“One thing I always really struggled with growing up is I never saw or heard anybody like me on TV or radio”—Dylan Alcott, former wheelchair tennis champion (Topsfield, 2023)

Introduction

It is clear that diverse representation in advertising is currently a prominent topic. We define representation in advertising as the fair and proportional portrayal of diverse groups in a particular promotional setting or medium (c.f., Eisend, 2022).Footnote 1 For simplicity, we also use the shorter term “diversity” to refer to this term. Research shows that many consumers want to see more diversity in marketing (Brodzik et al., 2021). Following the protests during COVID-19, many brands and marketers made bold commitments to diversity (Notte, 2023). However, many have struggled to keep them (Klara, 2023) and others are rethinking efforts amidst cutbacks (Alfonseca & Zahn, 2023) and potential legal challenges (The Economist, 2023). While it is undeniable that diverse representation in advertising has dramatically increased in recent decades, some reports are now indicating that ads have recently become less representative (e.g., Council, 2023).

Table 1 Table of diversity definitions
Full size table

Many brands are also being reminded that not all consumers value diverse representation. In the past year Disney, Anheuser-Busch, and the National Hockey League all dealt with fallout from their diversity efforts. Anheuser-Busch reportedly lost $6 billion in market capitalization in the week after partnering with a transgender influencer to promote its Bud Light brand (Hammer, 2023). Additionally, diversity efforts may be criticized by consumers who are supportive of the cause, but question the brand’s authenticity (e.g., “woke washing”, Vredenburg et al., 2020; or “diversity washing”, Baker et al., 2023). Such backlashes are said to be driving a decrease in mentions of diversity-related efforts by brands (Maurer, 2023). This reversal reflects a larger ongoing debate among both practitioners and academics about whether or not brands should aim to shape societal views on diversity (Eisend, 2010; Grau & Zotos, 2016; Holbrook, 1987; Lantos, 1987; Zayer & Coleman, 2015).

Complicating matters further, changes within the ad industry are occurring that have the potential to reshape how advertising is conducted. Rising reliance on data-driven decision-making can inadvertently intensify existing biases, like racism (Akter et al., 2022; Zou & Schiebinger, 2018). With the advent of technologies like deepfakes and generative AI, advertisers can now modify features such as a model’s age, gender, and race with ease (Campbell et al., 2022a). While these advancements allow for precision in audience targeting, presenting tailored ads representative of specific diverse groups, they also risk fostering a “diversity echo chamber”. Such an echo chamber may inhibit diverse representation by presenting ads repeatedly to a specific group that are curated to match the reality that group wants to see, whatever that might be. To put it succinctly, emerging tools offer advertisers greater flexibility in determining ad diversity, but this also introduces intricate challenges and avenues for potential criticism.

In light of these changes, it’s becoming imperative for companies to grasp how consumers perceive and engage with diversity, whether it be in a positive or negative manner.Footnote 2 However, the existing literature on diversity presents two challenges. Firstly, it’s scattered and lacks cohesion. Secondly, as recognition of various forms of diversity expands, the subject matter becomes more intricate. While deepening our understanding is beneficial, this diversifying landscape of knowledge complicates our ability to comprehensively discern how consumers interpret and react to diversity in advertising. Regardless of a brand’s stance on diversity, we believe it’s nonetheless beneficial to have a clearer understanding of how consumers respond to diverse representation in advertising.

We acknowledge that there are many exemplary systematic literature reviews (Ginder & Byun, 2015), literature reviews (Eisend, 2022; Grau & Zotos, 2016; Greco, 1987; Joseph 1982; Kumari & Shivani, 2012; Prieler, 2020; Taylor et al., 2019; Zayer & Pounders, 2022), and meta-analyses (Eisend & Hermann, 2019; Groesz et al., 2002) that address specific types of diversity representation. Yet, only two studies (Eisend, 2022; Eisend & Hermann, 2019) develop conceptual frameworks. While valuable, they feature a narrower scope of a single aspect of diversity. To the best of our understanding, no study has holistically assessed and synthesized the full body of literature on diversity representation in advertising. This article hopes to change that.

To achieve its aim, this article proceeds through eight distinct steps. First, we pinpoint the major forms of diverse representation that have been explored within the advertising literature. Second, using census data and multiple industry and academic studies, we confirm that diverse representation is often lacking within advertising. Third, we embark on a comprehensive review of existing literature on diversity representation in advertising, using both systematic and manual search methods, identifying a total of 337 relevant articles. Fourth, we then present overarching insights into the current trajectories of research on diverse representation, showing a discernible uptick in such scholarship. Fifth, inspired by methodologies from Crossan and Apaydin (2010) and Peñaloza et al. (2023), we conduct a deep qualitative analysis of the 337 articles. This analysis, reminiscent of grounded theory (Glaser & Strauss, 1967), seeks not only to summarize but also to make inferences regarding the potential generalizability of specific effects. This leads to our sixth step: the construction of a conceptual framework that encapsulates the generalizable theories, moderators, mechanisms, and social and commercial outcomes found in the current diversity representation literature. For rigor and completeness, we present a comprehensive table linking each of the 337 articles to our framework’s elements. Seventh, we then discuss and characterize the current state of scholarship on diverse representation. Eighth, and finally, we develop a comprehensive research agenda, paving the way for future inquiries into diverse representation in advertising.

Our investigation offers two primary conceptual contributions to the field of diverse representation in advertising: summarizing and integrating (see MacInnis, 2011). Conceptual contributions and their associated insights are valuable as they can catalyze shifts in perspectives and alter thinking (Hulland, 2020; MacInnis, 2011). First, we summarize by examining the entirety of research around a domain, or “an area of study”, rather than a construct, a procedure, or a philosophy of science issue (Palmatier et al., 2018). Our review offers what we believe is the first distillation of insights from the broader literature into conclusions about what is known in seven key areas of diverse representation. By summarizing insights from hundreds of articles, we provide not only a streamlined resource for academics delving into diversity research, but also a succinct overview for a broader audience of the current landscape of diversity studies, its evolutions, and trends (Palmatier et al., 2018). Our analysis of past work uncovers a series of themes that we believe are important to scholarship on diversity, as well as a series of suggestions for managers, policy makers, and industry professionals. In our summarizing of the literature, we also compile important descriptive information, both with respect to published work, but also in comparing advertising in the marketplace to certain societal base rates.

Our second conceptual contribution is integration. We leverage the results of our literature review to develop something new: a comprehensive framework that enables otherwise siloed knowledge and theory to be applied and extended across different facets of diversity (Hulland, 2020; Hulland & Houston, 2020). The framework encourages a more holistic view of the different forms of diversity, encouraging multiple forms of diversity to be considered concurrently and identifying overarching ideas (c.f., Breslin and Gattrell 2023). Identifying such commonalities and extrapolations provides increased theoretical depth across all the different areas where diversity is studied in advertising. Our framework also simplifies (Hulland, 2020), distilling existing knowledge into an accessible set of constructs and relationships either common across the many aspects of diversity that were previously examined separately, or which could reasonably be expected to generalize among them. Our synthesis directly equips advertisers with insight into best practices and strategies for managing both diversity in their advertising, as well as the potential challenges it can bring. Finally, the framework—and the accompanying granular data we provide on which articles provide support for each of its facets—spark a range of specific questions for future research that we outline.

In the subsequent sections, we first detail the methodologies we utilize in our review and qualitative analysis of the literature, before turning to describe our findings. We present our findings in a conceptual framework, and also highlight and discuss prominent themes evident within the literature. Lastly, we introduce a comprehensive research agenda on diverse representation in advertising.

