Backchannelmedia

Journal of Advertising

By Brumbaugh, Anne M; Grier, Sonya A

ABSTRACT:

Results of a "failed" experiment in which we compare and contrast responses of Asian, black, Hispanic, Indian (Asian Indian), white, and multiethnic American participants show systematic differences in ad processing among different groups toward the same advertisement. Rather than describe this research via implausible a priori hypotheses, we explain the results of the study post hoc by generating research questions that draw from extant literature on distinctiveness, race, ethnic identity, and culture. We discuss these research questions in the context of pluralistic, multigroup (greater than two) experimental designs, detailing methodological and theoretical considerations, in hopes of influencing the sophistication and utility of future advertising research.

Peruse a business research database in search of ethnicity and advertising research and you will find studies that pit just about any two groups against each other in an attempt to identify, produce, and explain differences between the two groups. Studies comparing black and white Americans (see Bush, Smith, and Martin 1999 for a review), Asians and Caucasians (Rose, Bush, and Kahle 1998), Hispanic and Anglo Americans (Deshpand and Stayman 1994), new Hispanic immigrants and long-acculturated Hispanics (Ueltschy and Krampf 1997), and indigenous and expatriate populations of the same ethnic group (Mehta and Belk 1991) have all yielded insights into the advertising processes and outcomes that are affected by ethnic group membership (defined and used herein as social characterization and self-identity based on shared race, nationality, ancestry, culture, language, and/or religion). In contrast to the vast number of papers comparing two groups, relatively few compare three or more distinct groups (for recent exceptions, see Kim and Kang 2001; Seitz 1998).

Nearly every aspect of research design in general (participant availability, sample size, administrative ease, methodological clarity) makes it clear why a predominance of research on ethnicity and advertising emphasizes two-group comparisons. In nearly all regions of the world, one ethnic group predominates numerically, socially, or both. It is this dominant group about which most is known in terms of advertising effects and to whom most marketing activity is targeted. This is also the group against which others are frequently compared in adverrising and other research. In the United States, whites are and will remain this group, despite dropping to less than 50% of the American population in 40 years or so: No other single ethnic group will eclipse non-Hispanic white Americans in number and buying power for the foreseeable future (Bureau of the Census 2001; Humphreys 2004). Thus, research that includes white Americans remains relevant, even in a multiethnic context. Comparing consumers in this segment, about which we know much (a vast majority of psychological research has been done with white participant populations; see Sue 1999), with blacks, Hispanics, or Asians is a pragmatic choice, as the latter two segments are growing fastest and the former two are largest in absolute terms. With two-group studies (such as those that compare whites with another ethnic group), sample sizes are manageable, and differences, should they obtain, are relatively easy to interpret.

However, the focus on two-group studies, particularly studies that compare the dominant group with another group, may limit the types of generalizations researchers can make about how ethnicity influences advertising responses and persuasion. Although this dominant group provides a convenient and frequently appropriate contextual reference against which to compare other ethnic groups (particularly for advertising research), we often do not consider how this group differs from the comparison group on bases other than ethnicity. Thus, although hypothesized differences obtain between the two groups, the underlying mechanism causing the differences (and perhaps misattributed to ethnicity) may remain a mystery.

In this paper, we explore the results of a "failed" two-group quasi-experiment in which we initially compare and contrast advertising responses of nondistinctive white and ethnically distinctive participants toward ads designed to target them on the basis of their (non)distinctiveness-the very experimental design we decry above. Initial predictions regarding this highly pragmatic 2x2 design were not borne out. However, when we explored the results for each ethnic group represented in our distinctive participant population (Asian, black, Hispanic, Indian [Asian Indian; hereafter referred to as "Indian"], and multiethnic Americans), we discovered some intriguing insights that we report herein. Our exploration of these results takes an unconventional form: Instead of trying to force a priori hypotheses that all will see as contrived, we describe the results of the study and then explain them post hoc. In the remainder of this paper, we detail these results and use them to generate questions that can and should be addressed in future research. These research questions are accompanied by suggestions for experimental designs that exploit, rather than avoid, the use of three or more racial, ethnic, or cultural groups to yield deeper insights into how ethnic group membership and identity influence advertising responses.

BACKGROUND

The initial intent of our study was to explore the impact of distinctiveness on advertising responses (see Grier and Brumbaugh 2004 for a review). Individuals who are distinctive relative to other people in a group (being numerically rare on some important, self-defining dimension) have been shown to be more likely to self- define on the trait that makes them distinctive (McGuire 1984). As a consequence, information about that trait (be it gender, Cota and Dion 1986; ethnicity, McGuire, McGuire, and Winton 1979; or wearing glasses, McGuire and Padawer-Singer 1976) is more likely to be used to process messages targeted on the basis of that trait than if the trait were nondistinctive. For example, ethnically distinctive individuals are more likely to interpret an ethnically targeted ad in terms of ethnicity than ethnically nondistinctive individuals (Deshpand and Stayman 1994), and to prefer ads with similar ethnically distinctive sources more than nondistinctive individuals in response to similar nondistinctive sources (Grier and Brumbaugh 1999). Nondistinctive individuals have been shown to prefer ads they feel target them based on a broader configuration of ad cues, not merely similar sources (Aaker, Brumbaugh, and Grier 2000).

Based on this large body of research in social psychology (McGuire, McGuire, and Winton 1979; McGuire 1984; McGuire and Padawer-Singer 1976) and increasing application in marketing (Forehand and Deshpand 2001; Forehand, Deshpand, and Reed 2002; Grier and Brumbaugh 1999; Grier and Deshpand 2001; Wooten 1995), we designed a study to explore felt similarity (Whittler 1991) and felt targetedness (Aaker, Brumbaugh, and Grier 2000) as mediating processes in distinctive and nondistinctive individuals' preferences for targeted advertising. Specifically, consistent with previous research (Aaker, Brumbaugh, and Grier 2000), we expected that ethnically distinctive individuals would prefer an ad targeting them on the basis of their ethnicity via a process of felt similarity and that ethnically nondistinctive individuals would prefer an ad targeting them on the basis of their ethnicity via a process of felt targetedness. We designed a quasi-experimental study (participant distinctiveness/ethnicity is measured, not manipulated) to explore these theses. Speculating that all distinctive participants (irrespective of the ethnic basis of their distinctiveness) would respond similarly to an ad targeted toward them, we recruited white Americans as our nondistinctive sample and ethnic minority Americans as our distinctive sample. Thus, in our initial analyses, we aggregated Asian, black, Hispanic, Indian, and multiethnic participants into a single distinctive category. This study failed in that the aggregate distinctiveness effects we hypothesized did not materialize. However, further examination of our data in which we disaggregated the effects of each ethnic group proved to be a success in generating insights into how to better investigate cross- cultural responses to targeted advertising.

