Abstract
Although research on ethnic diversity in advertising is extensive, its findings remain fragmented and often inconclusive. This limits practical guidance for marketers. Beyond marketing- and consumer-specific factors, differences between ethnic minority groups within a country and cross-national socioeconomic and cultural factors can play a role in shaping responses to ethnic diversity in advertising of consumers belonging to either ethnic majority or minority groups. This study develops and empirically tests a theoretical framework to examine the impact of ethnic diversity in advertising along with the factors that moderate this impact. The authors conduct a meta-analysis based on 1,276 effect sizes derived from 106 papers published between 1969 and 2024, encompassing 155 unique datasets from 21 countries. The findings reveal that ethnic diversity effects are small, but are moderated by country-level variables such as migration rates and rule of law, in addition to product, advertising, and consumer characteristics, and that some of these moderator effects work differently for ethnic minority compared with ethnic majority consumers. Furthermore, within the U.S. context, the sociopolitical power position of ethnic minorities enhances advertising effectiveness among majority consumers. These insights offer implications for both researchers and practitioners.
Many countries have experienced significant demographic shifts, resulting in increased ethnic diversity (Bai, Ramos, and Fiske 2020). Ethnic diversity refers to both the fragmentation and composition of multiple ethnic groups within a society (Dinesen, Schaeffer, and Sonderskov 2020). This growing diversity has become a critical consideration for marketers worldwide, prompting a greater incorporation of ethnically diverse representations in advertising in efforts to appeal to increasingly ethnically diverse consumer audiences (Unstereotype Alliance 2024). Recent research indicates that ethnic diversity in advertising—operationalized by the percentage of ethnic minority endorsers—has increased globally, reflecting the rising proportion of ethnic minorities in many societies (Rößner and Eisend 2023). Correspondingly, U.S. advertising and marketing spending in media targeting Hispanic, Black, and Asian American audiences rose by 5.7% to USD 34.64 billion in 2023 and is projected to increase by 8.3% to USD 45.83 billion in 2024 (PQ Media 2024). 1 Although multiethnic advertising and marketing accounts for approximately 5% of total ad and marketing spending (PQ Media 2024), most advertisers still perceive progress in diversity and inclusion as limited (Association of National Advertisers Alliance for Inclusive and Multicultural Marketing 2024). This is notable given that diversity in advertising has been associated with significant business benefits, including an optimistic 16.3% increase in long-term sales and a 15% rise in customer loyalty (Unstereotype Alliance 2024).
Beyond its managerial relevance, ethnic diversity in advertising continues to attract substantial academic attention (e.g., Davis, Smith, and Sevilla 2025; Khan et al. 2025). Numerous studies have examined the impact of ethnic diversity by comparing advertisements featuring either ethnic majority or minority endorsers, and by assessing responses from consumers. However, findings have been mixed and have shown positive, negative, and neutral effects of ethnic diversity in advertising (see Web Appendix A). These inconsistencies continue when distinguishing between ethnic minority and majority consumers. While ethnic minority consumers tend to respond neutrally or positively to ads featuring coethnic endorsers (e.g., Aaker, Brumbaugh, and Grier 2000; Shao et al. 2023), a pattern attributed to mechanisms such as targeting, perceived similarity, and identification, responses from ethnic majority consumers range from positive (e.g., Gan and Chen 2024) to neutral (e.g., Antioco, Smeesters, and Boedec 2012) to negative (e.g., Rajabi et al. 2017).
Previous efforts to explain the heterogeneity in ethnic diversity effects have focused on marketing-related factors (e.g., product type) and consumer differences (e.g., age, gender, ethnicity). For example, Desphandé and Stayman (1994) show that consumers’ reactions to endorsers of varying ethnicities depended on the ethnic composition of U.S. cities. Both minority and majority consumers showed a preference for endorsers of their own ethnicity—a tendency that is amplified when the respective group represents a numerical minority in a city. However, such findings are often context-specific and may not generalize across national or cultural boundaries. While data on ethnic diversity in advertising have been collected from various countries over several decades, cross-national differences remain underexplored. Furthermore, the composition and status and power positions of ethnic minority groups vary across countries (Omi and Winant 2014), which suggests that endorsers belonging to different minority groups might lead to varying effects on consumers within the same country. A more comprehensive understanding of the effects of ethnic diversity in advertising—along with how these effects are moderated by cross-national differences, ethnic diversity composition in a country, and marketing-related factors—is thus essential. Importantly, these moderator effects can differ between ethnic minority and majority consumers. Together, these insights can offer novel theoretical contributions and practical guidance for both domestic and international campaigns.
Against this backdrop, we develop a theoretical framework to reconcile the inconsistencies in existing research. We examine the effects of ethnic diversity in advertising and aim to explain the heterogeneity in findings. Specifically, we consider (1) between-country differences, (2) within-country diversity characteristics, and (3) marketing-related factors (e.g., product category, advertising medium). We test how these factors moderate the effects of ethnic diversity on all consumers, as well as minority and majority consumers separately. We explain differences in responses by minority and majority consumers by social categorization.
To explain between-country variations, we draw on two theoretical perspectives—exposure and contact and institutional theory—both of which posit that broader sociocultural contexts shape minority and majority consumer responses to ethnic diversity in advertising. Research in fields such as sociology has long highlighted the importance of socioeconomic and cultural factors in shaping the relationships between ethnic majority and minority groups (e.g., Dima and Dima 2016; Dinesen and Sonderskov 2015). However, advertising research has rarely considered comparisons between countries. When it has, such studies have typically included only a few countries and yielded limited generalizability (e.g., Gao, Xu, and Kim 2013; Lwin and Wee 2000). To examine within-country variation in ethnic diversity composition, we focus on U.S.-based data and analyze the effects of using different ethnic groups as endorsers, including consideration of their status and power positions.