Our investigation and method

Our investigation begins by identifying which representations of diversity exist, and which have sufficient research available to review. Next, instead of relying on one search method, we combine the rigor of a systematic literature review (Palmatier et al., 2018; Tranfield et al., 2003) with the nuance of a manual search (e.g., Peñaloza et al., 2023). Merging the data from both of these search strategies, we then report bibliometric findings on the present state of research on diversity representation. In our analysis stage, we develop an overarching framework that summarizes and integrates (MacInnis, 2011) generalizable theories, moderators, mechanisms, and social and commercial outcomes present in the current literature on diversity representation. To achieve this, we use a qualitative approach as our goal at this stage is a conceptual, rather than an empirical, consolidation of the literature (c.f., Crossan & Apaydin, 2010; Peñaloza et al., 2023). We next describe each of these four stages in more detail.

Stage 1: Conceptualizing representation of diversity in advertising

We first sought to define the areas of diversity to be included in our review. This is an important step given an absence of an agreed list of areas that comprise diversity (Ahmed, 2020; Glasener et al., 2019). To develop our frame for investigating diversity representation in advertising, two authors conducted initial and independent reviews of the literature using Scopus, Web of Science, and Google Scholar to identify existing conversations related to advertising and diversity. To ensure that all forms of diversity were probed for, the authors also independently conducted searches outside of the literature to understand possible types of diversity. Following their investigations, the author team met to review and discuss their findings.

Our initial review revealed seven well-researched areas to form a corpus of articles for a literature review: age, beauty, body size, gender, LGBTQIA+ , physical and mental ability, and race and ethnicity. There are limitations to any classification. In our case, the areas are occasionally overlapping (e.g., the examination of gender and LGBTQIA+  in individual articles, or both body size and beauty) or result in aggregation which may not always be desirable (e.g., race and ethnicity, or by not examining different LGBTQIA + identities separately). While our list of diverse identities is not fully exhaustive nor perfect, we believe it represents substantive coverage of areas of diversity representation in advertising where there exists sufficient research to warrant summarizing and integration. We acknowledge that other forms of diversity exist beyond the areas identified, and indeed we identify and call for research on such areas in our future research section. At a broader level, our list can be considered in terms of differences in surface-level and deep-level traits (e.g., Khan and Kalra 2021). Surface-level characteristics are easily observable (e.g., age, ethnicity, gender) while deep-level characteristics are less so (e.g., religious beliefs, neurodiversity, health status, intersectionality). We return to this point later.

Any efforts to make strong claims about there being sufficient or insufficient diversity has obstacles. When commentators ask brands to “represent society”, critics will argue for more specificity. Which society should be represented? What if the vast majority of a product’s current users are highly homogeneous? What is “a society?” The world? A country? Should advertisers represent different societies (e.g., each city an ad runs in)? Ourselves and others (see Eisend, 2022 for a discussion of the challenges of benchmarking representation) acknowledge that there is no perfect comparison, and that choice of benchmark can significantly influence results. We nonetheless found it informative to examine descriptive findings about diversity relative to population base rates and accessible industry data.

There are many studies and industry reports (e.g., Cooper, 2022; Drayton, 2022; Extreme Reach, 2023; Geena Davis Institute, 2020a, 2020b; Meta, 2021; Nielsen, 2021) as well as peer reviewed content analyses (e.g., Parashar & Devanathan, 2006; Simcock & Lynn, 2006; Taylor et al., 2019) that show advertisers tend to overrepresent majority groups and underrepresent many minority groups. Even when minority groups are present, they are often represented in ways that are unequal to their majority counterparts (Eisend, 2010; Taylor et al., 2019). We aggregated this data and compared it to US Census Data, finding a general underrepresentation in the US across the diversity areas where statistics could be obtained. This comparison is presented in Fig. 1. However, there are certain minority groups that are better represented. For instance, Black or African American people comprise 12.9% of models in ads and 12.4% of the US population. This pattern is similar for Asian people who comprise 5.9% of models in ads and 6.0% of the US population. We next overview research justifying each area of diversity’s historical under- and/or misrepresentation in advertising. Definitions of each area are included in Table 1. For those interested, a richer discussion of each type of diversity representation is available in Supplemental Web Appendix A, Web Tables 6 to 12.

Fig. 1
figure 1

Diversity representation in advertising

Full size image

Diversity in age

Older consumers have historically been underrepresented in advertising (Eisend, 2022; Miller et al., 1999; Nelson, 2005; Rocks, 2020). While definitions differ on the starting age of older consumers, they are often segmented into "young-old" (over 60) and "old-old" (75+) (Nunan & Di Domenico, 2019). Figure 1 reveals that consumers over 60 represent 22.2% of the US population but only 4.2% of models in ads.

Diversity in beauty

Advertising is known to often emphasize traditional beauty standards, which highlight specific physical attributes deemed attractive in society (Swami et al., 2008). Such standards are also enabled by the rise of digital retouching in post-production (Schirmer et al., 2018; Semaan et al., 2018). Non-traditional beauty challenges these standards by promoting more diverse and realistic representations.

Diversity in body size

Despite 41.9% of the US population being plus size, Fig. 1 shows only 7.2% of models in ads are plus-sized. Advertising historically has favored thin models since the 1950s, reinforcing a thin beauty ideal (Fay & Price, 1994; Häfner & Trampe, 2009). Such portrayals have led to negative self-perceptions among young consumers (Carels & Musher-Eizenman, 2010; Dittmar & Howard, 2004) and are driving plus-size individuals to challenge the advertising industry’s damaging norms (Scaraboto & Fischer, 2013).

Diversity in gender

Advertising has historically portrayed women in limited and lower-status roles (Wolin, 2003), even though as shown in Fig. 1 they constitute 50.4% of the US population but represent only 38.8% of ad models. While global views on gender are evolving beyond binary classifications (Eisend & Rößner, 2022), advertising has been slow to change (Shinoda et al., 2021; Taylor et al., 2019). Although some modern campaigns are beginning to challenge these entrenched norms (Chu et al., 2016).

Diversity in LGBTQIA+

As shown in Fig. 1, the LGBTQIA+ community makes up 5.6% of the population but only 1.7% of models in ads, has often faced underrepresentation or inauthentic portrayals in advertising (Fig. 1). This community has sometimes been depicted negatively, as predators, promiscuous, or figures of fun (Mogaji, 2015). LGBTQIA+ representations can elicit varied responses from both community members and the broader public (Cowart & Wagner, 2021; Ginder & Byun, 2015). We acknowledge that LGBTQIA+ includes aspects of gender, however for the purpose of our review we include all gender-related effects within the “gender” topic area.

Diversity in physical and mental ability

Despite the US Census indicating 26% of the population has a disability, merely 1.6% of models in ads are differently abled. While their representation in advertising has seen some growth (Haller & Ralph, 2001), individuals who are differently abled remain either underrepresented or are depicted in ways that emphasize dependency and perpetuate stereotypes (Bernardi & Alhamdan, 2022; Mirabito et al., 2016).

Diversity in race and ethnicity

Race is based on physical attributes, while ethnicity focuses on cultural and ancestral identity (Cornell & Hartmann, 2006). Despite increased racial representation in media (Bang & Reece, 2003; Bowen & Schmid, 1997; Bristor et al., 1995; Stevenson & Swayne, 2011), certain racial groups remain underrepresented or are portrayed stereotypically (Bowen & Schmid, 1997; Maher et al., 2008; Mitchell, 2020; Plous & Neptune, 1997). Even when well-represented, many groups are stereotyped, sexualized, or play minor background roles in advertising (Tartaglia & Rollero, 2015; Taylor et al., 2005).