STUDY

Method

A total of 236 undergraduate students (52% male, all 18 to 25 years old) from a Midwestern university were recruited via a campus electronic mail notice and posted flyers to participate in marketing research for five dollars. All were told that the purpose of the research was to evaluate a prototype of a new magazine for college students called OnCampus. The magazine contained three articles unrelated to the study and two fictitious color advertisements. The first was a filler advertisement for lunch meat containing no source cues and the second was one of two versions of the focal advertisement for spring break travel.

We designed one version of the focal advertisement (the "distinctive ad") to target black, Asian, and \Indian students by including young adult sources from each ethnic group (four total, two of whom were black) in the picture shown in the ad and listing one student club whose mission was to promote cultural understanding pretested to represent each ethnic group shown in the ad: the African American Union, the University Indian Student Association, and the Asian American Alliance. In the execution designed for white students (the "nondistinctive ad"), the ad contained four white young adults in the photo and cited three student groups whose memberships were shown in pretesting to be perceived as predominantly white: the Office of Greek Life, the Film Society, and the Trivia club. Each ad contained two males and two females. The text of the ad read, "For Fall Break. . . . Wouldn't You Rather Be Here? Langley Travel offers many spring break trips, including airfare, cruises, beach rentals and action vacations. Prices start at only $199 for 5 days, 4 nights. Contact the tclubs] and other student organizations for information on this special promotional offer." All other aspects of the advertisement, including the tropical beach photo, background color, and font were identical across conditions. Thus, the study was initially a 2 (ad distinctiveness: distinctive versus nondistinctive) ? 2 (subject distinctiveness: distinctive versus nondistinctive) between- subjects design.

Participants completed the study individually with the first author (a white female) at her office at scheduled appointment times.1 They were assigned randomly to either the distinctive or nondistinctive ad condition and were asked to read the magazine as they normally would. When finished, participants evaluated each advertisement, the editorial content of the magazine, the magazine's layout, and their overall perception of the magazine, consistent with the cover story. Ad measures included five questions regarding how similar participants felt to the sources depicted in the ad (α = .87; Whittler 1991), three questions regarding how targeted they felt by the ad (α = .84; Aaker, Brumbaugh, and Grier 2000), and three questions regarding their overall attitude toward the ad (a = .96; good/bad, favorable/unfavorable, like/ dislike).

In addition, participants completed demographic information that included an ethnic identity checklist for which they were allowed to specify identification with multiple ethnic groups. Students self- identified their ethnic group membership as Asian/Asian American, black/African American, Hispanic/Latino, white/European American, and/or other (with a blank to specify). Sixty-four self-identified uniquely (checking one box only) as Asian, 37 as black, 20 as Hispanic, and 83 as white. Fifteen students specified "other," with country of origin designated as India. We classified 17 students who specified two or more groups as "multiethnic" students. Responses from one student who identified as Native American were omitted.

Next, to provide a manipulation check of the target manipulations, participants completed an identical checklist to identify who they thought was the target for the ad. All participants correctly identified the intended targets for both advertisements as evaluated through their responses to this target- market checklist. To ensure that participants used the similarity, targetedness, and attitude toward advertising scales similarly, we evaluated between-group differences in these three items for the filler ad; there were no differences between participant groups on these measures.

Finally, we debriefed participants as to the true nature of the study, paid them, and dismissed them.

Initial Results: Failure

Results of the analyses were disappointing for our original theses that ethnically distinctive individuals would prefer an ad targeting them on the basis of their ethnicity via a process of felt similarity and that ethnically nondistinctive individuals would prefer an ad targeting them on the basis of their ethnicity via a process of felt targetedness. If distinctiveness as a construct were immutable across social groups (here, all ethnic minority groups of which participants were members), we would expect the interaction between participant distinctiveness and ad distinctiveness to be significant, but it was not. When we included dummy variables for each ethnic group represented in our sample nested under the distinctiveness condition, we found that they did achieve significance when interacted with ad distinctiveness and that the distinctiveness dummy variable did not contribute to the variance explained by the model. Our theoretical focus on distinctiveness had obscured interesting and insightful ethnic group differences.

Specifically, our results yielded little insight on distinctiveness as the underlying construct and more on the relationships between different ethnic groups and the representation of ethnicity in our ad stimuli. For example, the ethnic cues we selected were atypical of other ethnically targeted advertising in that our combination of different (but unambiguous) ethnic cues in a single ad is relatively infrequent. Furthermore, Asian and Indian populations are rarely targeted directly by advertising in the United States (Mandese 2003), and their inclusion in the study was likely to yield novel results. Therefore, as we revisited our data, we approached the analyses with the goal of understanding reactions to the ads by each individual ethnic group.

In doing so, we renamed our ad and participant factors to be more indicative of their underlying qualities and to aid in the clarity of the discussion of our results. The ad formerly known as the "nondistinctive ad" was now more appropriately called the "white ad," and the ad formerly known as the "distinctive ad" was now called the "multiethnic ad." Rather than aggregating across participant ethnicity to two levels based on level of distinctiveness (distinctive versus nondistinctive), we now looked at each of the six ethnic classifications individually. Thus, the design was now a 2 (ad ethnicity: white versus multiethnic) x 6 (participant ethnicity: Asian, black, Hispanic, Indian, multiethnic, or white) between-subjects design.