To test our theoretical framework, we conduct a meta-analysis of 1,276 effect sizes related to ethnic diversity in advertising, drawn from 106 papers published between 1969 and 2024. These studies comprise 155 unique datasets from 21 countries, offering a unique opportunity to examine the influence of time-varying country-level moderators such as migration rates. To explore within-country dynamics, we pay particular attention to the United States, which accounts for over 60% of the data in our sample.
Our findings offer empirical, theoretical, and managerial contributions. First, we synthesize prior results and show that the overall effects of ethnic diversity in advertising are modest, while they are significantly more positive for ethnic minority consumers than for majority consumers. This supports the strategic use of ethnic diversity in advertising to engage minority audiences. Moreover, since such advertisements do not negatively affect majority consumers, a key concern among advertisers (Schlegelmilch, Khan, and Hair 2016) is mitigated. Second, we identify important country-level moderators, demonstrating that migration rates and institutional strength (e.g., rule of law) influence consumer responses to ethnic diversity in advertising. Third, we find that the impact of minority endorsers is contingent on their power and social status within a given national context. These results clarify how ethnic diversity functions within and between countries and underscore the importance of adapting advertising strategies to both local and international contexts. Finally, we identify several marketing-related factors that moderate advertising effects, thereby providing actionable insights for practitioners seeking to optimize ethnically inclusive campaigns. Overall, these findings can help marketers navigate the trade-off between inclusive versus targeted ethnic advertising strategies, addressing the longstanding challenge of “targeting without alienating” (Johnson and Grier 2011). Ethnic diversity in advertising can be effectively used not only in targeted media but also in mainstream media without risking backlash from majority consumers.
Conceptual Background
Definitions and Research Framework
Ethnicity is defined as a social construct representing “clusters of people who have common culture traits that they distinguish from those of other people” (Smedley and Smedley 2005, p. 17). These traits refer to a common language, place of origin, historical background, traditions, values, beliefs, and a sense of collective identity. In line with Dinesen, Schaeffer, and Sonderskov (2020), ethnic diversity refers to the presence of multiple distinct ethnic groups within a society, capturing the degree of ethnic fragmentation and composition. In advertising, ethnic diversity is typically implemented through the inclusion of endorsers who belong to ethnic minority groups. Due to limited consumer attention and cognitive processing during advertising exposure, identification of endorsers’ ethnicity is often guided by salient visual cues, such as facial features, skin tone, eye color, and hair texture. Less commonly, advertisers employ cultural cues such as clothing, names, or references to heritage to signal ethnicity (e.g., Rößner, Gvili, and Eisend 2021).
Consumer perceptions of and responses to ethnic minority representations in advertising are shaped by various attributes associated with those minority groups. These include their relative population size, social status and power, categorical distinction from the majority, group-specific characteristics, and the nature of their treatment by the majority population (Seyranian, Atuel, and Crano 2008). Ethnic group distinctions can be either manifest (e.g., observable traits or group characteristics) or latent (e.g., perceived social power or status), each of which can influence advertising responses in different ways.
Figure 1 outlines the research framework guiding this meta-analysis. The central relationship under investigation is the effect of ethnic diversity in advertising on persuasion (i.e., positively valenced responses toward advertising, including attitudes and behavioral responses). The framework incorporates four categories of moderator variables to account for heterogeneity in advertising effects: (1) minority versus majority consumers, (2) country-level variables, (3) ethnic diversity variables, and (4) other (including marketing) variables. For the first category, we distinguish between ethnic minority and majority consumers, drawing conceptually on social categorization theory. Moderators in the second to fourth categories are theorized and analyzed both as main effects and in interaction with consumer groups (i.e., minority vs. majority). To explain country-level differences in response to ethnic diversity in advertising, we draw on two complementary theoretical perspectives: exposure and contact, and institutional theory. In addition, other factors—including time, product and brand characteristics, consumer demographics, media, and methodological differences—are considered to account for further variability across studies and effect sizes. The composition of and relationship between different ethnic diversity groups, endorsers, and consumers are country-specific and difficult to assess by a meta-analysis, as most studies capture only few selected minority groups in a specific country. To better understand within-country effects of ethnic diversity, we investigate variations in advertising effectiveness based on the specific identity of the ethnic endorser and the ethnic group's relative social status and power. We use data from the United States, which comprise a large proportion of our meta-analytic dataset, and assess multiple ethnic minority group endorsers and consumers.
Research Framework.
Explaining Responses of Minority and Majority Consumers to Ethnic Diversity: Social Categorization
People tend to think categorically about others (Macrae and Bodenhausen 2000). Social categorization is a fundamental psychological process through which individuals classify others into distinct groups, allowing them to anticipate behavior and situate themselves within the social world (Tajfel 1978). These categorizations often draw on stereotypes—simplified, generalized beliefs about social groups formed through prior experience and selective exposure to environmental cues (Fiske and Neuberg 1990). While stereotypes can be either positive or negative, those applied to outgroup members tend to be more negative in valence, whereas ingroup stereotypes are typically more favorable (Allport 1954).
Social categorization theory provides a valuable lens for understanding consumer responses to ethnic diversity in advertising. For ethnic majority consumers, exposure to minority-group endorsers can activate implicit biases, resulting in automatic negative reactions—even when individuals consciously support principles of diversity and inclusion. These reactions are often not rooted in overt prejudice but rather in ingroup favoritism—the unconscious tendency to show trust, empathy, and positive regard toward those perceived as similar or belonging to the same social group (Hewstone, Rubin, and Willis 2002). Consequently, majority-group consumers may withhold positive evaluations of diverse advertisements not necessarily because they dislike the minority outgroup, but because they reserve positive affect for the ingroup.