Stage 2: Literature review process

Having identified the areas of diverse representation to examine, we then reviewed the literature in two ways: first employing a systematic literature review (Palmatier et al., 2018; Tranfield et al., 2003) and then conducting a manual search (e.g., Peñaloza et al., 2023) of all work up until August 2022. Figure 2 overviews the process we now describe in detail.

Fig. 2
figure 2

Overview of literature review process

Full size image

Systematic literature review

The first part of the systematic literature review involved data collection, where we sought to identify articles based on keyword searches in relevant databases. We conducted our search using both the Scopus and Web of Science databases, as they provide the best coverage for social science research (Norris & Oppenheim, 2007). Our keyword search comprised three components. First, we included generic advertising search terms to focus our review on advertising effects. Second, we included specific diversity area search terms. However, given the fact that researchers often used different terms to discuss diversity areas (e.g., gay, homosexual, LGBTQIA+ are all possible search terms in the context of LGBTQIA+ alone), we employed a wide initial pool of diversity terms for each diversity area in order to maximize the inclusion of relevant studies. Third, we limited our search to peer-reviewed articles in marketing and psychology journals that occur on the Chartered Association of Business Schools (ABS) journal listing that are rated as 2 or above. We did not limit the search to any particular year and used all years available in the databases. We provide detailed search syntax in the web appendix (Supplementary Web Appendix A, Web Tables 2 to 4).

This search process resulted in 1,956 (Scopus) and 2,660 (Web of Science) articles, which when de-duplicated comprise 2,534 articles. Each author systematically and separately reviewed each article based on title and abstract. We selected only articles focused on advertising-relevant effects and in which one of the seven diversity areas was a key component of the article (resulting in an inter-rater agreement of 93%). Articles with diversity as a central focus of the work, but where diversity representation was not involved were excluded. This included, as examples, articles where children were the respondents to general advertising, or articles examining general response differences along sex or ethnic lines to advertising, but where diverse representation in advertising was not manipulated or examined. This process resulted in the identification of 234 articles for inclusion from the systematic review.

Manual literature search

During the systematic literature review process, it became apparent that some key articles we were familiar with were not included, given they did not meet the journal quality criteria based on the ABS listing or failed to include words used in our code. Hence, we conducted a subsequent cross-check with Google Scholar to catch any topically relevant or highly cited articles that the systematic review process had excluded. We then removed duplicate articles to identify unique articles from both the systematic and manual searches. This manual search resulted in the identification of an additional 103 articles, resulting in a final data set of 337 unique articles when combined with the 234 articles from the systematic review.

Authors then reviewed each article to code it into the seven diversity areas. Several of the articles in the data set were coded as being relevant to multiple diversity areas (e.g., gender and body size), resulting in 42 articles being coded across more than one diversity area. This process resulted in the following profile of articles: age = 44 (13%), beauty = 63 (19%), body size = 47 (14%), gender = 97 (29%), LGBTQIA +  = 27 (8%), physical and mental ability = 22 (7%), and race and ethnicity = 79 (23%). A complete list of the articles comprising the data set is presented in Supplementary Web Appendix B.

Stage 3: Bibliometric findings

We next report bibliometric findings across all of the diversity areas reviewed in our dataset. Our database encompasses articles up to mid-2022 and published in marketing and psychology journals ranked as 2 or above on the ABS journal listing. Web Table 5 in Supplementary Web Appendix provides a summary of the research on diversity in advertising by journal and type of diversity. While articles appear in a broad range of leading journals, not surprisingly, the largest proportions of articles come from more specialized advertising journals (i.e., Journal of Advertising, n = 66, 17%; International Journal of Advertising, n = 56, 15%; and Journal of Advertising Research, n = 25, 7%), followed by Psychology & Marketing, n = 19, 5% (full details are provided in supplementary web appendix Table 5).

Figure 3 plots the total number of articles published between the period 1964 and 2023, showing a sharp increase in publications during this period. Figure 3 also provides a break-out of this data by type of diversity, showing that the research on diversity in advertising is (and has been) dominated by research on gender (30%), beauty (19%), and race/ ethnicity (24%). Research on diversity in terms of LGBTQIA+ (8%) and physical and mental ability (7%) represent the lowest propositions of diversity research.

Fig. 3
figure 3

Total diversity in advertising publications by year

Full size image

Stage 4: Qualitative analysis of literature on representation

In our final analysis stage, we qualitatively analyze the seven identified areas of diversity representation in advertising to summarize and integrate (MacInnis, 2011; Post et al. 2020) the current literature. The 337 articles identified in our literature review comprise our dataset. Our goal is development of a conceptual framework that synthesizes what is currently known about diversity in advertising, how it operates, and how consumers react to it. In doing such integration we hope to encourage researchers to think outside of their silo and approach diversity from a more holistic vantage point (c.f., Breslin & Gatrell, 2023).

Our methodological approach follows that of Crossan and Apaydin (2010), and more recently Peñaloza et al. (2023). Data analysis was also informed by the grounded theory principles outlined by Glaser and Strauss (1967) and further detailed by Strauss and Corbin (1998). We employed the constant comparative analysis approach recommended by Spiggle (1994) as well as the pattern-matching technique described by Yin (1994), first categorizing the information and then identifying overarching themes. Articles were first reviewed individually and then compared and analyzed collectively. This process proceeded in two stages, which we describe in more detail next and summarize in Fig. 4.

Fig. 4
figure 4

Overview of qualitative analysis process

Full size image

In phase one, the 337 articles in our data set were analyzed in terms of each diversity area. Following an inductive process, each author separately reviewed the set of articles within each diversity area to identify and code potential moderators, mediators, and outcomes. The authors then met to discuss and identify common moderators, mediators, and outcomes present to each diversity area. This led to the development of seven common coding schemes—one for each area of diversity examined. We then re-reviewed each set of articles against the refined list of moderators, mediators, and outcomes developed for each respective diversity area. Detailed findings are available in Supplementary Web Appendix A, Web Tables 6 to 12.

In phase two, we shifted to look across diversity areas to identify moderators, mediators, and outcomes that could generalize across all diversity forms. We met to discuss and review our coding, resulting in a revised set of generalizable moderators, mediators, and outcomes that formed an initial framework. We then re-reviewed all articles against this framework, resulting in a table quantitatively connecting each effect to relevant articles. This table, shown in Table 2, was then revised to only include effects that either had strong empirical support, or based on inductive reasoning could reasonably be expected to extrapolate widely (Hulland, 2020). The final list of generalizable moderators, mediators, and outcomes appear in the framework shown in Fig. 5, which we describe in more detail in the next section.

Table 2 Connecting mechanisms, moderators, and outcomes with the literature in our data set
Full size table
Fig. 5
figure 5

A framework synthesizing existing understanding of diversity in advertising

Full size image

A unified framework of consumer response to diversity in advertising

Our qualitative analysis of the 337 articles that comprise the existing literature on representation of diversity in advertising lead to multiple outputs. In Web Tables 6 to 12 (Supplementary Web Appendix A) we summarize the current state of literature in each of the seven diversity areas we examine. In Fig. 5 we present a conceptual framework that details generalizable mechanisms, moderators, and outcomes gleaned from our examination of the disparate diversity literatures. It is important to note that inclusion of a specific effect or mechanism in the framework does not imply that it has been studied across all of the different forms of diversity possible in advertising. Rather, inclusion in the framework means that it has been studied in at least one domain, but we believe can be reasonably expected to occur across other forms of diversity. Table 2 demonstrates rigor (Hulland, 2020; Palmatier et al., 2018) by providing granular detail linking each aspect of our conceptual framework to specific articles from the 337 that we examined. This is not only useful as a summary of research on specific effects or topics, but more importantly exposes opportunities for future research. This is because Table 2 makes it very clear which effects in our framework are relatively understudied both in general, and in specific areas.