A Priori Caveats

It should be noted that in disaggregating our participants into self-identified ethnic groups in this revised quasi-experiment, several shortcomings of our study become salient. First, for some ethnic groups (notably Indians and multiethnic Americans), cell sizes become smaller than convention would suggest is appropriate. As a consequence, we used Levene's Test of Homogeneity of Variance to ensure that we could pool observations across ethnic groups. Results show that the presumption of homogeneity of variance was not rejected for any of our three dependent variables: For similarity, F(11, 224) = 1.00, p > .44; for targetedness, F(11, 224) = 1.02, p > .43; for attitude toward the ad, F(11, 224) = .84, p > .60. Furthermore, we use PROC GLM in SAS and Type III sums of squares to control for unequal cell sizes (SAS 2004; Tybout et al. 2001) and report these results throughout.

Second, although previous research conducted with student samples has yielded important insights into the impact of ethnicity on advertising and other phenomena (Brett, Lee, and Yang 2004; Bush, Smith, and Martin 1999; Forehand and Deshpand 2001), student samples should always be used with caution (e.g., Liefeld 2003). This is particularly true when studying ethnicity, as both the student's country of origin (international or not) and contextual factors (ethnic orientation of the study situation) may confound or otherwise affect the influence of felt ethnic identity on advertising responses. International students are not typical immigrants, nor are they simply residents of their home country, and these distinctions should be kept in mind when interpreting our results. For example, in this study, we did not distinguish between Asian versus Asian American students if participants selected "Asian/ Asian American"; we did not distinguish between Hispanic American and international students from Latin America if they selected "Hispanic/Latino"; and for international students from India, we did not use a "hyphen-American" analog. Prior research has shown that Asian American college students differ from Asian students attending college in the United States in terms of acculturation in subtle, but potentially important, ways (Rhee, Uleman, and Lee 1996; Tsai, Ying, and Lee 2000). More generally, within-group heterogeneity is an important consideration when interpreting any study results and is an issue we consider explicitly in our questions for future research.

Finally, the context surrounding the administration of this study (and, in fact, all studies related to race, ethnicity, and culture) may subtly influence our results. The relevance of ethnicity and its influence on study results and interpretation frequently depend on the specific situation or phenomenon of interest, the ethnic context associated with administration of the study, and/or the ethnicity of the experimenter. For example, prior research shows that the racial or ethnic identity of the experimenter does not have a straightforward influence on responses but, rather, interacts with other influences such as language, gender, and type of question (Annis and Corenblum 1986; Webster 1996). Furthermore, participants' ethnic identity as similar or dissimilar to the ethnic nature of the study context influences participants' responses (Deshpand and Stayman 1994; Stayman and Deshpand 1989). Therefore, as suggested by prior literature, we included consideration of these contextual factors to interpret the results of our study. We also followed advice from previous research on context and experimenter effects (e.g., Venkatesan 1967) to include the same instructions, the same conditions, and the same experimenter across all administ\rations of the study.

Revised Results: Success

Due to this exploratory nature of our investigation, we report effects significant top < . 10 (Asham 1988). When differences between cells are consistent with prior research (i.e., when white nondistinctive [ethnic minority distinctive] participants prefer, feel more targeted by, and feel more similar to the characters in the white [multiethnic] ad), we used one-tailed t tests. Degrees of freedom for ANOVA (analysis of variance) models are (11, 235) and (1, 235) for contrasts. In addition, a separate regression was run for each ethnic group to assess the relative impact of a match between participant and ad (coded as 1 for white participants who viewed the white ad and for ethnic minority participants who viewed the multiethnic ad or 0 otherwise), targetedness, the interaction between targetedness and participant/ad match, similarity, and the interaction between similarity and participant/ad match on attitude toward the ad. Results are shown in Table 1.

We expected nondistinctive white participants to prefer the white ad over the multiethnic ad and for targetedness , to drive this effect. Indeed, white participants felt more targeted by the white ad (M = 5.30) than the multiethnic ad (M = 3.16, F = 40.83, p < .01), more similar to the sources in the white ad (M = 3.91) than to the sources in the multiethnic ad (M = 3-47, F = 2.45, p < .06), and more favorably disposed toward the white ad (M = 4.14) than the multiethnic ad (M = 3.08, F = 10.66, p < .01). Targetedness was related to attitude toward the ad among white subjects (β = .29, p < .03), an effect that was stronger for the white ad (β = .46, p < .06).

Asian, black, and Indian subjects served as distinctive populations targeted by the multiethnic ad. We expected them to prefer the multiethnic ad over the white ad and for felt similarity with the sources depicted in the ad to drive this effect. Asian subjects felt both more targeted by the multiethnic ad (M = 4.89) than the white ad (M = 4.36, F = 1.98, p < .08) and more similar to the characters in the multiethnic ad (M = 4.25) than to the characters in the white ad (M = 3.51, F = 5.26, p < .01), but they did not like the multiethnic ad more than the white ad (p > .20). Targetedness was associated with attitude toward the ad among Asian subjects only for the multiethnic ad (β = .37, p < .08). Consistent with prior research, black subjects felt more similar to the characters in the multiethnic ad (M = 4.15) than to characters in the white ad (M = 3.29, F = 3.92, p < .03) and liked the multiethnic ad more (M = 4.59) than the white ad (M = 3.96, F = 1.65, p < .10). Furthermore, similarity was related to ad attitudes among black subjects only for the multiethnic ad (β = .65, p < .03); black subjects felt no more targeted by the multiethnic ad than the white ad (p > .20). Indian subjects felt more similar to characters in the multiethnic ad (M = 4.05) than to characters in the white ad (M = 2.71, F = 3.95, p < .03) and felt similarity to characters was associated with their ad attitudes for both ads (β = 1.06, p < .03).