Country-specific influences on ethnic diversity effects can differ between minority and majority consumers, as both the salience of social categories and the stereotypes that are associated with them vary (Rhodes and Baron 2019). For example, majority-group consumers in ethnically more homogeneous societies such as China, South Korea, and Japan tend to respond more negatively to multiethnic advertising (Baek, Lee, and Oh 2022). In contrast, consumers in multicultural societies like Canada often associate such advertising with inclusiveness and social openness (Strebinger et al. 2018). Ethnic minority consumers are more likely to respond positively to advertisements featuring minority-group endorsers, whom they categorize as members of their ingroup (Tajfel 1978). This identification promotes feelings of representation, recognition, and self-relevance, which in turn enhance positive advertising responses (Campbell et al. 2025).
Explaining Between-Country Effects of Ethnic Diversity on Minority and Majority Consumers with Country-Level Variables
Drawing from sociological and related interdisciplinary research, we propose two overarching theoretical perspectives to account for country-level differences in consumer responses to ethnic diversity: (1) exposure and contact and (2) institutional frameworks. 2
The first theoretical perspective refers to exposure and contact. Individuals’ exposure to ethnically diverse environments significantly shapes their acceptance of ethnic minorities (Dinesen and Sonderskov 2015), thereby influencing attitudes toward ethnic diversity in advertising. Exposure and contact vary across countries due to differences in ethnic compositions in a country, which in turn has differential effects depending on ethnic group membership. Two opposing predictions emerge for the impact of such exposure. First, increased exposure to ethnic outgroups may lead to negative reactions among ethnic majority individuals. This is attributed to perceived competition over limited resources such as employment. This heightens intergroup threat—the perception that outgroups pose a challenge to the welfare or goal attainment of the ingroup (Riek, Mania, and Gaertner 2006). In contrast, for minority consumers, exposure to minority-group endorsers can reduce perceived threat, as they are less likely to see themselves in competition with majority members. As a result, increased exposure to ethnic diversity is expected to decrease (increase) ethnic diversity effects on persuasion for majority (minority) consumers; that is, exposure leads to less (more) favorable responses toward ethnic diversity among majority (minority) consumers.
Conversely, intergroup contact theory (Allport 1954) posits that contact with members of different ethnic groups reduces prejudice and promotes favorable attitudes. Meta-analytical evidence supports the benefits of intergroup contact (Pettigrew and Tropp 2006), attributing them to factors such as interethnic cohesion, cross-group friendships, and increased familiarity. Accordingly, environments that facilitate exposure of majority consumers to minority-group endorsers can enhance positive responses, while minority consumers benefit from identification with the endorsers (Tropp and Pettigrew 2005). From this perspective, increasing ethnic diversity should yield positive persuasion effects for both groups.
To empirically assess these mechanisms, we use two proxy variables: net migration rate 3 and ethnic fractionalization (Drazanova 2020), and test the interaction with minority versus majority consumers. Net migration rate reflects the likelihood of exposure to outgroups due to migration flows, while ethnic fractionalization captures the diversity of ethnic groups within a population. The former is more closely linked to perceptions of intergroup threat, as it signals an increase in the number of outgroup members. The latter captures ethnic variety without necessarily indicating increases in any minority group's size.
The second theoretical framework refers to institutions, which are “the humanly devised constraints that structure political, economic and social interaction. They consist of both informal constraints (sanctions, taboos, customs, traditions, and codes of conduct), and formal rules (constitutions, laws, property rights)” (North 1991, p. 97). Institutions establish societal expectations and shape individual behaviors. The institutional perspective suggests that country-level rules and norms influence the societal acceptance of ethnic diversity but vary between minority and majority group members, who are differently affected by the level of acceptance of ethnic diversity. Burgess and Steenkamp (2006) identify three interrelated institutional subsystems: regulative, cultural, and socioeconomic.
The regulative system refers to the formal rules and legal frameworks that govern behavior. Stable democratic institutions are linked to greater tolerance (Peffley and Rohrschneider 2003). Strong regulative systems promote equality and freedom, reducing perceived intergroup differences and fostering inclusivity (Lehmann and Seitz 2016). Weak regulative institutions, by contrast, may reinforce perceived disparities, prompting minority consumers, but not majority consumers, to respond more favorably to diverse advertising due to increased solidarity with other underrepresented groups. We measure the regulative environment using rule of law indicators and use a democracy index to conduct a robustness check.
The cultural system encompasses societal values and norms. Two of Hofstede’s (2001) cultural dimensions are particularly relevant: power distance and achievement values (also known as masculinity). Power distance—the extent to which unequal power distribution is accepted—discourages diversity, as it reinforces hierarchical group distinctions (Van der Vegt, Van de Vliert, and Huang 2005). While societal norms of high achievement-oriented cultures include “ego orientation” and “money and things are important,” societal norms of low achievement-oriented cultures cover “relationship orientation” and “quality of life and people are important” (Hofstede 2001, p. 299). High achievement-oriented cultures emphasize individual success, assertiveness, and toughness over interpersonal relations, empathy, and altruism, often at the expense of inclusivity, as has been shown for gender equality and stereotyping in advertising in different countries (Eisend 2010). High achievement values are associated with lower tolerance for diversity in advertising (Eisend and Hermann 2019) and negative attitudes toward immigration (Leong and Ward 2006). In cultures high in power distance and achievement values, ethnic majority consumers may respond less favorably to minority-group endorsers. Conversely, minority consumers may experience heightened group identity and solidarity, leading to more positive responses toward ethnic diversity (Bernstein 1997).