Our conceptual framework (Fig. 5) operates at the consumer-level, although like many models, it includes individual effects that are part of social forces and group identification (e.g., Hydock et al., 2020). Our hope is that creating this framework can break down silos and help create a more unified and purposive community of research focused on advancing diversity in adversity. We encourage future research that challenges, adds to, and augments this framework. We highlight components of this framework in more detail next.

Mechanisms through which diversity representation operate

Our review leads us to the conclusion that representations of diversity can cause effects through several different mechanisms. Habituation refers to the effect of becoming accustomed or acclimatized to seeing diversity in advertising. Habituation is driven by how often a consumer views diversity in advertising. Increasing habituation generally has positive effects.

Social comparison refers to either an explicit or implicit comparison that can occur when a difference exists between a viewer and the consumer group(s) depicted in an ad. Both representation and lack of representation of diversity in advertising can trigger social comparison, which can be either positive or negative depending on whether a viewer feels superior or inferior to what is depicted (e.g., Bower, 2001; Richins, 1991). Identification is similar but refers to the effect of seeing oneself represented in advertising. Identification generally has positive outcomes (e.g., Bond & Farrell, 2020; Hudders & De Jans, 2022).

Reactance refers to reactions that occur in response to a perceived loss of freedom. How a particular consumer group is depicted (or not depicted) in advertising can trigger reactance among viewers and generally results in negative outcomes (e.g., Åkestam et al., 2017). Depictions of diversity can also elicit perceptions of authenticity, which refers to how genuine or real an ad is perceived to be. Increased authenticity typically results in more positive outcomes (Becker et al., 2019; Shoenberger et al., 2020a, 2020b). Likewise, diversity can affect the perceived trustworthiness of an ad or spokesperson (e.g., Su et al., 2021).

Moderators on the effects of diversity representation

Multiple moderators that affect response to representations of diversity are studied within the literature. These can be related to three different aspects: depicted diversity, the viewer, or what is advertised. Within the category of depicted diversity, accuracy of representation refers to whether a diverse group is depicted in a manner that is consistent with their lived experience. Increased accuracy of representation generally enhances positive effects of depictions of diversity. Locus of control refers to whether an individual is seen as being in control of a form of diversity. For example, body size is viewed by some as being related to a person’s individual choices, rather than genetic or societal factors. Increased locus of control of a form of diversity generally results in less positive reactions. Extremeness refers to the degree or strength of diversity shown. Overtness refers to the extent a form of diversity is shown or featured within an ad. While extremeness and overtness are conceptually different, they both tend to elicit heightened responses from viewers (e.g., Meyer et al., 2020; Smeesters & Mandel, 2006).

In terms of characteristics of the viewer, social desirability refers to the extent to which a viewer has a desire to present themselves favorably relative to societal norms and expectations. Higher levels of social desirability typically cause a viewer to mute some responses to diversity while strengthening others. Schema (in)congruity refers to the extent to which a depiction of diversity matches—or does not match—what a viewer expects. Depictions that are more consistent with a viewer’s existing schemas are generally responded to more positively (e.g., Baxter et al., 2016). Culture is an important determinant in shaping consumer perspectives of, and responses to, diversity (Green, 1993; Maynard & Taylor, 1999). Culture affects a viewer’s values, beliefs, and experiences which in turn can shape how they view diversity and inclusion (Hofstede, 2001). Time refers to the fact that consumer perceptions of, and reactions to, diversity evolve over time (Bond & Farrell, 2020). This can in part be explained by habituation, whereby greater exposure results in individuals being less reactive or attentive to those differences (Levy & Glimcher, 2012). Finally, political or social dominance orientation affects responses to diversity, as those who are more conservative (or higher in social dominance orientation) generally have worse reactions to policies that benefit minority groups and report higher levels of prejudice and sexism (Pratto et al., 1997, 2000).

Several ad-related factors can also influence response to representations of diversity. The product category being advertised, and product-model fit can both impact how viewers respond to diversity (e.g., Greco, 1988; Häfner & Trampe, 2009). For instance, research finds that an androgynous model leads to better ad outcomes when representing a luxury, as opposed to non-luxury, brand (Cowart & Wagner, 2021). Likewise, the context in which an advertisement appears can affect response (e.g., Bian & Foxall, 2013; Peck & Loken, 2004), with ads featuring larger models placed in traditional vs. new magazines shown to have differential effects on ad outcomes (Peck & Loken, 2004). Breadth of targeting refers to a viewer’s perception of how broad or narrow an ad’s targeting is and viewers, particularly those in a diverse group, can react differently based on who an ad is perceived to be targeting (e.g., Aaker et al., 2000; Johnson & Grier, 2011).

Social effects of diversity representation

Representations of diversity can have multiple effects on both marginalized and non-marginalized consumers. Representation can trigger a feeling of inclusion and self-esteem (e.g., Dimofte et al., 2015; Henderson et al., 2023; Prieler, 2020). Inaccurate or absent representation can create, reinforce, or challenge negative stereotypes and stigma (e.g., Davies et al., 2002; Dias de Faria & Moreira Casotti, 2019; Houston, 2022; Phillips, 2022). For marginalized consumers, poor representation can also trigger shame (e.g., Pounders, 2018; Tunaley et al., 1999) along with coping behaviors. For non-marginalized consumers, viewing diversity can evoke a range of emotions including fear and aversion, disgust and repulsion, and pity and empathy (e.g., Corrigan et al., 2003; Yeh et al., 2017).

Commercial effects of diversity representation

Featuring diverse consumers in advertising has multiple impacts on advertisers and brands. First, representations of diversity can change the attention paid to an ad (e.g., Baxter et al., 2016; Hardin, 2003). In some cases, this can lead to distraction if attention becomes focused on model(s) rather than the advertised product or brand. Both effects can influence recall. Representation of diversity can affect attitudes (e.g., toward the model, ad, brand, product) (e.g., Micu et al., 2009) and behavioral intentions. For instance, diverse representation can cause consumers to purchase or not purchase products as a show of solidarity or as a boycott (e.g., Bradley & Longino, 2001; Eisend, 2022). More broadly, diverse representations can also shape the brand and product associations that consumers have in their minds (e.g., Liljedal et al., 2020). These associations can be positive or negative, influenced by both the consumer’s membership in a diverse group and their pre-existing beliefs about that group (e.g., Shavitt, 2019). Diversity in advertising can affect perceptions of brand/ad morality and workplace attractiveness as well (e.g., Khan & Kalra, 2021).

Discussing themes from our analysis of diverse representation in advertising

Research has a predominantly Western perspective

The first theme identified from our analysis is that most of the research on diverse representation in advertising is Western-centric or focused on developed nations, such as North America and Europe. This pattern is apparent in the manual coding of the study composition for each of the articles in our data set, shown in Table 3. This coding identifies 79% of studies as conducted in Western and developed nations (compared to 16% conducted in non-Western and 14% in developing nations). While we should not abandon Western research, there is an argument that we need more research into the effects of marketplace diversity and inclusion (Arsel et al., 2022) worldwide.

