The Influence of Emotional Appeals across Cultures - CiteSeerX

Empathy versus Pride: The Influence ofEmotional Appeals across Cultures

JENNIFER L. AAKERPATTI WILLIAMS*

This research examines the persuasive effect of emotional appeals on membersof collectivist versus individualist cultures. The results of two experiments demon-strate that ego-focused (e.g., pride, happiness) versus other-focused (e.g., empa-thy, peacefulness) emotional appeals lead to more favorable attitudes for mem-bers of a collectivist culture, while other-focused versus ego-focused emotionalappeals lead to more favorable attitudes for members of an individualist culture.Experiment 2 was conducted to examine the psychological mechanism underly-ing these effects. The results indicated that the generation of and elaboration ona relatively novel type of thought ( individual thoughts for members of a collectivistculture, collective thoughts for members of an individualist culture) account forthe persuasive effects found in this research. These results are interpreted withinan ability-motivation framework, and theoretical implications involving cross-cul-tural persuasion effects are discussed.

A cultural orientation and emotion to examine how emo-tional appeals operate differently across cultures.

considerable amount of research in consumer behav-ior has examined how emotions are used in persua-

sion appeals. Researchers have focused on how emotions In addition, we explore the psychological mechanismunderlying the persuasion effects of emotional appealsinfluence affective responses (Batra and Ray 1986), at-

tention to the advertisement (Olney, Holbrook, and Batra across cultures. In experiment 1, we demonstrate that theeffect of emotions on attitudes is related to perceptions1991), brand and advertisement memories (Friestad and

Thorson 1993), and attitudes (Edell and Burke 1987). of self and other, which tend to vary systematically acrosscultures. In experiment 2, we examine the moderatingHowever, while recent research in cultural psychology

suggests that cultural orientation has a significant effect effect of reference type on the persuasive effect of emo-tional appeals for members of individualist versus collec-on processes of persuasion (Aaker and Maheswaran 1997;

Triandis 1989) as well as patterns of emotions (Markus tivist cultures and show that the persuasive effects ofemotional appeals are mediated by the different type ofand Kitayama 1994; Matsumoto 1989), very little work

has examined the differential effect of emotional appeals thoughts generated and elaborated on by members of thetwo cultures.on persuasion across cultures. A growing interest in the

globalization of marketing activities and the cross-culturaluse of advertisements in particular makes this gap in the

EMOTIONS, CULTURE,literature even more significant (see Han and Shavitt1994). This research attempts to bridge the research on AND THE SELF

Emotions have been viewed as a universal set of inter-nal processes that are largely hardwired, arising when an

*Jennifer L. Aaker is assistant professor of marketing at the Anderson event that is relevant to the concerns of an individualSchool of Management, University of California, Los Angeles, 110

occurs (Darwin 1896). Since many concerns are sharedWestwood Boulevard, Box 951481, Los Angeles, CA 90095-1481. Pattiacross cultures, emotional experiences tend to be cross-Williams is assistant professor of marketing at the Stern School of

Business, New York University. This research was funded by Jose de culturally similar. In fact, research on emotions has docu-la Torre and the Center for International Business, Education, and Re- mented universal consensus in evaluations of basic emo-search at University of California, Los Angeles. The authors thank Susan tions such as evaluation, activity, and potency (e.g.,Chiu, Loraine Lau, and Durairaj Maheswaran for their comments on

Osgood, May, and Miron 1975), as well as specific emo-earlier drafts, and particularly Ivy Fei, Sally Lee, and Bernd Schmitt,the BAT Chair of Marketing at China Europe International Business tions such as anger, fear, empathy, and surprise (e.g.,School in Shanghai (China) , for their help in data collection. Finally, Matsumoto 1989). However, although the recognitionthe authors extend great amounts of thanks to Bob Burnkrant, Debbie and experience of many emotions appears to be cross-MacInnis, and three JCR reviewers.

culturally robust, significant differences exist as well.241

q 1998 by JOURNAL OF CONSUMER RESEARCH, Inc. Á Vol. 25 Á December 1998All rights reserved. 0093-5301/99/2503-0003$03.00