The socioeconomic system relates to the influence of economic structures and entrepreneurial activity. Income inequality, often measured by the Gini index, signals how evenly resources are distributed. High inequality tends to foster perceptions of a competitive, fragmented society, reducing trust and increasing social intolerance (Dima and Dima 2016; Uslaner 2002), which is in line with the preceding arguments of intergroup threat and perceived competition over limited resources. Conversely, entrepreneurship and innovation—measured by population-weighted patent registrations—are associated with openness, talent attraction, and creativity, all of which foster inclusivity and positive attitudes toward diversity (Florida 2003; Freitag and Rapp 2015; Qian 2013). Therefore, we expect responses to ethnic diversity in advertising to deteriorate with rising inequality and to improve with innovation for both minority and majority consumers. To control for broader economic conditions and as a robustness check, we also control for GDP growth rate and GDP per capita, both of which have been linked to greater societal tolerance and diversity acceptance (Eagle, Macy, and Claxton 2010).
Explaining Effects of Ethnic Diversity Variables on Minority and Majority Consumers Within a Country
Ethnic diversity is a multifaceted construct shaped by cultural complexities and specific historical trajectories. Ethnic groups can be distinguished not only by overt characteristics such as group membership but also by more nuanced, latent factors—particularly those related to power and status. These factors influence individuals’ access to social, economic, and political resources, as well as their broader standing within the societal hierarchy (Omi and Winant 2014).
The composition and relative positions of ethnic minority groups vary across countries. To explore within-country differences in consumer responses to ethnic diversity, we focus on the United States—the only country in our dataset with sufficient data points to distinguish between major ethnic minority subgroups. Specifically, we examine three groups: Black, Hispanic, and Asian Americans. These groups differ not only in demographic representation—19.1% Hispanic/Latino, 12.6% Black, and 6.1% Asian American as of 2022 (USA Facts 2024)—but in their historical experiences and socioeconomic positioning (Humes and Hogan 2009).
The social positioning of Black Americans has been profoundly shaped by the transatlantic slave trade and its enduring legacy of institutionalized racism, including slavery and legalized segregation. Although significant legal and political advances were achieved during the Civil Rights Movement of the 1960s, structural inequalities persist, particularly in wealth accumulation, education, and the criminal justice system (Humes and Hogan 2009). The Hispanic/Latino population has deep roots in the United States, linked to historical territorial shifts following the Mexican-American War in 1848, and continued migration from Latin American countries, particularly Mexico, Cuba, and Puerto Rico. Hispanics often encounter systemic barriers such as lower educational attainment, labor market discrimination, and multifaceted discrimination based on language, phenotype, and generational status (Cano et al. 2021). By contrast, Asian Americans—though subject to exclusionary policies in earlier periods—experienced a shift following the Immigration and Nationality Act of 1965. It eliminated national origin quotas and enabled the immigration of highly educated individuals (Humes and Hogan 2009). Today, Asian Americans enjoy higher average levels of education and income, comparable to or exceeding those of White Americans, and in contrast to other minority groups.
To empirically assess ethnic diversity effects, we run two models: one that investigates group categorization influences (i.e., whether the ethnic minority endorser is Black, Hispanic, or Asian American) and one that investigates differences in power and social standing, for which we use education and income of the endorser minority group as proxies. These structural indicators help capture the hierarchical positioning of ethnic groups within the United States. Research on social dominance orientation (Jost and Thompson 2000) provides insight into how such hierarchies influence persuasion: Majority group members tend to score higher on social dominance orientation and are more likely to endorse social hierarchies, whereas minority group members are more egalitarian. Powerful minority endorsers may be perceived by their own group as atypical exemplars that do not fit existing schemas, which can lead to negative responses (Campbell et al. 2025), as identification and ingroup favoritism is reduced (Tajfel 1978). Research shows that individuals who succeed in counterstereotypical, high-status domains can experience social backlash, as their success challenges existing group hierarchies and norms (Phelan and Rudman 2010). Backlash arises when such atypical exemplars are perceived as violating prescriptive status expectations, leading to resistance and reduced identification within their own ingroup (Rudman et al. 2012). Consequently, majority consumers may respond more positively to minority endorsers with high status, while minority consumers may not favor such high-status minority endorsers due to perceived distance or lack of shared identity.
Effects of Other Variables
Table 1 provides an overview of the moderator variables illustrated in Figure 1, along with their operational definitions, theoretical explanation, and expected effects. The group of other variables includes several moderators pertaining to time, consumers, endorsers, products, media channels, and methodological approaches. 4 Temporal variation in effects of ethnic diversity in advertising acknowledge that immigration trends, economic shifts, and political developments can cause fluctuations over time (Meuleman, Davidov, and Billiet 2009). Recent meta-analytic work (Lenk, Hartmann, and Sattler 2024) finds a temporal trend showing increased acceptance of minority endorsers by White majority consumers over time. However, given the complex and dynamic nature of these factors, the direction of temporal influences remains difficult to predict. Apart from the moderators related to methodological differences, we do not offer specific hypotheses for the other variables, neither for their main effects nor for the potential interaction with minority/majority consumers. An exploratory analysis of their influence can still yield valuable managerial insight and inform more effective marketing decisions, as the marketing-specific factors are endogenous to marketers and thus subject to strategic manipulation.
Explaining Effects of Ethnic Diversity in Advertising: Moderators and Their Expected Effects.
Method
Paper Retrieval
We identified both published and unpublished studies that quantitatively assessed the effects of ethnic diversity in advertising—specifically, the impact of ethnic minority endorsers relative to ethnic majority endorsers—on consumer responses. To identify relevant papers, we performed a keyword search in electronic databases (Web of Science, EBSCO, ProQuest, Google Scholar, and SSRN). 5 We scanned two seminal articles (Aaker, Brumbaugh, and Grier 2000; Desphande and Stayman 1994) and reviews of the extant literature (Kareklas and Polonsky 2011; Sierra, Hyman, and Heiser 2010, 2012) for references, as well as articles citing them. Next, we reviewed the reference lists of obtained papers. The search encompassed all studies published up to December 2024. Our approach follows procedures used in recent meta-analyses (e.g., Roschk and Hosseinpour 2020; Rosengren et al. 2020).