Table 3 Characterizing the literature in our data set
Full size table

Such a dominant Western perspective raises the question as to whether existing knowledge is transportable to non-Western contexts. For instance, should companies operating in many middle eastern countries feel compelled to increase representation of individuals with diverse sexual orientations? Particularly given it is illegal to express such an identity in some of those countries. For instance, Eisend and Hermann (2019) note that around 70 countries criminalize homosexuality. However, such views are liberalizing. In 2003, only two countries had laws allowing for same-sex marriage (Belgium and The Netherlands); only 20 years later, that number is now 30 (Masci et al., 2023). Even several of the most traditional countries have liberalized recently with respect to DEI, such as Saudi Arabia recently allowing women to travel abroad alone, drive vehicles, and live independently without consent—all part of the Saudi royal family’s stated goal of modernizing and reducing extremism (The Week, 2021).

Additionally, Table 3 highlights skews in the unit of analysis. First, the bulk of work uses the general population (52%) rather than diverse groups (20%) as the unit of analysis. Also, student samples, or those with a high degree of education, represent a high proportion of respondents (31%). These sample characteristics potentially skew our understanding of response to diverse representation. As advertising globalizes and more research is done with non-WEIRD samples (Henrich et al., 2010), there will be a broader voice and inclusivity of diverse perspectives in research. Doing so should also be of value theoretically; for example, highlighting differences between groups in their reactions to various content. Finally, there is only a limited number of studies that make comparisons across different cross-cultural samples; even fewer draw comparisons between Western and non-Western, or developed and developing, nations (i.e., An and Kim 2007; Cheng, 1997).

Research is largely non-experimental and non-behavior based

A second theme identified is that research on diversity representation tends to be non-experimental and non-behavior based. While Table 3 shows that a wide variety of methodologies have been employed in the study of diversity in advertising (i.e., quantitative 61%, qualitative 14%, content analysis 19%, literature review 7%, or conceptual 2%), most studies classified as quantitative were non-experimental, and thus offered a limited ability to draw causal conclusions. Fewer tested the purported psychological processes directly. Content analyses, however, were very common in the data, a method that is relatively rare in marketing more broadly (and perhaps even in other advertising research more specifically).

Additionally, there is a systemic focus on attitudes and intentions as outcomes of interest following ad exposure, rather than real behavior. There were virtually no studies that used field experiments, secondary purchase data, or true longitudinal designs. Purchase intentions were used instead of incentive compatible choices or consequential purchase decisions. There is opportunity for advertising researchers to run experiments or utilize A/B testing (Campbell et al., 2022b), for instance on social media or online shopping platforms. Further, self-reported claims about attention are not as persuasive as actual eye-tracking (Muñoz-Leiva et al., 2019). Ultimately, we echo Morales et al. (2017) that drawing more meaningful conclusions about effectiveness requires a more complete set of dependent measures, consequential choices, and marketplace evidence.

Research on diversity topics is unevenly distributed

A third theme identified is that there is a relatively narrow focus on diversity representation areas within past research. As demonstrated in Fig. 3, researchers tend to have focused on topics such as gender, race/ethnicity, and beauty, while limited work explores physical/mental ability and sexual orientation (LGBTQIA+). Further, areas of focus have been somewhat limited in scope, for instance research on gender has typically considered gender as binary (Eisend & Rößner, 2022). In essence, diversity research has overemphasized surface-level relative to deep-level diversity characteristics. In other words, the more visually apparent aspects of diversity seem to be better-researched than those less visually apparent. Furthermore, there is also limited research on certain forms of diversity. As in other domains, such as workplace diversity (Corrington et al., 2020), diversity research on individuals who vary in terms of religious beliefs, political ideology, economic status, and even undocumented immigrants tends to be lacking. Further, despite the known importance of intersectionality (Crenshaw, 2017), there are few studies that investigate intersectionality in the context of advertising or contrast responses to different forms of diversity. In other words, our review reveals that most articles examine a single aspect of diversity, and shy away from making comparisons when contrasting effects may be possible.

Research generally (but not always) views diversity representation as positive

A fourth theme we identify is that the overriding tone in the literature is positive towards diversity, although negative effects are also shown. In other words, researchers find (and report) that diversity in advertising is generally viewed favorably by consumers and advertisers. A general conclusion is that diversity tends to be viewed as good for business, and that firms should strive for greater representation. However, it is possible that bias exists. For instance, certain (more liberal) perspectives may be favored in the literature, either because these are easier to publish or because researchers who personally favor increased diversity are more likely to engage in scholarship in this domain (Duarte et al., 2015; Konnikova, 2014). Additionally, lessons from medicine and public health show that messaging that targets one group can either be ineffectual or result in backfire effects for other segments (c.f., anti-smoking and anti-drug advertising research, Bolton et al., 2008; Evans-Polce et al., 2015; Fishbein et al., 2002; Pechmann & Wang, 2010), suggesting that there is likely more heterogeneity in response than is captured by the homogeneous samples typically utilized.

Importantly, this theme requires more nuance for several reasons. First, stereotypes are pervasive, powerful, and feature both conscious and unconscious elements (Fiske, 1998). That an advertisement is powerful enough to overcome such effects, and do so with consistency, seems like an unrealistically tall order, and may reflect the short duration between exposure and response common in the literature. Relatedly, there is limited work looking at the effect of or habitation on response, or longer-term effects, both of which would require longitudinal studies (save for content analyses) not typically present in our analysis. A second reason to be hesitant towards a staunch pro-diversity expectation is that the marketplace has shown pushback towards diversity campaigns. Campaigns from Disney, Anheuser-Busch, and the National Hockey League (mentioned in the introduction) are some such examples, but there are seemingly several examples from around the globe each year. If diversity is so positive in advertising, the question arises as to why negative effects can occur in the marketplace when brands engage in it. This is especially important given the well-known social desirability concerns (e.g., Middleton & Jones, 2000) that exist when asking directly about diversity, where a danger is a gap between stated preferences (which often support diversity efforts) and revealed preferences (which often don’t).

While diversity research highlights its benefits, there are a range of negative effects that are identified. As examples, Lin and McFerran (2016) show that messaging around presenting overweight bodies as normal can reduce consumers’ desires to engage in healthier behaviors. Campbell et al. (2016) show a similar finding with cartoon characters and children as viewers. Baek et al. (2022) show that East Asian consumers (in an ethnically homogeneous location) can prefer non-inclusive (versus inclusive) messaging. El Hazzouri and Hamilton (2019) find that ethnic minority consumers respond more favorably to public health messaging from a Caucasian than a member of their own race, as did obese consumers to messaging featuring a normal weight (versus obese) model. Outside of the domains of beauty and body size, there is a scarcity of studies exploring when and why consumers sometimes prefer traditional (or aspirational) models vs. more realistic ones (or ones more like themselves).

Even if responses to diversity are not always positive, and indeed can sometimes be negative, this is more reason for work to examine strategies and tactics to better understand negative response, or consumer hesitancy to diversity representation. Such articles are a rarity and there is opportunity to better understand the multiple perspectives on diversity representation that exist in the marketplace.

If advertising has a role, is it to mold or just mirror society?

Finally, we identify a recurrent theme which poses the question: should advertisers mirror or mold society? While advertisers have a history of using stereotypes in advertising there is debate as to whether advertisers can influence an audience’s perceptions, attitudes, and values. Indeed, mixed views exist as to the evolution of stereotypes in ads, with some arguing that advertisements are becoming less stereotypical (Wolin, 2003) and others that they have become worse (e.g., Milner & Higgs, 2004). Regardless, Eisend (2010) suggests that any decreases tend to be due to developments in certain countries, like Japan, rather than being across the board.