Hispanic and multiethnic subjects provide an interesting counterpoint to the (non)distinctive (non)target design instantiated by white, Asian, black, and Indian subjects in the study. Neither stimulus ad targets Hispanics based on ethnicity explicitly. On the one hand, we might expect Hispanics to respond favorably to the general multicultural theme of the multiethnic ad even though they are not included in the sources depicted and the clubs mentioned. On the other hand, they might feel more targeted by the white ad because of shared race (though not ethnicity in terms of the Anglo/Hispanic distinction). Results show that Hispanic subjects generally reacted no differently to the white ad versus the multiethnic ad in terms of felt targetedness, similarity, or attitude toward the ad, marginally preferring the multiethnic ad over the white ad in the regression analysis (β = 2.25, p < .08). Felt similarity to characters in either ad was related to ad attitudes (β = ,98, p < ,01) for Hispanic subjects,

In contrast, multiethnic subjects may have felt targeted by one or both ads depending on their ethnic background, Like Hispanic subjects, they reacted no differently to the white ad versus the multiethnic ad in terms of felt targetedness, similarity, or attitude toward the ad, Similarity was associated with ad attitudes for them for the white ad (β = 1.39, p < .04), but not for the multiethnic ad (p > ,20), Felt targetedness, however, was related to attitudes among multiethnic subjects for the multiethnic ad (β = 1.21, p ,06), but not for the white ad (p < ,20),

Discussion

Table 2 organizes and summarizes the results of our study, Results for black and white subjects are entirely consistent with our expectations and previous research, showing that whites prefer the ad targeted toward them via a process of targetedness, whereas blacks prefer the ad targeted toward them via a process of similarity, It is the results among other ethnic groups that generate deeper insights into how ethnicity, race, and identity may influence responses to targeted advertising (Tharp 2001) and vice versa (Halter 2000).

The other ethnic groups represented in the multiethnic ad-Asians and Jndians-also felt more similar to sources in this ad than to those in the white ad, as did black subjects, This similarity translated into higher attitude toward the ad for Indian subjects, as it did for black subjects but not for Asians. seemingly, despite feeling more similar to sources in the multiethnic ad, Asians processed the ads in a manner closer to how whites processed the ad targeted for them. These results call into question how ethnic distinctiveness influences responses to targeted processing among groups other than black Americans.

Results among the ethnic groups in our subject pool not represented in the multiethnic ad provide insights into the boundary conditions of how ethnic distinctiveness influences ad processing. Results for Hispanics, who were excluded on the bases of both ethnicity and race from the multiethnic ad, were generally null. If they felt similar to sources in either the multiethnic or white ad, it favorably influenced ad attitudes; targetedness did not have such an influence, however. For multiethnic subjects, who generally identified as white plus a minority ethnic group, neither ad influenced similarity, targetedness, or attitude toward the ad more than the other. Their ad attitudes were favorably influenced by targetedness in response to the multiethnic ad and similarity in response to the white ad, however, contrary to research among whites and blacks.

QUESTIONS FOR FUTURE RESEARCH

Our results raise more questions than they answer. These are questions that have received little or no attention, however, because they are obscured when researchers consider results for only two groups, either by design or because of reviewer suggestion. In this section, we identify plausible explanations for our results, accompanied by suggestions for future research.

The Role of Status in Ad Response

Asian and white participants responded to their respective targeted ads somewhat similarly, with felt targetedness (versus similarity with sources depicted) driving ad attitudes. One plausible explanation for this result may be related to the status of these groups in American society. Ethnicity and majority/ minority representation are nearly always confounded with status, and research suggests that the status of viewers affects how they respond to advertising (Grier and Brumbaugh forthcoming; Grier and Deshpand 2001). Furthermore, research shows that the status of characters depicted in an ad influences how viewers respond to that ad (Grier and Brumbaugh 1999! Grier and Deshpand 2001). Higher status spokespersons may be viewed more credibly (Eagly, Wood, and Chaiken 1978), and advertising featuring these spokespersons may be more effective, In addition, consumption behavior is driven and defined by the highest status group in a society, which serves as a default model or norm for consumers (Grier and McGill 2000).

In the United States, where we conducted our study, whites and Asians are perceived to be of higher status than blacks and Hispanics (Grier and Brumbaugh 1999). As a sought-after target market because of relatively high education and disposable income (Bureau of the Census 2001; Taylor and Stern 1997), Asians may behave more like whites (a similarly attractive target market) in response to targeted advertising. Most ethnic advertising research, it might be argued, compares groups differing not only in ethnicity, but also in status. Measuring, controlling for, or otherwise considering the status of three or more different groups to rule out status as an alternative explanation for ethnicity's influence on results is imperative. Study contexts that offer a plurality of ethnic groups that differ in both ethnicity and status (e.g., Asian, black, Coloured, and white populations in South Africa; multiracial classifications in Brazil) would be helpful in this regard. Experimental designs that cross alternative manipulations or manifestations of status (e.g., occupation) with ethnicity would enable researchers to parcel out the differential effects of each. Utilizing individual differences in perceptions of groups' status as a moderator of the effect of status and ethnicity on ad outcomes would also yield great insight (Grier and Brumbaugh forthcoming; Grier and Deshpand 2001). We propose that researchers consider the following:

RQ1: Status is frequently confounded with ethnicity, but likely has a different effect on identity and self. How does status (as distinct from ethnicity, race, and numeric majority I minority status) affect the influence of ethnicity on ad processing?

Social Cat\egory (Dis)Aggregation

A second possible explanation for our results among Asians and whites may be that targeting on the basis of Asian or white ethnicity has little meaning for members of these segments. Target marketing is a balancing act in which marketers try to pick the optimal point between disaggregating consumers into segments that behave similarly in response to a given marketing activity, and aggregating consumers into segments large enough to be cost- effective to target. Some superordinate categories (e.g., black) may be more diagnostic compared with subordinate categories within them (e.g., Haitian American) because the superordinate category has social, historical, or cultural significance, and influences how members of the category think and behave. In contrast, some subordinate categories (e.g., Korean American, Chinese American) may be more diagnostic than superordinate categories that may have little meaning to members of the groups contained therein (e.g., Asian American). For Asians in our study, "Asian" may represent a superordinate category that subsumes and obfuscates country culture of origin and/or differences in number of years spent in the United States. Group membership may have meaning for individuals in this category only at the subordinate level of country.