We included studies that provided quantitative comparisons of ethnic minority versus majority endorsers in advertising and measured their impact on at least one consumer response variable. Studies were excluded if they fell outside this scope or did not provide sufficient data for the computation of effect sizes, and where such data could not be obtained from the original authors. Eligible studies were required to be written in English. All studies are based on experimental designs, thereby ensuring causal inference and directional clarity in the effect sizes.
To avoid duplication, we defined a “paper” as any unique document presenting original analyses (e.g., journal articles, working papers, conference proceedings, or unpublished theses). A single paper could contain multiple datasets, also labeled as “studies.” For our analysis, we used these datasets as the primary unit of analysis. Each dataset could contribute one or more effect sizes, depending on how many consumer response variables it addressed.
Coding
To ensure generalizability, we focused on 13 consumer response variables that were sufficiently represented across studies (i.e., each had at least ten reported effect sizes). These response variables, along with their definitions, are presented in Table 2. Two independent coders (i.e., one author and a trained research assistant who was unaware of the research hypotheses) assigned each effect size to the appropriate response category. Intercoder reliability was high (Cohen's κ = .958; agreement rate = 96.5%), and any discrepancies were resolved through discussion. Moderator variables are presented in Table 3. Coders received instructions in line with the definitions and description in Table 1 and Table 3. Where coding ambiguity existed, two independent coders conducted the assessments. Intercoder reliability scores are noted in the table. Disagreements were discussed until consensus was reached. For variables involving low-inference judgments, coding was conducted by a single coder and subsequently validated by one of the study authors. With the exception of power distance and achievement values, all country-level variables and ethnic group characteristics were time-varying. We matched values to each dataset based on the country and year of data collection. For example, if a study was conducted in the United States in 2020, we assigned the U.S. value for net migration in that year. Similarly, if an Asian endorser was used in a 2020 study, we assigned the 2020 per capita income for Asians in the United States. In cases where the year of data collection was not reported, we estimated it by subtracting three years from the publication year—an approach based on the typical time lag between data collection and publication.
Consumer Response Variables.
Moderator Variables.
For definitions of the variables, refer to Table 1.
Descriptives refer to the effect sizes and datasets that were used for the main meta-regression model.
The intercoder reliabilities (agreement rate [kappa]) were as follows: minority and majority (100%), endorser gender (100%), brand type (95.1% [.89]), product type (99% [.99]), product involvement (98% [.96]), and product benefit (91.1% [.88]).
For ethnic fractionalization, participant age, and participant gender, missing value imputations were applied, following the procedure in former meta-analyses (e.g., Babic Rosario et al. 2016; Peng et al. 2023). Ethnic fractionalization has 40 missing values (effect sizes); participant age, 549; and participant gender, 129. We apply alternative imputation techniques as robustness checks.
The descriptive data refer to the reduced set of data collected in the United States.
We used the correlation coefficient as an effect size metric. A positive value indicates that positively (negatively) valenced consumer responses such as attitude toward the brand (negative cognitions) were more (less) favorable toward ethnic minority endorsers than toward ethnic majority endorsers. A negative value indicates that positively (negatively) valenced consumer responses were less (more) favorable toward ethnic minority endorsers than toward ethnic majority endorsers. Alternative statistical measures (e.g., Student's t-tests) were converted into correlation coefficients using established procedures (e.g., Borenstein et al. 2021; Morris and DeShon 2002). We contacted the authors when insufficient statistical information was provided. All effect sizes were corrected for measurement unreliability. If a study did not report reliability statistics or used single-item measures, we applied the average reliability coefficient for the respective construct, aggregated across all studies.
The final database included 106 papers with 155 distinct datasets (studies), providing 1,276 effect sizes. The sample included peer-reviewed journal articles, unpublished theses, and conference proceedings, thereby mitigating the risk of publication bias (see Web Appendix B for an overview). Data were collected across diverse geographic contexts, including countries from all continents except South America—specifically Australia, Belgium, Canada, China, France, Germany, Greece, India, Indonesia, Iran, Israel, Malaysia, Qatar, Singapore, South Africa, South Korea, Spain, Taiwan, Turkey, the United Kingdom, and the United States.
Integration of Effect Sizes
We integrated the effect sizes using a mixed-effects multilevel model to account for the hierarchical structure of the data and for multiple effect sizes derived within a single dataset. We estimated the following model:
Meta-Regression
To examine potential moderating effects, we conducted multilevel meta-regression analyses. We aggregated all attitude variables (attitude toward the ad, attitude toward the brand, attitude toward the company, attitude toward the product, attitude toward the spokesperson), and behavioral response into a “persuasion” variable, following prior research (e.g., Eisend and Tarrahi 2022). In line with prior meta-analyses (e.g., Weingarten and Goodman 2021), we controlled for differences between dependent variables and included dummy variables for each attitude variable, using “behavioral response” as the reference category. The estimated model is expressed as follows.
Model robustness was assessed through various diagnostic checks. First, the distribution of effect sizes conformed to the assumption of normality (see Web Appendix C). Second, we examined multicollinearity, which is a common issue in cross-national research and meta-analyses (Steensma et al. 2000). The highest variance inflation factor (VIF) was observed for power distance (6.29 in the persuasion model, 6.23 in the attitude model). Removing this variable reduced the maximum VIF to 3.14 and 3.11, respectively, though results remained stable. Thus, we retained power distance in the model, as specified previously. Correlations among moderators are reported in Web Appendix D.