The mirror argument suggests that advertisers reflect existing values in society (Holbrook, 1987). If advertising is reflecting society, the case can be made that it is not actively creating stereotypes; rather it simply reinforces those that already exist (Lantos, 1987). Under such conditions, it could be argued there is little wrong with the creation of messaging such as Uncle Ben, Mrs. Butterworth’s and other racist advertising portrayals. Of course, society’s views have changed over time, resulting in many brands to rebrand (e.g., Aunt Jemima in 2020).

In contrast, the mold argument suggests advertisers can shape an audience’s values (Pollay, 1986, 1987). One such way to mold values may be through challenging negative stereotypes, which can increase or normalize diversity in ads. Recent examples of such stereotypes being challenged can be found in social media accounts such as “Retirement House,” “Old Gays,” or Ross Smith’s “Granny” that find humor in generational divides (Locke, 2022). The Retirement House TikTok account has more than 3.8 million followers and features 70- and 80-year-olds doing their own takes on popular TikTok trends. That said, there is relatively limited understanding as to if and how advertising can break, or challenge, negative stereotypes.

Eisend (2010) provides a rare empirical test, finding support for the mirror over the mold argument, showing that in the context of gender, marketers react to gender-related developments in society and use existing values in a society to promote their brands rather than trying to alter these values. Even if this is so, it is not known to what extent these results carry over to other forms of diverse representation in advertising or are unique to representations of gender.

Developing a research agenda for diverse representation in advertising

This article develops a comprehensive framework for understanding how consumers respond to diverse representation in advertising. Our framework makes several contributions. First, it spans multiple forms of diversity effects in advertising, thus making it easier for researchers to start examining emerging forms of diversity, some of which are discussed below. Second, the framework ties together and organizes disparate research streams in the context of diversity in advertising. While each of these streams is relevant to understanding responses to diversity in advertising, synthesizing these literature streams provides a starting point for further theorizing and expansion on the topic of diversity in the context of advertising. Third, the use of diversity in advertising presents numerous additional questions pertaining to its implications for consumers and advertisers. Our framework assists in developing a research agenda to address these areas in the context of diversity, equity, and inclusion evolution within the marketplace (Arsel et al., 2022; Harmeling et al., 2021). To pollinate ideas for future studies pertaining to diversity in advertising, the following sections provide a brief overview of these ideas arranged according to several broad areas and summarized in Table 4.

Table 4 Potential research questions pertaining to diversity in advertising
Full size table

Better understanding diverse representation

Our review reveals that diversity representation in advertising has largely focused on surface-level characteristics, as Fig. 1 shows. Again, there are two perspectives on this. The first is that it is more natural for advertising to examine surface-level characteristics than deep-level ones. This is because advertising is commonly (although not always) a brief, non-verbal interaction with a consumer. A consumer develops a relationship with very few spokespeople, and surface-level features have been a historical staple for hiring in the modeling and acting industries (often precursors for casting in advertisements) for decades. Further, as the focus of the ad is commonly on a product and the model is mostly instrumental to that goal, it is harder (and often impossible) to discern deep-level diversity in them. Income is often used as an example of a deep-level characteristic, but there are many non-verbal cues (e.g., patterns of speech, dress) that aid third parties in making inferences about a person’s income. These are stripped away in most advertising, such that the model is representing the brand, not choosing their own words or clothing in ways that could enable the reader to discern their socioeconomic status. Other deep-level characteristics are even more problematic to discern in an advertising context, unless an advertiser calls them out specifically in the ad’s execution (e.g., a print ad noting a model’s autism), in which case they cease to be deep-level to the same extent. Hence, research could explore how to make deep-level diversity characteristics more apparent in advertising and how consumers respond to such characteristics.

Further, existing research has tended to tackle diversity from a siloed and Western perspective; hence, there is an opportunity to expand the scope of diversity research. First, there is opportunity to expand existing notions of diversity contexts to include broader perspectives. For instance, it is necessary to broaden the concept of gender beyond existing classifications of diverse groups and explore new ways to characterize and operationalize groups (Eisend & Rößner, 2022). While there is a clear need to consider gender beyond dichotomous gender roles (Peñaloza et al., 2023), it may also be possible to further break down and understand other forms of diversity, such as considering old age in terms of more nuanced subcategories (Prieler, 2020), such as young old, older old, and so on. Such understanding and operationalization might be applied across numerous diversity classifications. Further, there is opportunity to expand our understanding of diverse representation into non-Western contexts. While some research does exist in these contexts, as we have shown it is relatively sparse and limited in its application.

An additional area of possible research expands on a recent trend in advertising of brands using models that are either androgynous (Garst & Bodenhausen, 1997) or ethnically ambiguous (Booyse & Swanepoel, 2017). There is an opportunity to understand the effect of such ambiguity in the context of advertising. As an example, models with blended ethnic backgrounds have gained traction in the fashion industry, in a category the industry calls “ethnically ambiguous” or “exotic beauty” that transcends race or geography (Rabimov, 2022). Essentially, diverse and ambiguous models seem to have tremendous appeal to advertisers (Rabimov, 2022). Future research could set out to further understand this effect and consider if and how these effects might extend to other diversity groupings, such as LGBTQIA+ .

Further, moving away from a siloed approach to understanding the effect of diversity in the context of advertising could be beneficial. In doing so, research could better understand the role of intersectionality—or the ways multiple diversity elements come together and mutually shape and influence ad response. For instance, there is opportunity to better understand advertising in the context of people from ethnic groups who are also members of the LGBTQIA+ community. There is opportunity for much more research here.

Finally, there is opportunity for research to move beyond content analysis and counts in the context of diversity in advertising (Taylor, 2022). Future research might consider providing a deeper and more nuanced view of how diverse communities are portrayed. While many prior studies strive to evaluate the presence of negative stereotypes in advertising, they often fail to explain the processes behind these (Taylor, 2022). Further, future studies should seek to expand our understanding of diversity contexts that remain under-researched, such as diversity in the context of Indigenous people and decolonization efforts (Green, 1993), religion and spirituality (Waller & Casidy, 2021), political affiliation (Robideaux, 2002), and income (Hamilton, 2012).

Understanding diverse responses to diversity

While there’s a growing emphasis on broadening representation, advertisers grapple with multiple challenges in their pursuit. One pressing concern is understanding the array of consumer views and sensitivities related to representation. How might advertisers improve their insights into the range of consumer views and sensitivities related to diverse representation? Even as many brands and consumers advocate for more inclusive portrayals, it’s essential to foster dialogue that appreciates alternative viewpoints on the subject. Another hurdle is navigating the balance between those who advocate for more inclusive portrayals and the imperative to understand and respect those holding alternative viewpoints. What strategies can foster this mutual respect and understanding?

Furthermore, it’s important to acknowledge that not all cultures and nations share a Western perspective on diversity. How can research be tailored to appreciate that many cultural and national responses to diversity can differ from more prevalent Western perspectives? Global brands, with their vast outreach, can often find themselves at the crossroads of varied opinions. Their actions, no matter how well-intentioned, can be met with fervent responses, underscoring the complexity of the issue. How can these brands effectively chart a course in the face of varied opinions on diversity, especially when their initiatives might elicit strong consumer responses? More broadly, what additional unrecognized challenges linked to diverse representation exist, and if so, how do they operate?

