.033.

Mediation tests revealed that none of the indirect pathways were significant. For the both-genders condition, the indirect effect via attitudes toward the ad was b = 0.06, 95% CI [−0.19, 0.34], via brand attitude b = 0.05, 95% CI [−0.14, 0.26], and through the sequential pathway attitudes toward the ad → brand attitude → purchase intention b = 0.03, 95% CI [−0.10, 0.15]. Comparable indirect effects for the male-only condition relative to the female-only condition were also nonsignificant.

When controlling for both mediators, the direct effect of the both-genders condition on purchase intention remained significant, b = 0.49, p = 0.021, suggesting that the advantage of featuring both male and female models is not explained by attitudes toward the ad or brand attitude.

To assess robustness, we estimated two simple mediation models (PROCESS Model 4), treating attitudes toward the ad and brand attitude as sole mediators. As shown in

Table 3. Simple mediation analyses of attitudes toward the advertisement and brand attitude as sole mediators of the effect of model composition on purchase intention.

Comparison	Mediator	Direct Effect (b)	Direct p	Indirect Effect (b)	95% CI	Sig.
Male vs. Female	Attitudes toward the ad	0.10	ns	−0.05	[−0.38, 0.25]	ns
Both vs. Female	Attitudes toward the ad	0.55	0.016 *	0.08	[−0.30, 0.46]	ns
Male vs. Female	Brand attitude	0.32	ns	−0.27	[−0.59, 0.05]	ns
Both vs. Female	Brand attitude	0.51	0.033 *	0.13	[−0.23, 0.50]	ns

* p < 0.05, ns = not significant.

, neither variable significantly mediated the effect of model composition on purchase intention. For attitudes toward the ad, indirect effects were nonsignificant (both vs. female: b = 0.08, 95% CI [−0.30, 0.46]; male vs. female: b = −0.05, 95% CI [−0.38, 0.25]). Similarly, for brand attitude, indirect effects were nonsignificant (both vs. female: b = 0.13, 95% CI [−0.23, 0.50]; male vs. female: b = −0.27, 95% CI [−0.59, 0.05]). Across both models, the direct effect of the both-genders condition on purchase intention remained significant (p < 0.05), indicating that neither attitudes toward the ad nor brand attitude fully explains the advantage of balanced gender portrayals.

4. Discussion

The aim of this study was to examine whether sexual appeals in advertising, specifically nudity, affect consumer attitudes toward the advertisement, the brand, and purchase intentions. To isolate the effects of nudity and model gender, the study employed a carefully controlled design in which all advertisements featured the same brand, the same product line, and the same professional models. This consistency across conditions enhances internal validity and distinguishes the study from previous research that relied on varied stimuli, which can confound interpretation. Building on this design, we tested three a priori hypotheses (H1–H3) concerning (i) nudity level (swimwear vs. outdoor wear), (ii) model composition (male, female, both), and (iii) participant gender as a potential moderator.

Consistent with H1, varying nudity level (body coverage; swimwear vs. outdoor wear) did not improve attitudes toward the ad and showed no main effects on brand evaluations or purchase intentions. By contrast, in line with H2, model composition yielded small but reliable omnibus differences for brand attitude and purchase intention, whereas attitudes toward the ad were unaffected by any manipulation. Interpreted through the Elaboration Likelihood Model (Petty & Cacioppo, 1986), when argument strength is minimal, as in fragrance advertising, peripheral cues tied to who is depicted (e.g., featuring both a male and a female model) may shape downstream evaluations more than the degree of body coverage. This pattern aligns with evidence that advertising inputs often produce small average effects (Eisend & Tarrahi, 2016) and challenges the assumption that more nudity is inherently persuasive (Reichert et al., 2007; Wirtz et al., 2018; Halldórsdóttir et al., 2020).

The results showed no significant differences in attitude toward the ad as a function of model clothing, suggesting that nudity level by itself may not drive evaluations. At the same time, advertisements that presented a male and a female model together yielded more favorable brand attitudes and purchase intentions than single-model ads. This gender-balanced configuration is rarely tested with real advertisements and therefore adds a small but novel contribution to the literature. Our pattern differs from Halldórsdóttir et al. (2020), who reported negative effects of nudity on ad attitudes, and is closer to studies showing that nudity alone does not consistently improve effectiveness (Wirtz et al., 2018; Reichert et al., 2007). Descriptively, ads with only a female model were evaluated least favorably, while the both-genders condition performed best; however, pairwise contrasts did not reach significance and should be interpreted with caution.