To assess within-country heterogeneity, we conducted a subsample analysis using only U.S.-based studies. Building on Equation 2, we excluded all country-level moderators and added three ethnic minority group indicators (Black, Hispanic, Asian) to capture endorser ethnic diversity. Then, we reran the same model without the dummy variables but added either education or income. Additionally, interaction terms between each diversity variable and consumer ethnicity (minority) were added individually. The model specification for these analyses is as follows, where DIVERSITY refers to either a specific ethnic group (i.e., Black, Hispanic, Asian) or socioeconomic characteristic (i.e., education, income):
Results
Effect Size Integration
Table 4 reports the results of the effect size integration. When we aggregated the data across both ethnic minority and majority consumers, five consumer response variables yielded statistically significant average effect sizes. Specifically, exposure to ethnic diversity significantly increased (1) attitude toward the brand, (2) attitude toward the company, (3) attitude toward the product, (4) memory, and (5) model identification. However, when the results were disaggregated by consumer group, distinct response patterns emerged. Among ethnic minority consumers, exposure to ethnic diversity led to significantly more favorable outcomes for attitude toward the ad, attitude toward the brand, attitude toward the company, behavioral responses, positive cognitive responses, model identification, and perceived relevance, and it reduced negative cognitive responses. In contrast, ethnic majority consumers demonstrated significantly increased negative cognitive responses and decreased positive cognitive responses in response to ethnic diversity. Differences between the two consumer groups can be inferred from the nonoverlapping confidence intervals. In addition, we directly tested for differences between ethnic minority and majority consumers (see Web Appendix E). These analyses confirm that minority consumers, compared with their majority counterparts, exhibit more positive attitudes toward the advertisement, company, and spokesperson; stronger behavioral responses; lower levels of negative cognitive responses; higher levels of positive cognitive responses; reduced intolerance; greater model identification; and stronger ad relevance in response to ethnic diversity.
Effect Size Integration.
The multilevel Q-statistic refers to heterogeneity between datasets.
The fail-safe N is the number of additional statistically nonsignificant correlations needed to render the results nonsignificant.
The regression test assessed funnel plot asymmetry, and a statistically significant result indicated asymmetry.
Tests for homogeneity revealed significant heterogeneity for all consumer response variables except self-image evaluation. The removal of outliers had minimal impact and resulted in only slight changes to the effect sizes for attitude toward the brand and self-image evaluation.
To assess publication bias, we calculated fail-safe N values for all statistically significant average effect sizes. The results suggest that the findings are robust. However, several funnel plot asymmetry tests yielded statistically significant results, with most showing a negative association. This indicates that effect sizes tend to be larger in studies with smaller sample sizes and suggests the presence of small-study effects, which may stem from publication bias. Nevertheless, small-study effects may also arise from differences in study design between small and large studies or from random variation. Moreover, the power of regression-based tests for funnel plot asymmetry is limited, especially when the number of effect sizes is small—as is the case for certain outcome variables in Table 4 (Sterne, Gavaghan, and Egger 2000). Finally, the sample size moderator in the meta-regression model was not statistically significant, providing no further support for the presence of small-study effects.
Meta-Regression
Tables 5 and 6 presents the results of the persuasion meta-regression models (Equation 2) that include statistically significant interaction effects (p < .05). Comprehensive results for all meta-regression models—covering both persuasion and attitudes, and including all tested interaction effects—are available in Web Appendix F. Each model explained a significant proportion of heterogeneity in the effect sizes, as confirmed by model test statistics. The residual heterogeneity remained statistically significant, and variance component estimates indicated that within-dataset variance exceeded between-dataset variance.
Influence of Moderator Variables on the Effects of Ethnic Diversity in Advertising on Persuasion: Meta-Regression Results—Part 1.
We predicted the correlations by setting each dummy variable to 0 and 1. For continuous moderators, we predicted correlations at one standard deviation below the mean, at the mean, and at one standard deviation above the mean. All other variables in the model were held at their sample mean values.
The number of 891 effect sizes is smaller than the 1,019 persuasion-related effect sizes in Table 4 due to missing values in some country moderators.
Influence of Moderator Variables on the Effects of Ethnic Diversity in Advertising on Persuasion: Meta-Regression Results—Part 2.
The number of 891 effect sizes is smaller than the 1,019 persuasion-related effect sizes in Table 4 due to missing values in some country moderators.
Across the various models, several main effects emerged consistently. Rule of law, year of data collection, minority/majority consumer, celebrity endorser status, and attitudes toward the spokesperson all significantly predicted effect size variations. In line with findings from the effect size integration, these effects were generally stronger among ethnic minority consumers. Specifically, the rule of law was associated with attenuated advertising effects, while effect sizes increased over time. Advertising featuring noncelebrity endorsers yielded stronger effects, as did advertisements evaluated on the basis of attitudes toward the spokesperson and, to a lesser extent, attitudes toward the company.
Consistent interaction effects were observed across both the persuasion and attitude models, particularly between several moderator variables and minority/majority consumer: net migration, rule of law, year, participant age, participant gender, endorser gender, and brand type. As illustrated in Figure 2, the effects of ethnic diversity in advertising increase with net migration for minority consumers but decrease for majority consumers, consistent with predictions from intergroup threat theory. Rule of law moderates these effects by reducing the divergence in response between majority and minority consumers. The increasing impact of ethnic diversity over time is evident only among majority consumers. Additional interaction patterns reveal that participant age negatively moderates effects among minority consumers but not among majority consumers. Further, a higher proportion of male consumers strengthens the effects for minority groups but weakens them for majority groups. The gender of the ethnic minority endorser also matters: Female endorsers tend to be more effective among majority consumers, whereas male endorsers have a stronger influence on minority consumers. Finally, the differential effect of ethnic diversity is stronger for real brands than for fictitious brands or products.