Finally, a crucial question is how to overcome consumer hesitancy to diverse representation, particularly among majority groups. Representation in advertising media can have positive effects for minority and marginalized groups, and while advances have been made in recent years, there remains a need to develop understanding as to how to increase diversity and support for more of it. This is an important question for marketing given that the discipline often strives to address issues that have larger impacts on society, for instance addressing issues pertaining to obesity or mental health and well-being (Lee & Kotler, 2011). Extending this stream of research to investigate public stigma surrounding some of the diversity contexts in advertising would be fruitful. Specifically, research could investigate reasons for diversity stigma in advertising as well as develop understanding as to the associated message frames that could address or reverse these perceptions. Past research considers the ways in which message frames can encourage or support behavior change with a view to improve public health, safety, the environment, or community well-being (Sirgy & Lee, 2008). One way that consumers may learn to better tolerate different views on diversity is via perspective-taking, or the active effort to put oneself in another person’s shoes (Darwall, 1998). Perspective-taking is a form of empathy that allows an individual to imagine another person’s situation. To what extent can advertisers use this technique on audiences to encourage consumers to better understand and support those with differing views?

Exploring additional mechanisms, moderators, and outcomes

Our framework identifies a range of mediating mechanisms and moderating factors; however, there is certainly opportunity to explore additional mechanisms, moderators, and even outcomes.

Additional mechanisms

In terms of potential mechanisms, there may be different effects based on the overtness of diversity that is depicted. Prior research shows effects of similarity and aspiration in advertising (Naderer et al., 2021), however extreme levels of diversity can have an avoidance or repelling effect. As an example, in the case of socio-economic diversity, research shows that extreme levels of low social-economic status can have negative effects on consumer reactions and market participation (Park et al., 2023). These effects may be like the uncanny valley, whereby a more extreme depiction of diversity (e.g., a model who is a double amputee) leads to consumer response becoming more negative. There may be certain levels of diversity salience (van Knippenberg et al., 2004) that have mediating effects on consumer outcomes.

Developing understanding of viewer responses based on personality factors is important. In the context of organizational research, team personality is shown to be an important mediator between team diversity and performance (Homan et al., 2008). In this respect, teams, and individuals can differ in their attitudes and feelings toward working in diverse teams (Ely & Thomas, 2001). In the context of organizational teams, Homan et al. (2008) draw on the five-factor personality model, specifically the factor “openness to experience”, represented an individual’s degree of broad-mindedness, such as being open to novel experiences, and not being conservative (McCrae & Costa, 1987). Incorporating this concept in advertising research on diversity may offer fruitful contributions, as the effect of diversity salience may be contingent upon openness to experience.

Additional moderators

In terms of potential moderators, research might consider if advertisers are “blamed” for showcasing certain forms of diversity, as in the case of prior research which indicates that stigma can “rub off” on others (Patel et al., 2022). Drawing on attribution theory, Patel et al. (2022) show that consumers attribute a child being overweight to the child’s parents and accordingly stigmatize those parents (see also Jones, 1999). These findings illuminate the psychology underlying stigma toward parents of children with other potentially stigma-evoking identities and call into question whether similar effects could occur with brands. For instance, might brands who employ models of high weight be similarly stigmatized or “blamed”, with the brands ultimately experiencing negative effects from using diverse models?

While diversity is historically underrepresented in advertising, in recent times diversity has become more common across a range of media. As brands increasingly include diversity in ad campaigns, the question arises as to when and how the use of diverse models is perceived as tokenistic. Indeed, it may be the case that there is an inverted-U relationship between the degree of diversity in advertising and effectiveness. In this way, underrepresentation is viewed negatively, whereby not enough diversity has negative effects, but similarly overrepresentation may also have negative effects, being perceived as being overdone, tokenistic, or opportunistic.

Additional outcomes

While existing research reveals some negative effects to diversity in ads—such as the potential for distraction—there is opportunity to better understand the potential for negative effects. Further knowledge is also required pertaining to the relationship between stigmatized models and advertisers. For instance, while it is known that endorsers have associations that can transfer to the brand, it is possible that associations with models may also transfer to the brand. This poses a challenge for brands that engage diverse models in ads that may have certain negative associations or stereotypes (e.g., obese models). However, with increased use by a range of advertisers, the associations with these groups could become “re-branded” or de-stigmatized.

Accurately representing diversity

Echoing prior research (e.g., Eisend et al., 2023), our review highlights the predominant focus of research on a relatively narrow set of diversity areas (i.e., gender, race and ethnicity) and highlights opportunities to expand our knowledge of diverse groups. One question that arises is which diverse groups should be represented? The case could be made that consumers representing all forms of health and behaviors could be represented. On the contrary, critics could argue that people engaging in unhealthy behaviors (e.g., some drug users or obese consumers) should not be represented in equivalence with their prevalence. Future research might consider the degree to which behaviors that may be unhealthy and something that society is aiming to reduce, should be represented. Or should there be certain diverse elements (i.e., potentially harmful behaviors) that should not be represented? Either way, researchers need to exert care and clearly differentiate stigmatizing the behavior from stigmatizing or excluding the person. This point has been made with respect to food choice (Lin & McFerran, 2016; Puhl & Heuer, 2009) as well as in smoking research (Bresnahan et al., 2013), and drug use (Lloyd, 2013).

For advertisers, the process of moving towards more accurate representation may be aided through better understanding of the effect of habituation. Habituation, termed the simplest form of learning, refers to behavioral response decrement that results from repeated stimulation (Rankin et al., 2009). In the context of diversity, repeated exposure to ads with diverse models or representing diverse groups may lead to individual habituation and greater acceptance of diversity in general. Understanding how this process works and the limits of representation that can occur without triggering reactance may enable advertisers to subtly adjust representation in their ads. In a similar vein, accurately representing diversity comes with a balance between employing diversity in a meaningful way and not overreaching or being tokenistic. It may be that overreaching draws too much attention to diversity efforts. Parallels can be drawn to research showing disclaimers about digitally enhanced ads may not actually reduce body dissatisfaction or negative mood (Borau & Nepomuceno, 2019). These effects can be explained in that ad disclaimers about a digitally enhanced model draw more visual attention, leading to more negative effects (Bury et al., 2016). Research should consider if and how advertisers can combat these possible effects in the context of diversity, finding a balance between too little diversity and too much (i.e., tokenistic) diversity for a given context.

One strategy might be for advertisers to be less intentional, or more natural, in their representation of diversity. In this way advertisers could strive for a more random allocation of diversity into ads. For instance, if advertisers were to characterize the composition of their audience and then use this to randomly feature models in ads, diversity might feel more natural and less “forced” to some segments. Developing ads without considering the specific models that will be used might also make diversity seem more incidental as well as ensure full representation. Implementing a more natural approach ultimately carries philosophical implications. What population should be sampled from to ensure diversity (Eisend 2022)? Should it be one’s current consumers, a target market, a region, a nation, or the planet? Each would have very different implications for diversity and representation. In other words, the prominence of some groups would increase, but others decrease, depending on the aperture of the lens used. Do advertisers have an obligation to oversample historically underrepresented groups in advertising, including those identifying as differently abled?

Such a move to be less intentional, or more natural, would be consistent with the literature and should have minimal impact on sales. Over 60 years ago, Yankelovich (1964) noted that demographics were poor predictors of behavior. Forty years later, at a point where media was far more fragmented and society more diverse and inclusive than it was in the 1960’s, Yankelovich and Meer (2006) still commented:

“...market segmentation - which, if properly applied, would guide companies in tailoring their product and service offerings to the groups most likely to purchase them. Instead, marketing segmentation has become narrowly focused on the needs of advertising, which it serves mainly by populating commercials with characters that the viewers can identify with - the marketing equivalent of central casting” (p.1).