Some categories (e.g., white) may simply be irrelevant to self- definition and identification, and thus consumption. Although "whiteness" has received increasing theoretical attention in recent years (e.g., Nakayama and Martin 1999), its meaning for the average member of the category and its role in informing advertising processes may have little relevance as a superordinate category (cf. McGuire, McGuire, and Winton 1979). In using whiteness as a norm against which to compare other groups in a myriad of studies, researchers may have created a consumption category that really does not exist in the minds of most white consumers. For whites in our study, "white" may represent an irrelevant category in terms of self- identity.

Thus, in using whiteness and Asianness as the implicit bases for targeting in our study, we may have picked categories that have no meaning for these populations. Further research into the most relevant, salient, and highly instantiated cultural categories into which one may categorize oneself or others would yield theoretical insight into how consumer identity influences consumption and persuasion, as well as practical insight into how best to target ethnically diverse consumers. For example, researchers could evaluate how different levels of subjects' identification within a hierarchy of cultural categories affect processing of targeted ads (e.g., American, Asian American, Chinese American; American, white versus European American, Italian American; American, black versus African American, Haitian American) to assess which is most diagnostic under different conditions. Manipulations of felt ethnicity at these different tiers of possible identification compared to a control condition in which felt ethnicity is not made salient might suggest which tier in the hierarchy is most influential. Thus, we ask scholars to investigate the following:

RQ2: Categorizing individuals as members of a specific cultural, ethnic, or racial group for the purposes of target marketing is based on convenience, historical practice, and other criteria not necessarily related to individuals' identification with the category to which marketers assign them. What is the meaning of superordinate (e.g., Asian, black) and subordinate (e.g., Chinese American, Caribbean American) cultural, ethnic, or racial categories frequently used in target marketing and what are individuals' reactions to ads targeting them on the basis of this categorization?

Visual Markers of Ethnic Group Membership

Our results show that black and Indian subjects in our sample responded similarly to the multiethnic ad, generally preferring it over the white ad and causing members of both groups to feel similar to the sources depicted, which, in turn, was related to ad attitudes. Why should these two groups respond similarly to the multiethnic ad, whereas Asians did not? Perhaps the visually obvious answer is the correct one. Because of their (relatively) darker skin, members of these two groups are perhaps more visually salient than members of other ethnic groups in this study (Eberhardt et al. 2004). Results raise questions about the relation between one's own ethnic identity and the spontaneous categorization of visually distinctive members of ethnic groups.

For members of an ethnic group viewing an ad targeted on the basis of their ethnicity, ethnic cues (including the source depicted) activate ethnicity-specific knowledge that only they possess that aids in interpreting the ad and influences how (un)favorable their attitude toward the ad (and ultimately, the product) will be (Brumbaugh 2002). Most certainly, the visual salience, novelty, or subtlety of ethnic cues (again, including the source depicted) will influence if and how this cultural knowledge is activated and deployed. For consumers viewing an ad intended for another ethnic group, ethnic cues either convey no meaning and are, for all intents and purposes, disregarded (Grier and Brumbaugh 1999), or they activate information these consumers have about the ethnic group. Indeed, this information may be stereotypical and lead to negative marketing consequences (Crockett, Grier, and Williams 2003). How are these processes different when the ethnic group depicted is highly visually distinctive (very dark-skinned with Afrocentric features) versus visually ambiguous (olive-skinned with brown hair)? When people in the target market decode the in-group target quickly because of greater visual salience, do they rely more on social group information to process the ad (thus inducing similarity effects) than if they don't decode the in-group target quickly (thus inducing targetedness effects)? If people outside the target market decode the out-group target quickly, will they be more likely to stereotype the ad source than if they had to rely on other nonsource cues?

It may be that the visual salience of physical similarity that drives the similarity effect for these participants is quite distinct from ethnic meaning and culture in terms of ad processing (Appiah 2001; Brumbaugh 2002). Research that explicitly distinguishes between the visual markers of ethnicity as advertising cues and the richer meanings associated with ethnicity (either as one's own ethnicity or as stereotype/sociotype information about others) via operationalizations that use more than two groups would allow researchers to investigate more clearly the underlying mechanisms by which targeted advertising and nontarget marketing operate. For example, by manipulating stimuli such that some ads feature ethnic sources with highly visually salient physical cues and some feature ethnic sources with less visually salient physical cues for multiple ethnic groups (that is, an experimental design that crosses ethnic group membership in several groups with level of visual salience of ethnic source cues in an ad), we might be able to discern whether and to what extent the visual salience of ethnic cues influences the speed with which and the degree to which people use such cues to make "me/not me" evaluations. Researchers might consider the following question in their future research:

RQ3: Different ethnic groups may differ in their physical appearance. How does the visual salience of physical markers of ethnic group membership influence ad processing (similarity versus targetedness; heuristic versus systematic) and the potential effectiveness of targeted advertising?

Ethnicity? Race? All of the Above? None of the Above?

Definitional issues regarding race and ethnicity have not been resolved (see Glazer 2002 for a thought-provoking view). Researchers, including the authors, frequently obfuscate this distinction, sometimes appropriately (as when race is the sole driver of ethnicity as in black/white comparative studies) and sometimes inappropriately (as when Hispanics, who may be black or white in terms of race, are compared with white and/or black others).

Though the term "race" was previously used for distinct biological designations between peoples with different phenotype markers (Keita and Kittles 1997), advances in biological research show that within-group variations in these markers are greater than between-group variations. Recent operationalizations and definitions emphasize the social aspects of race over biological aspects (Helms and Talleyrand 1997; Yee et al. 1993). Although some argue that race itself is no longer a relevant construct-that it should be replaced by ethnicity alone (e.g., Yee 1983)-this distinction remains important in light of the rise in Hispanic and biracial populations in the United States, whose members may identify as black, white, both, or neither (Bureau of the Census 2001).