Interaction Plots: Meta-Regression Results.
Robustness checks, reported in Web Appendix G, support the stability of these findings. First, we reestimated the models after removing power distance, which had the highest VIF. Second, we recoded the minority/majority variable to distinguish between majority-only versus minority or mixed consumer samples. Third, we substituted rule of law with democracy (due to high collinearity between the two), and we introduced GDP growth and GDP per capita as alternative economic indicators, removing multicollinear variables as necessary. 7 We tested alternative imputation methods—median of nearby values and linear interpolation—for variables where mean imputation was applied (participant age, participant gender, and ethnic fractionalization). Then, we examined an interaction between net migration and ethnic fractionalization. Finally, we estimated a noncontextualized model excluding all country-level variables. Across all robustness tests, the results remained largely consistent with the main findings presented in Tables 5 and 6, affirming the validity of the meta-regression results.
Over 60% of the datasets in our sample were collected in the United States, a proportion that aligns with the country distribution reported in previous meta-analyses (e.g., Geyskens, Steenkamp, and Kumar 1998). Table 7 presents the results of the U.S.-focused regression model (Equation 3), with corresponding results for the attitude models reported in Web Appendix H.
Influence of Moderator Variables on the Effects of Ethnic Diversity in Advertising on Persuasion: U.S. Data.
The results show that persuasive effects do not systematically vary by specific ethnic group categorizations within the United States, for both minority and majority consumers. However, significant interaction effects emerged between education and minority/majority consumer, as well as between income and minority/majority consumer. As illustrated in Figure 3, higher education and income levels of ethnic minority endorsers lead to stronger persuasive effects among majority consumers, consistent with theoretical expectations regarding perceived power and status. The same characteristics diminish persuasive effects among minority consumers. Other significant main effects—specifically for celebrity status, minority/majority consumer, and attitudes toward the spokesperson—were generally consistent with the broader findings reported in Tables 5 and 6, further supporting the robustness of our model.
Interaction Plots: U.S. Data.
Discussion
This meta-analysis examined the effects of ethnic diversity in advertising, revealing small effects. The effects are influenced by multiple moderating variables related to country context, ethnic diversity characteristics, and factors linked to time, product, media, consumer, context, and method and often vary across consumer groups. The findings offer valuable implications for both academic research and managerial practice, which are summarized in Table 8.
Summary of Key Findings, Research Implications, and Managerial Implications.
Research Implications
Although the average effects of ethnic diversity in advertising are relatively modest, thereby questioning the optimistic estimates of business benefits due to diversity in advertising as indicated by recent industry reports (Unstereotype Alliance 2024), they are more pronounced and positive for ethnic minority consumers compared with majority consumers. This finding clarifies mixed results found in prior research and highlights the necessity of disaggregating effects by consumer groups. The persuasive effects explored here may extend to other important outcomes—such as trust, engagement, or brand attachment—that warrant further empirical investigation, as they remain understudied in the context of ethnic diversity in advertising.
Prior literature emphasizes the need to consider societal as well as commercial outcomes of diversity in advertising, given that consumers often respond favorably to socially responsible brands (Eisend 2019). While this meta-analysis primarily focused on commercial outcomes—reflecting the emphasis in the primary studies—it also included two noncommercial outcomes: self-evaluation and tolerance. Although we did not find significant effects for these, this does not undermine the potential societal relevance of ethnic diversity in advertising. Future studies should explore broader societal outcomes, such as intergroup attitudes, ethnic identity affirmation, perceived social norms, and civic engagement.
Importantly, this meta-analysis incorporates multiple country-level moderators simultaneously, avoiding the limitations of earlier work that adopted simplistic cultural frameworks (e.g., Gao, Xu, and Kim 2013; Lwin and Wee 2000). Our results support theories referring to both contact/exposure and institutions. Perceived intergroup threat explains divergent consumer responses: As threat increases, majority group consumers respond less favorably, while minority group consumers respond more favorably to ethnic diversity in advertising. Strong institutions, characterized by effective rule of law, reduce both the differential impact across consumer groups and the overall effect of ethnic diversity, likely due to enhanced equality and lower perceived intergroup differences. The findings highlight the need to consider sociopolitical and economic environments in both interpreting prior studies and designing future research.
Within-country moderator analyses revealed that differences in consumer responses to specific ethnic minority groups are best explained by power and status differences (i.e., education and income difference of the endorser's ethnic group) rather than mere group categorizations (i.e., whether the endorser is Black, Hispanic, or Asian American). Higher status and power of minority endorsers enhances effects for majority consumers but reduces them for minority consumers—likely because powerful minority endorsers are perceived by their own group as atypical exemplars that do not fit existing schemas and status expectations, leading to backlash and reducing identification and ingroup favoritism (Rudman et al. 2012). This is consistent with findings from recent meta-analytic work (Lenk, Hartmann, and Sattler 2024), which show stronger ingroup preference among U.S. consumers for same-race endorsers, and a temporal trend showing increased acceptance of minority endorsers by White consumers over time. This parallels our finding that effects of ethnic diversity (related to multiple ethnic minority groups) on majority consumers have increased in recent years.