/ 9h10$$de01 11-17-98 09:17:00 cresa UC: Con Res

242 JOURNAL OF CONSUMER RESEARCH

These differences tend to be attributed to systematic cul- levels of arousal, flow, or integration (e.g., Batra and Ray1986; Edell and Burke 1987; Holbrook and Batra 1987;tural variations in the concerns of individuals, which arise

from distinct self-construal patterns (Triandis 1993). Kamp and MacInnis 1995; Stout, Homer, and Liu 1990).Despite this increased interest in the effect of emotionson persuasion, however, differences in emotions basedDifferences in Self-Construal in Individualiston their primary referent of focus have not been examined

versus Collectivist Cultures in consumer behavior research. The recognition of thisdistinction in cultural psychology has led to the termsIndividualism-collectivism is a cultural-level variable‘‘ego-focused’’ and ‘‘other-focused,’’ which refer to ‘‘thereferring to the extent to which members of a culture tenddegree to which specific emotions systematically vary into have an independent versus interdependent construalthe extent to which they follow from, and also fosterof the self (e.g., Hofstede 1980).1 In individualist culturesor reinforce, an independent versus interdependent self ’’(e.g., the United States, Australia, and Canada), an inde-(Markus and Kitayama 1991, p. 235). Ego-focused emo-pendent self-construal, which refers to the self as compris-tions (e.g., pride, happiness, frustration, anger) tend toing a unique set of internal attributes including motiva-be associated with an individual’s internal state or attri-tions, traits, and values, tends to be fostered. In contrast,butes, to the exclusion of others, and are consistent within collectivist cultures (e.g., China, Japan, and Taiwan)the need for individual awareness, experience, and expres-an interdependent self-construal, which refers to the selfsion. In contrast, other-focused emotions (e.g., empathy,as inseparable from others and social context, tends to bepeacefulness, indebtedness, shame) tend to be associatedfostered.with others in a social context or close others ( i.e., family,As a result of these distinct self-construals, the con-friends, coworkers, political or religious groups, socialcerns of members of individualist versus collectivist cul-class, or an ideological or national entity that is centrallytures often vary. Individualists tend to be concerned withimportant to the individual’s self-definition [Triandisseparating one’s self from others, displaying qualities of1994]) and are consistent with the need for unity, har-uniqueness and not being influenced by others. Conse-mony, and the alignment of one’s actions with those ofquently, the personal and subjective part of one’s emo-another.2tional experience, to the exclusion of others, is salient.

Importantly, cultural orientation appears to influenceCollectivists tend to be concerned with affiliating withthe ability to recognize these two emotion types due toclose others, maintaining connectedness, and blending thedifferent levels of accessibility of ego-focused versusself /other boundary (Singelis 1994). Consequently, theother-focused emotions across cultures. More specifically,interpersonal and intersubjective aspects of emotional ex-acknowledging certain emotions is discouraged in someperience are salient. In this light, ‘‘the self-structure func-cultures, resulting in lowered levels of accuracy in theirtions as an individualized orienting, mediating, interpre-identification (Matsumoto 1989). Similarly, research ontive framework giving shape to what people notice andthe self demonstrates that culturally self-relevant versusthink about, to what they are motivated to do and toself-irrelevant emotions tend to be easier to recognizehow they feel and their ways of feeling’’ (Markus andsince there is a greater presence of cognitive categoriesKitayama 1994, p. 92).with which to appraise and process them (Markus, Mo-reland, and Smith 1985). Lazarus (1982) argues that this

Persuasion Appeals and the Distinction cognitively based appraisal process, which tends to occurbetween Ego-Focused and Other-Focused automatically, involves the recognition of certain aspects

of a stimulus situation as emotionally relevant to an indi-Emotionsvidual. Thus, culturally derived knowledge structures can

A growing amount of research has examined the role be utilized in the appraisal process, driving the recogni-of emotions in consumer behavior (cf. Holbrook and tion of which aspects in a situation are self-relevant andHirschman 1982). This work has primarily distinguished consequently likely to be emotionally significant for thebetween emotions based on their content, valence, and individual. As a result, having an interdependent versus

independent self emphasizes particular domains of con-cern, leading to different levels of accessibility of ego-