Although our PROCESS models did not support mediation via attitude toward the ad or brand attitude, the pattern is consistent with peripheral-cue processes under ELM: featuring both genders may broaden self-relevance and reduce perceived stereotyping or objectification, thereby lowering offense or reactance and supporting more favorable judgments (Petty & Cacioppo, 1986; Huhmann & Limbu, 2016; Reichert et al., 2007). Under low-elaboration conditions typical of fragrance advertising, such cues can tip evaluations even when argument strength is minimal. Consistent with classic persuasion models (MacKenzie & Lutz, 1989; Mitchell & Olson, 1981; Spears & Singh, 2004), we therefore tested attitude toward the ad and brand attitude as mediators; however, the indirect paths were not significant (Table 2 and Table 3), and the both-genders contrast on purchase intention remained direct when controlling for both mediators. These mechanisms are speculative in the present data. Future studies should measure arousal/offense and objectification directly and consider inclusivity signals to stakeholders (Trkulja et al., 2024).

We advise against treating greater nudity as a default lever; in low-argument contexts, model composition is the more reliable cue. Featuring both a male and a female model may lower perceived offensiveness/stereotyping and broaden self-relevance (Huhmann & Limbu, 2016; Terlutter et al., 2021). Effects are category- and brand-dependent. They are more plausible in hedonic categories and when the creative fits the brand, and a misfit can backfire, especially on social platforms (Choi et al., 2022; Wirtz et al., 2018). On social media in particular, maintaining alignment with brand identity and stakeholder expectations supports engagement and reputation (Fayvishenko et al., 2023; Herman et al., 2021; Trkulja et al., 2024). These findings also contribute to ongoing debates in marketing ethics and diversity by showing that gender-balanced portrayals, beyond their persuasive impact, may align advertising practices with broader calls for equitable and inclusive representation (Huhmann & Limbu, 2016; Trkulja et al., 2024).

The significant omnibus effects for model composition were small. This pattern is consistent with evidence that advertising inputs often yield modest average effects (Eisend & Tarrahi, 2016). Practically, any advantage of featuring both a male and a female model should be seen as incremental, potentially meaningful when aligned with brand identity and audience expectations (Choi et al., 2022) and when delivered with sufficient reach and frequency, but unlikely to overcome category or platform mismatches (Wirtz et al., 2018).

The implications of our findings are likely category dependent. Sexual content varies strongly by product category (Reichert et al., 2012), and prior work shows that it does not consistently help in more utilitarian categories such as computers or laundry detergent (Wirtz et al., 2018). Our pattern, an advantage for both-gender compositions under low-argument conditions, therefore seems most plausible for hedonic/experiential categories like fragrance and fashion, especially when aligned with brand identity. Misfit can weaken consumer connection, particularly on social platforms (Choi et al., 2022). In practice, teams should expect small, incremental gains in congruent categories and little benefit where sexual appeals clash with the product’s functional positioning.

Our data come from Icelandic university students, so cultural generalization is limited. Prior work shows that responses to sexual content and to perceived offensiveness vary across cultures and platforms (Dianoux & Linhart, 2010; Terlutter et al., 2021; Choi et al., 2022; Khalid et al., 2023), and that the prevalence and portrayal of nudity differ across countries (Matthes & Prieler, 2020). Thus, the small advantage we observe for both-gender compositions may be stronger, weaker, or even reversed in markets with different modesty norms or identity expectations. For practice, we recommend local pretesting and creative localization (e.g., calibrating body coverage and model composition to brand fit and local norms) when adapting campaigns beyond Western European contexts.

This study has several limitations. Although brand, product line, and professional models were held constant, other executional features necessarily varied across ads (e.g., seasonal setting, lighting, posture/gaze) and may act as peripheral cues under low elaboration (Petty & Cacioppo, 1986). We did not measure prior familiarity or baseline attitude toward Dolce & Gabbana. Such baselines could shift overall levels despite random assignment. Outcomes reflect responses to a single, immediate exposure when advertising effects can depend on repetition and frequency (Singh & Cole, 1993; Vakratsas & Ambler, 1999). Finally, all focal variables were self-reported, and therefore, adding implicit or behavioral measures would strengthen inference.