Studies that compare black and white Hispanics with white Anglo and black African-American populations, for example, would enable researchers to disentangle race and ethnicity (to the extent that they are separate constructs) and their impact on target-marketing efforts. Alternatively, crossing race (the black/white distinction) with country (United Kingdom/United States, for example) would similarly separate race from local socioculturel forces. We encourage researchers to address this issue in their research:

RQ4: Marketing-related definitions of race and ethnicity are gray areas that frequently overlap. What is the relation between race and ethnicity in terms of consumer identity and the ability of targeted advertising to induce self-referencing, internalization of the ad message, and other favorable targetmarketing effects?

Ads for Everyone, Ads for None

Multiethnic advertising that features multiple ethnic sourcesand cues aimed at courting several distinct and unrelated target markets is on the rise as marketers recognize that mainstream media are watched by an increasingly diverse audience (Reid 2003). As such, increasing advertising includes members of different ethnic minorities depicted together along with white actors and models (Taylor and Stern 1997; Wynter 1996). However, scant research exists on the effectiveness of such advertisements and the processes in which viewers of different groups engage when they see such advertisements. What little research does exist shows that advertising responses change when an ethnic source is combined with nonethnic sources and that consumers react differently to different combinations of ethnically diverse sources. For example, black actors induce higher ethnic identification among black viewers when depicted alone than when depicted in an identical scene alongside white actors (Brumbaugh and Grier 2001). Different combinations of ethnically diverse sources interact with target medium (ethnic- targeted or mainstream) to strengthen these effects (Green 1999). From the present study, we learned that general multiethnic advertising does not work if the target group was not depicted explicitly (as was the case for our Hispanic subjects), nor is a multiethnic theme enough to trigger favorable target-marketing effects among our multiethnic subjects. Clever experimental manipulations of combinations of the ethnic groups depicted in advertisements, their relative numbers in the ad (i.e., minority, majority, one-of-a-kind status), and the roles they play would inform how source cues are used to induce target-marketing effects among different target populations. Experimental instructions directing subjects from different targeted and nontargeted ethnic groups to process the ads in different ways (e.g., systematic versus heuristic; with cognitive loads versus without cognitive loads) would add insight into the processing mechanisms that underlie target-market effects. Future research might consider the following:

RQ5: What causes asymmetries in ad processes and attitudes between different ethnic groups in response to ads constructed to be monocultural white, monocultural ethnic, multiethnic including white, and multiethnic exclusive of white?

CONCLUSION

We believe that restricting within-study comparisons to two groups has limited the types of generalizations researchers can make about how ethnicity influences advertising responses, and by extension, the types of insights learned about how ethnicity affects message processing, persuasion, source effects, and other topics of interest to consumer and advertising researchers, psychologists, anthropologists, sociologists, policymakers, and marketing practitioners. Our study highlights very clearly how relevant between-group differences in the influence of ethnicity on responses to targeted advertising may be lost when observations for "irrelevant" ethnic groups are omitted or otherwise not considered.

As a result, two-group studies, though valuable in identifying and illustrating descriptive differences between the groups, yield less incremental theoretical insight as research in this area matures. When differences arise between black and white consumers, for example, we are hard-pressed to identify the mechanism underlying the effect. Is it due to enduring in-group cultural beliefs (e.g., ethnic identity, Forehand and Deshpand 2001), transitory in-group cultural beliefs (e.g., situational ethnicity, Stayman and Deshpand 1989), outgroup beliefs related to race (e.g., stereotypes, Triandis 1994), status (Grier and Deshpand 2001), numeric minority representation (Aaker, Brumbaugh, and Grier 2000), or something else? For comparisons between whites and Hispanics, the possibilities expand to include country of origin (Winsberg 1994), level of acculturation for first-generation immigrants (Ueltschy and Krampf 1997), intergenerational influences (Fisher 1995), ability and/or desire to assimilate/acculturate (Pealoza 1994), language facility (Deshpand, Hoyer, and Donthu 1986), language preference independent of ability (Fisher 1994), and more.

We believe the solution lies in the use of experimental designs that use three or more groups for comparison. To be sure, the results of such designs and studies with three or more groups are more difficult to interpret than studies with two groups. It has been suggested to us on more than one occasion to omit the "third group" from research-both in the United States and abroad-in the interest of eliminating noise and improving clarity. Some suggest that certain American groups (e.g., Asians and Hispanics) are troublesome from a research perspective because of their varied geographical backgrounds and that other international groups (e.g., South African Coloureds) are also inconvenient for research because they lack an analogous group in the United States. Ironically, we believe that research investigating these very groups may make some of the most significant contributions to research on ethnicity and culture. For example, by comparing Asian Americans from different countries to Hispanics from different countries in a nested design (country of origin nested under ethnicity), one could separate country effects from ethnicity effects. By comparing Mandarin- preferring Asian Americans to English-preferring Asian Americans to English-preferring Hispanics to Spanish-preferring Hispanics (English versus other language preference crossed with ethnicity), one could distinguish language from ethnicity.

Our study merely begins to illustrate the types of insights that can be learned from multiple group, nested, or crossed designs that allow researchers to disentangle ethnic identity from other confounding forces, including social status, race, numeric minority/ majority status, and distinctiveness. By highlighting contextual influences and the varied confounded meanings associated with different racial and ethnic groups and the dynamics between them, such research may help bridge the gap between the etic perspective implicitly or explicitly reflected in research that compares minority groups with white consumers, and the emic perspective associated with a richer contextualization of ethnic meaning for members of the group themselves (cf. Firat and Venkatesh 1995). It is our belief that research that considers the effects of a third (or fourth or fifth . . .) group to be more than mere noise in the data will enhance theory and practice relevant to advertising in increasingly multiethnic, multiracial, and multicultural marketplaces.

The authors thank Jennifer A. Aaker for her help and insight with this research and Dawn Iacobucci for her comments on a previous version of the paper.

NOTE

1. In a dominant-culture setting in which ethnic participants are a minority, as on this college campus (and most in the United States), it is unlikely that an experimenter of the dominant culture would alter the level of ethnic salience among ethnic participants (cf. Brumbaugh 2002; Stayman and Deshpand 1989). We consider this possibility, however, in the "Questions for Future Research" section.