Moreover, social categorization may explain why ethnic minority celebrities tend to have weaker persuasive effects: They are often perceived as atypical and not representative of their group, thereby weakening identification and prototype-based effects. A lack of prototype congruence is particularly likely in cultural contexts where ethnic minorities have historically lower status or face structural exclusion, such as in France or Germany: Minority celebrities may be perceived as unrepresentative exceptions rather than group prototypes, activating contrast effects or stereotype threat rather than identification. However, prototype congruence and identification vary depending on the social status of ethnic minorities. For instance, in countries with highly selective immigration policies such as Canada or Australia, high-status migration is more common, which could increase the identification of ethnic minority consumers with same-ethnicity celebrities. Our results further show that younger and male minority consumers respond more favorably to diversity in advertising, possibly due to a stronger need for identity affirmation. Additionally, male minority endorsers are more effective with minority consumers, while female endorsers work better with majority consumers. This may be attributed to stereotypical gender roles. Male endorsers may convey strength and competence (appealing to status needs), while female endorsers may be perceived as warm and nurturing, qualities broadly favored by general audiences (Fiske, Cuddy, and Glick 2007). Finally, we found that majority consumers respond less favorably to ethnic diversity when the advertised brand is real rather than fictitious, potentially due to stronger expectations or brand schemas based on traditional advertising portrayals for real brands and products. Real brands can activate preexisting brand knowledge and stereotypical expectations (e.g., German or U.S. brands that are associated with tradition and national identity), which can amplify the diagnosticity of diversity cues and make them more likely to be scrutinized for authenticity or consistency. In contrast, fictitious brands that lack prior schemas allow for more flexible interpretation of diversity cues, potentially reducing dissonance or resistance. In sum, these moderator effects provide opportunities for more thorough investigation into the proposed underlying mechanisms.
Managerial Implications
From a practical standpoint, the findings challenge the efficacy of using ethnic diversity in advertising aimed primarily at majority consumers. Despite concerns among advertisers about alienating majority audiences (Schlegelmilch, Khan, and Hair 2016), our results suggest that featuring ethnic minority endorsers does not undermine advertising effectiveness—particularly when the target audience includes minority consumers. While the average effects are modest, they are notably more substantial among minority audiences, making ethnic diversity advertising especially viable in markets with significant minority populations. Given increasing ethnic diversity in many countries—for example, Hispanic/Latino individuals comprise 19.1% of the U.S. population (USA Facts 2024)—targeted ethnic diversity advertising is both relevant and necessary. It seems particularly effective for products specifically relevant to minority consumers (e.g., food, films) but may also be adapted to mainstream products.
Because advertising strategies are typically implemented on a national level, advertisers require country-specific insights, especially when contextual factors (e.g., migration rates, institutional strength) exert countervailing influences. To this end, we conducted an additional country-level prediction analysis (Web Appendix I). The results suggest that ethnic diversity advertising is least effective in countries with both strong institutions and high migration rates (e.g., Sweden, Singapore, Netherlands), but more effective in countries with stricter migration controls (e.g., Austria) or weaker regulatory frameworks (e.g., Mexico). Although national categorizations are imperfect, patterns in immigration histories and public sentiment regarding immigration regulation can help explain the results. Advertisers should carefully interpret them, as local demographic and cultural factors may moderate advertising effectiveness. For instance, majority consumers in countries with low migration rates might respond more favorably to ethnic diversity, but the group of ethnic minority consumers itself might be too small to be targeted efficiently.
Endorser selection is also critical. High-status ethnic minority endorsers resonate more with majority audiences but less with their own group. These dynamics may change over time and vary by country, depending on the composition of ethnic groups and their relative power. For example, the U.S. has a high ethnic fractionalization index (Drazanova 2020), indicating frequent contact with diverse ethnic groups and making status differences more salient. Similar high fractionalization exists in countries such as Belgium, Canada, Indonesia, Spain, or the United Arab Emirates. The findings may not apply in more homogeneous societies with low migration rates and low ethnic fractionalization (e.g., Japan), where such distinctions are less visible or relevant. Thus, the findings should be generalized with caution across different countries, taking into account their unique demographic and migration characteristics.
Additional moderators suggest avoiding celebrity endorsers in ethnic diversity advertising. Focusing on younger and male minority audiences can increase the effectiveness of ethnic diversity advertising. Male minority endorsers are most effective for targeting minority consumers, while female endorsers appeal more to majority consumers. To mitigate backlash among majority audiences, brands may benefit from using both minority and majority endorsers in initial campaigns to foster familiarity before transitioning to minority-only representation.
Limitations and Future Research
This study has several limitations. Although the meta-analysis includes data from countries across multiple continents, South America is not represented, and the dataset skews toward Western, industrialized nations. This is a common limitation in the field and highlights the need for broader geographic replication, particularly for a topic as context-sensitive as ethnic diversity. Future studies should collect data from underrepresented regions and test more granular country-level moderators.
Moreover, current implementations of ethnic diversity in advertising often rely on visible markers such as skin color. This oversimplifies the construct and risks obscuring more nuanced identity cues. Future research should explore alternative indicators of ethnicity, such as names, symbols, or short narrative descriptions that signal ethnic affiliation.
Our analysis also does not account for within-country regional variation. Ethnic diversity tends to be higher in urban areas than in rural regions, likely influencing consumer responses. Unfortunately, most primary studies do not report regional data, limiting the ability to make subnational recommendations. Future research should address ethnic diversity advertising effectiveness below the national level to inform more targeted marketing strategies.
Supplemental Material
sj-pdf-1-jmx-10.1177_00222429251408344 - Supplemental material for Ethnic Diversity in Advertising: A Meta-Analysis
Supplemental material, sj-pdf-1-jmx-10.1177_00222429251408344 for Ethnic Diversity in Advertising: A Meta-Analysis by Martin Eisend, Anna Rößner and Erik Hermann in Journal of Marketing
Footnotes
Coeditor
Jan-Benedict E.M. Steenkamp
Associate Editor
Marnik Dekimpe
Ethical Considerations
There are no human participants in this study and informed consent is not required.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research received financial support in the form of a research grant from the German Research Foundation (Deutsche Forschungsgemeinschaft), Germany (grant number EI 508/21-1).
Data Availability
The data that support the findings of this article are publicly available in the JM Dataverse.
Notes
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