Because all stimuli came from a single brand (Dolce & Gabbana), the study should be viewed as a tightly controlled, single-brand case designed to isolate the effects of nudity and model composition. This choice safeguards internal validity but constrains external validity. The results likely generalize best to fragrance advertising and brands with a sexualized image. Future research should replicate the design with additional brands (including less sexualized ones), other product categories, different media formats (e.g., video, social), and across cultural contexts to evaluate the scope of generalizability.

Although the use of a carefully controlled design (featuring the same brand, product line, and models) is one of the strengths of this study, this approach does limit generalizability. To address this issue, we suggest that the study should be replicated using different brands to examine whether brand type plays a role.

A further limitation is that we did not include a manipulation check to confirm participants’ perceptions of the stimuli. Although the use of professionally produced advertisements with clear visual differences helps reduce this concern, future research should incorporate manipulation checks or pretests to ensure that participants interpret the stimuli as intended.

In addition, the study did not measure emotional valence or arousal, which are central mechanisms in how sexual appeals affect consumer responses. Future research should incorporate such measures to more precisely capture the affective processes underlying ad and brand evaluations.

Although our sample size exceeded the a priori requirements for detecting medium effects (Halldórsdóttir et al., 2020), it may not have been large enough to reliably detect very small effects (η² < 0.01). The unequal group sizes between male and female participants further limit the power for interaction effects, which should be interpreted with caution. Another potential limitation is that some constructs were measured using single-item scales.

We used static print executions. Effects may differ in video or interactive formats. Given current practice, we view interactive digital/social replications as most urgent, followed by studies that measure brand familiarity, include repeated exposures, and add implicit/behavioral (and optional biometric) indices to unpack mechanisms.

Building on these limitations, a priority for future research is to move beyond static print and test these effects in interactive digital and social formats (short video, Stories/reels, influencer placements), where platform norms and engagement dynamics may alter outcomes (Choi et al., 2022; Nadlifatin et al., 2022; Wirtz et al., 2018). Randomized A/B experiments can orthogonally manipulate model composition and body coverage, include manipulation checks, and track platform metrics (e.g., click-through, dwell time, shares) alongside attitudes and intentions. A second priority is systematic cross-cultural work using the same creative localized across countries to assess transportability and cultural moderation, given known differences in responses and portrayals of nudity (Dianoux & Linhart, 2010; Terlutter et al., 2021; Matthes & Prieler, 2020; Choi et al., 2022; Khalid et al., 2023).

To identify mechanisms, future studies should combine self-report with process measures such as eye-tracking and biometrics (e.g., facial expression, skin conductance, heart rate, EEG) to test whether gender-balanced creative reduces offense/objectification and broadens self-relevance under low-elaboration conditions (Petty & Cacioppo, 1986; Reichert et al., 2007). Moderators worth mapping include cultural values (Khalid et al., 2023), sexual self-schema (Reichert et al., 2007), gender identity and sexual orientation, as well as brand familiarity and category type (hedonic vs. utilitarian; Reichert et al., 2012; Wirtz et al., 2018). Designs should also measure brand familiarity, and cleanly separate body coverage from other executional features (e.g., setting/season).

Finally, long-term effects merit attention. Longitudinal or field experiments can examine frequency and repetition (Singh & Cole, 1993) to test for wear-in, desensitization, or wear-out (Vakratsas & Ambler, 1999), linking exposure histories to lagged brand outcomes (attitudes, choice, willingness-to-pay). This agenda would clarify when and why sexual appeals work, for whom, and on which platforms, and whether the small advantages of both-gender compositions cumulate to commercially meaningful effects.

5. Conclusions

This study examined sexual appeals in advertising with tightly controlled, real-brand stimuli. Nudity level did not improve attitudes toward the ad and showed limited impact on brand evaluations or purchase intentions. By contrast, model composition, presenting both a male and a female model, was associated with small omnibus improvements in brand attitude and purchase intention. In practice, these effects should be viewed as incremental, contingent on brand fit and category context. Any benefits must be weighed against creative and production costs, and verified through pretesting in the intended media environment. While gender-balanced portrayals may align with contemporary expectations around equitable and inclusive representation, our data do not permit causal claims about broader outcomes such as brand trust. Overall, the results challenge the assumption that “more nudity” is inherently effective and point instead to model composition as a modest, context-dependent cue that can be leveraged, especially in low-argument settings, subject to rigorous testing and local adaptation.