REFERENCES

Aaker, Jennifer L., Anne M. Brumbaugh, and Sonya A. Grier (2000), "Non-Target Market Effects and Viewer Distinctiveness: The Impact of Target Marketing on Advertising Attitudes, "Journal of Consumer Psychology, 9(3), 127-140.

Annis, Robert C., and Barry Corenblum (1986), "Effect of Test Language and Experimenter Race on Canadian Indian Children's Racial and Self-Identity "Journal of Social Psychology, 126(6), 761-773.

Appiah, Osei (2001), "Ethnic Identification on Adolescents' Evaluations of Advertisements, "Journal of Advertising Research, 41 (5), 7-22.

Asham, H. (1988), "Kuiper's P-Value as a Measuring Tool and Decision Procedure for the Goodness-of-Fit Test," Journal of Applied Statistics, 15 (3), 131-135.

Brett, A. S. Martin, Christina Kwai-Choi Lee, and Feng Yang (2004), "The Influence of Ad Model Ethnicity and Self-Referencing on Attitudes," Journal of Advertising, 33 (4), 27-37.

Brumbaugh, Anne M. (2002), "Source and Nonsource Cues in Advertising and Their Effects on the Activation of Cultural and Subcultural Knowledge on the Route to Persuasion," Journal of Consumer Research, 29 (September), 258-269.

______, and Sonya A. Grier (2001), "Alternative Bases for Character Identification: When Culture Is Not a Factor," paper presented at the Association for Consumer Research Conference, Austin, TX, October.

Bush, Alan J., Rachel Smith, and Craig Martin (1999), "The Influence of Consumer Socialization Variables on Attitude Toward Advertising: A Comparison of African-Americans and Caucasians," Journal of Advertising, 28 (3), 13-24.

Cota, Albert A., and Kenneth L. Dion (1986), "Salience of Gender and Sex Composition of Ad Hoc Groups: An Experimental Test of Distinctiveness Theory," Journal of Personality and Social Psychology, 50 (April), 770-776.

Crockett, David, Sonya A. Grier, and Jacqueline A. Williams (2003), "Coping with Marketplace Discrimination: An Exploration of the Experiences of Black Men," Academy of Marketing Science Review, 4, available at www.amsreview .org (accessed February 5, 2006).

Deshpand, Rohit, Wayne D. Hoyer, and Naveen Donthu (1986), "The Intensity of Ethnic Affiliation: A Study of the Sociology of Hispanic Consumption, "Journal of Consumer Research, 13 (September), 214-220.

______, and Douglas M. Stayman (1994), "A Tale of Two Cities: Distinctiveness Theory and Advertising Effectiveness," Journal of Marketing Research, 31 (January), 57-64.

Eagly, Alice H., Wendy Wood, and Shelly Chaiken (1978), "Causal Inferences About Communicators and Their Effect on Opinion Change," Journal of Personality and Social Psychology, 36 (April), 424-435.

Eberhardt, Jennifer L., Phillip Atiba Goff, Valerie J. Purdie, and Paul G. Davies (2004), "Seeing Black: Race, Crime, and Visual P\rocessing," Journal of Personality and Social Psychology, 87 (6), 876-893.

Firat, A. Fuat, and Alladi Venkatesh (1995), "Liberatory Postmodernism and the Reenchantment of Consumption," Journal of Consumer Research, 22 (3), 239-267.

Fisher, Christie (1994), "Hispanics Indicate Enduring Preference for Native Language," Ad Age, 65 (7), 26-29.

_____ (1995), "Marketing to Hispanics: Younger Hispanics Get the Word," Ad Age, 66 (4), 29-30.

Forehand, Mark R., and Rohit Deshpand (2001), "What We see Makes Us Who We Are: Priming Ethnic Self-Awareness and Advertising Response," Journal of Marketing Research, 38 (3), 336-348.

______, ______, and Americus Reed II (2002), "Identity Salience and the Influence of Differential Activation of the Social Self- Schema in Advertising Response," Journal of Applied Psychology, 87 (6), 1086-1099.

Glazer, Nathan (2002), "Do We Need the Census Race Question?" Public Interest, 149 (Fall), 21-31.

Green, Corliss L. (1999), "Ethnic Evaluations of Advertising: Interaction Effects of Strength of Ethnic Identification, Media Placement, and Degree of Racial Composition," Journal of Advertising, 28 (1), 49-64.

Grier, Sonya A., and Anne M. Brumbaugh (forthcoming), "Compared to Whom? The Impact of Status on Third Person Effects in Advertising Persuasion,"Jo of Consumer Behaviour.

______, and ______ (1999), "Noticing Cultural Differences: Ad Meanings Created by Target and Non-Target Markets," Journal of Advertising, 28 (Spring), 79-93.

______, and ______ (2004), "Consumer Distinctiveness and Advertising Persuasion," in Diversity in Advertising, Jerome D. Williams, Wei-Na Lee, and Curtis P. Haugtvedt, eds., Hillsdale, NJ: Lawrence Erlbaum, 217-236.

______, and Rohit Deshpand (2001), "Social Dimensions of Consumer Distinctiveness: The Influence of Social Status on Group Identity and Advertising Persuasion," Journal of Marketing Research, 38 (May), 216-224.

______, and Ann L. McGill (2000), "How We Explain Depends on Who We Explain: The Impact of Social Category on the Selection of Causal Comparisons and Casual Explanations," Journal of Experimental Social Psychology, 36, 545-566.

Halter, Marilyn (2000), Shopping for Identity: The Marketing of Ethnicity, New York: Schocken Books.

Helms, Janet E., and Regine M. Talleyrand (1997), "Race Is Not Ethnicity," American Psychologist, 52(11), 1246-1247.

Humphreys, Jeffrey M. (2004), "The Multicultural Economy 2004: America's Minority Buying Power," Georgia Business and Economic Conditions, 64 (3).

Keita, S. O. Y, and Rick A. Kittles (1997), "The Persistence of Racial Thinking and the Myth of Racial Divergence," American Anthropologist, 99 (3), 534-544.