Marketing classes still teach students to represent a typical consumer in a segment, give them a name, some visuals, and the like. Since this person is likely to represent an actual (or perhaps aspirational) member for the segment, it is possible that advertising often misses the mark and under diversifies, but occasionally over diversifies and appears tokenistic or diversity-washing (Vredenburg et al., 2020).

Finally, more accurately representing diversity in the context of advertising might be to consider the ways in which multiple meanings can be drawn on. The use of models that elicit multiple models might have broader effects for a wider group of people. Such a strategy might be akin to that of polysemy in advertising, or “the occurrence of multiple meanings for the same advertising message across the members of an audience” (Puntoni et al., 2010, p. 51). By drawing on a variety of diversity elements and cues in ads, advertisers can engage in strategic ambiguity and might be able to ensure that the ad message reaches different audiences in different ways, with purposeful messages for each audience. In addition to the ad itself, advertisers can also imbue ads with meaning through the advertising medium used to transmit the ad (Malthouse et al., 2007).

Better understanding why advertising isn’t more diverse

Our analyses lead to the conclusion that advertising, for the large part, remains lacking in diversity (see Fig. 1). There is an opportunity for future research to develop and test theoretical explanations as to why such gaps persist, despite awareness and repeated calls for change from within and outside the industry. One explanation may be that the majority is often depicted simply because they are the majority and thus statistically will appeal to the largest group of consumers. Further, since even today media are still relatively large, concentrated, and monolithic (Pollay, 1987), perhaps it is simply more efficient to produce messages aimed at conformity and a single standard even if they may not resonate perfectly with all segments?

There are likely other explanations that exist as well, since these points alone paint an insufficient picture of why diverse models remain underrepresented, as none of these arguments hold up well to historical scrutiny. While still not representative, the ad industry has become more diverse over time, with females and minorities nearly absent at one point. Media is also now more fragmented, so media and marketers have an increased ability to target multiple segments. Marketers have long known that there are benefits to segmentation, and that responses to marketing efforts are improved when consumers are targeted in segments as opposed to en masse (Holbrook, 1987). So why have more fragmented media, increased ability to target multiple segments, and changing demographics not made diversity more present in advertising?

One potential explanation may lie in borrowing theory from economics, namely that there could be a tragedy of the commons at play (Hardin, 1968). If Whites make up the majority of America (61.5%), the modal advertising model should (statistically speaking) be white, as should the modal consumer. A problem arises when and if all firms act in this same manner, resulting in a production mechanism for advertising that is structurally designed to under-represent minority groups when scaled up to the aggregate. In other words, similar to a “Prisoner’s Dilemma” strategy game (Axelrod, 1980), a problem arises if each firm acts in its own (perceived) self-interest. This is since each advertiser is individually incentivized toward actions that, taken together, collectively hurt everyone akin to a negative externality in economics.

Similarly, perhaps when diversity is valued, organizations seek to signal their greater openness than rivals, with increasing and increasing levels of diversity, resulting in some groups being overrepresented. However, this signaling means that some groups are systemically left out more than others. Since individuals are being selected in advertising as models based on their demographic traits and ability to offer “evidence” supporting an organization’s diversity or positive stance towards it, individuals possessing highly visible, surface level traits are overrepresented. This means that individuals who are diverse in terms of more deep-level characteristics will continue to remain vastly under-represented. Does such an explanation help explain why diversity is not uniformly represented in advertising? If not, what other theories can be drawn on or developed to understand current levels of diversity representation in advertising?

Understanding the effects of technology and personalization

While there are more and more niche media to target with each passing year, technology itself provides an interesting opportunity for increased diversity in advertising. For instance, the effect of diversity in the context of digitally manipulated advertising, such as artificial intelligence, will likely evolve to play an increasingly prominent role in advertising (Campbell et al., 2022a). In this context questions arise as to how consumers might respond to manipulated advertising when diversity is artificially controlled. The manipulation of diversity in influencers has already begun, with the world’s first virtual influencer with Down Syndrome—Kami—recently created and engaging with consumers on Instagram (Kiefer, 2022). As technology develops, the question arises as to whether increased hyper-reality leads to ads and brands being perceived positively, negatively, or just “different”; in this way does being clearly different from reality lead people to be accepting of diversity? Or is it possible that manipulated advertising is viewed as a form of “blackfacing” a model, leading to negative perception of the brand?

Another area technology will likely impact is personalization in advertising (e.g., Sahni et al., 2018), whereby technology could match consumers with models they are likely to respond more favorably to. Such matching could be based on a variety of factors that go beyond mere demographic characteristics. One can imagine a world where AI can assist personalization in such a way that consumers are exposed to ads that are more representative because that is what consumers prefer. AI is already being used by Levi’s to increase the diversity in its models (Pasquarelli, 2023; Shanklin, 2023). Others have written about the potential benefits for diversity in other domains that technology will soon offer (Lobel, 2022).

However, there are also cautionary warnings regarding the effect of technology. What if personalization leads people to prefer less rather than more diversity? What if (mostly ideal beauty) human models continue to get jobs, and diverse models are replaced by AI? It’s very conceivable that those who are least tolerant and accepting of diverse models would become the least likely to be exposed, perhaps entrenching explicit and implicit biases. Research is needed that examines these potential effects.

Conclusion

The multifaceted nature of diversity in advertising paints a complex picture. While the literature predominantly underscores the benefits of diversity there has been consumer resistance to brands engaging diverse models in advertising. Such tension raises questions about the objectivity of our current understanding of the impact of diverse representation in advertising. This article contributes in several ways. Beyond confirming that diverse representation is often lacking within advertising, we provide a review across multiple forms of diversity and diverse representation in advertising and present overarching insights into the current trajectories of research on diverse representation. From that, we develop a framework that unifies the different knowledge bases, highlighting common moderators, mechanisms, and social and commercial outcomes across the fragmented literature. We conclude by discussing the current state of scholarship on diverse representation and offer a suggested program of future work, as well as recommendations for advertisers.

Just as we have written about the importance of viewing diversity holistically, diversity in advertising should be viewed alongside diversity in other media. For example, diversity in acting roles may also be increasing in motion pictures, with leading roles being given to actors from underrepresented groups (including the Asian superhero, Shang-Chi). There may also be more diverse directors, screenwriters, and scripts. Indeed, there are likely lessons from experiences in other media and domains that are relevant but outside the scope of our review. In addition to diverse representation advertising, diversity representation in other forms of media may also affect consumer expectations and norms in the context of diversity representation.

In essence, much remains unknown. For instance, consumers, especially those from diverse backgrounds, don’t offer a monolithic view. While there’s a clear demand for accurate representation, the parameters of this accuracy remain under-researched. Further, the role of advertising in society is a further complication. As consumers seek to have new forms of diversity represented (i.e., surface vs deep-level traits), should advertisers strive to reflect the existing societal fabric, or should they aspire to actively shape societal norms? The prioritization of surface-level diversity over deep-level characteristics in advertising signals a deeper systemic issue, emphasizing the need for a more nuanced understanding and appreciation of diversity in all its forms.

In sum, as the advertising landscape continues to evolve a balanced and informed approach is essential. With this article we hope to both recognize the progress that has been made and the challenges that lie ahead. The advertising industry is poised at an intersection where technology will soon become available to advertisers that could aid, but could possibly detract from, diversity efforts. Such questions are important to help advertisers understand the pros and cons to such technology adoption in ad creation. We hope this article plays a part in shaping diversity in advertising going forward.