Kim, Youn-Kyung, and Jikyeong Kang (2001), "The Effects of Ethnicity and Product on Purchase Decision Making," Journal of Advertising Research, 41 (2), 39-48.

Liefeld, John P. (2003), "Consumer Research in the Land of Oz," Marketing Research, 15 (D, 10-15.

Mandese, Joe (2003), "Asian Market Defies Classification," TelevisionWeek, 22(37), 11.

McGuire, William J. (1984), "Search for the Self: Going Beyond Self-Esteem and the Reactive Self," in Personality and the Prediction of Behavior, Robert A. Zucker, Joel Aronoff, and Albert I. Rabin, eds., New York: Academic Press, 73-120.

______, Claire V. McGuire, and Ward Winton (1979), "Effects of Household Sex Composition on the Salience of One's Gender in the Spontaneous Self-Concept, "Journal of Experimental Social Psychology, 15 (1), 77-90.

______, and Alice Padawer-Singer (1976), "Trait Salience in the Spontaneous Self-Concept, "Journal of Personality and Social Psychology, 33 (6), 743-754.

Mehta, Raj, and Russell Belk (1991), "Artifacts, Identity, and Transition: Favorite Possessions of Indians and Indian Immigrants to the United States, "Journal of Consumer Research, 17(4), 398-411.

Nakayama, Thomas K., and Judith N. Martin (1999), Whiteness: The Communication of Social Identity, Thousand Oaks, CA: Sage.

Pealoza, Lisa (1994), "Atravesando Fronteras/Border Crossings: An Ethnographic Account of the Consumer Acculturation of Mexican Immigrants," Journal of Consumer Research, 21 (1), 32-54.

Reid, Alasdair (2003), "What Is the Best Way to Target Ethnic Minorities?" Campaign (July 11), 8-10.

Rhee, Eun, James S. Uleman, and Hoon Koo Lee (1996), "Variations in Collectivism and Individualism by Intergroup and Culture: Confirmatory Factor Analysis," Journal of Personality and Social Psychology, 71 (5), 1073-1054.

Rose, Gregory M., Victoria D. Bush, and Lynn Kahle (1998), "The Influence of Family Communication Patterns on Parental Reactions Toward Advertising: A Cross-National Examination, "Journal of Advertising, 27 (4), 71-85.

SAS (2004), "Hypothesis Testing in Proc GLM," in SAS 9.1.3 Help and Documentation, Cary, NC: SAS Institute.

Seitz, Victoria (1998), "Acculturation and Direct Purchasing Behavior Among Ethnic Groups in the U.S.: Implications for Business Practitioners, "Journal of Consumer Marketing, 15(1), 23-31.

Stayman, Douglas M., and Rohit Deshpand (1989), "Situational Ethnicity and Consumer Behavior," Journal of Consumer Research, 16 (3), 361-371.

Sue, Stanley (1999), "Science, Ethnicity, and Bias: Where Have We Gone Wrong?" American Psychologist, 54 (12), 1070-1077.

Taylor, Charles R., and Barbara B. Stern (1997), "Asian- Americans: Television Advertising and the Model Minority," Journal of Advertising, 26 (Summer), 1-15.

Tharp, Marye (2001), Marketing and Consumer Identity in Multicultural America, Thousand Oaks, CA: Sage.

Triandis, Harry C. (1994), Culture and Social Behavior, New York: McGraw-Hill.

Tsai, Jeanne L, Yu Wen Ymg, and Peter A. Lee (2000), "The Meaning of 'Being Chinese' and 'Being American': Variation Among Chinese American Young Adults, "Journal of Cross-Cultural Psychology, 31 (3), 302-332.

Tybout, Alice, Brian Sternthal, Geoffrey Keppel, Joseph Verducci, Joan Meyers-Levy, James Barnes, Scott Maxwell, Greg Allenby, and Jan- Benedict Stencamp (2001), "Analysis of Variance," Journal of Consumer Psychology, 10(1/2), 5-25.

Ueltschy, Linda C., and Robert F. Krampf (1997), "The Influence of Acculturation on Advertising Effectiveness to the Hispanic Market," Journal of Applied Business Research, 13 (2), 87-101.

U.S. Bureau of the Census (2001), Projections of the Resident Population by Race, Hispanic Origin, and Nativity: Middle Series, 2050 to 2070, available at www.census.gov/population/projections/ nation/summary/np-T5-G.txt (accessed February 5, 2006).

Venkatesan, M. (1967), "Laboratory Experiments in Marketing: The Experimenter Effect," Journal of Marketing Research, 4 (2), 142- 146.

Webster, Cynthia (1996), "Hispanic and Anglo Interviewer and Respondent Ethnicity and Gender: The Impact on Survey Response Quality," Journal of Marketing Research, 33 (1), 62-72.

Whittler, Tommy E. (1991), "The Effects of Actor's Race in Commercial Advertising: Review and Extension,"Journal of Advertising, 20 (1), 54-60.

Winsberg, Margin (1994), "Specific Hispanics," American Demographics, 16 (2), 44-52.

Wooten, David B. (1995), "One-of-a-Kind in a Full House: Some Consequences of Ethnic and Gender Distinctiveness," Journal of Consumer Psychology, 4 (3), 205-224.

Wynter, Leon E. (1996), "TV Commercials Use More Minority Actors," Wall Street Journal (November 6), Bl.

Yee, Albert H. (1983), "Ethnicity and Race: Psychological Perspectives," Educational Psychologist, 18(1), 14-24.

______, Halford H. Fairchild, Fredric Weizmann, and Gail E. Wyatt (1993), 'Addressing Psychology's Problem with Race," American Psychologist, 48(11), 1132-1140.

Anne M. Brumbaugh (Ph.D., Duke University) is an assistant professor of marketing, Babcock Graduate School of Management, Wake Forest University.

Sonya A. Grier (Ph.D., Northwestern University) is an associate professor of marketing, Kogod School of Business, American University.

Copyright American Academy of Advertising Fall 2006

(c) 2006 Journal of Advertising. Provided by ProQuest Information and Learning. All rights Reserved.