1Individualism-collectivism has been treated as a dichotomous cul- focused versus other-focused emotions, and thereby in-tural level variable by many researchers and as an individual differencevariable (e.g., idiocentrism-allocentrism, independent-interdependent)by others. In this research, we focus on individualism-collectivism as a

2It should be noted that the ego-focused versus other-focused emotioncultural variable rather than an individual difference variable for tworeasons. First, the cultural level conceptualization is consistent with variable differentiates between emotions that otherwise may be per-

ceived as similar if one were considering only valence or arousal. ForHofstede (1980) and Triandis (1989), which provide the basis fortheory development. Second, the within-culture variance of these con- example, although anger and shame are similarly negative and arousing,

anger tends to be less interpersonally engaged than shame. Similarly,structs tends to be small relative to the between-culture variance (Tri-andis 1993). Thus, the theoretical predictions can be better tested by while happiness and empathy are both positive and moderately arousing,

empathy tends to be more interpersonally engaged than happiness (Ki-treating individualism-collectivism as a dichotomous cultural level vari-able. tayama and Markus 1995).


243EMOTION AND CULTURE

fluencing the accessibility patterns of both emotional ap- the frequency with which they felt the emotions in every-day life. American participants reported greater frequencypraisal and arousal (Frijda and Mesquita 1994).

To clarify the distinction between ego-focused and of experiencing socially disengaged emotions (e.g., pride,anger) , while Japanese participants reported greater fre-other-focused emotions, consider the following story: A

graduate student lost an award competition, unsuccess- quency of experiencing socially engaged emotions (e.g.,friendly feelings, feelings of indebtedness) .fully appealed it, and subsequently failed to get an aca-

demic job. Soon after, he entered his graduate program Finally, the literature on emotions and persuasion dis-criminates between the recognition of emotions depictedoffice and shot his advisor, the person who handled the

appeal, several bystanders and himself (as cited in Morris in advertisements and the felt emotional responses thatsuch depictions can evoke. The recognition of the emotionand Peng [1994]) . What emotions will a reader of this

story feel most intensely? In individualist cultures, the depicted in an appeal tends to represent a cognitivelybased appraisal process and does not necessarily lead topersonality traits or internal attributes of the student tend

to be highlighted (e.g., ‘‘very bad temper,’’ ‘‘darkly dis- arousal or the actual experience of emotion (Arnold1960). However, emotional appeals tend to be createdturbed man’’; Morris and Peng 1994, p. 958), which

may result in relatively more intense feelings of anger or to evoke emotional reactions in consumers that, in turn,influence attitudes (Holbrook and Batra 1987). In thedisgust than in collectivist cultures. In collectivist cul-

tures, others in the social context or the situation in which present set of experiments, we examine the cultural differ-ences in the ability to appraise certain types of emotionalthe student found himself tend be highlighted (e.g., ‘‘did

not get along with his advisor,’’ ‘‘isolation from Chinese appeals as culturally relevant. As a result, we focus onthe types of emotions depicted in advertising appealscommunity’’; Morris and Peng 1994, p. 958), which may

result in relatively more intense feelings of understanding rather than the emotional responses evoked in responseto those appeals.or empathy than in individualist cultures.

Similarly, consider a scenario in which an individualwins a prize in a competitive contest. Ellsworth (1994)

Cultural Differences in Ability: Differentialsuggests that more intense feelings of pride for excellingmay be experienced if the winner is from an individualist Effectiveness of Ego-Focused and Other-versus collectivist culture, while more intense feelings of Focused Emotional Appealsguilt for having been helped by others and gaining thefocus of attention may be experienced if the winner is In this research, we posit that the cultural differences

in accessibility of emotion types may lead to differentialfrom a collectivist versus individualist culture.In both scenarios, the relative felt intensity of these levels in the ability to recognize and experience these

emotions, thereby affecting their effectiveness in persua-emotions differs across cultures due to a varying focusof attention typically held by members of individualist sion contexts. Specifically, the heightened accessibility of

ego-focused emotions in individualist cultures, or other-cultures (where the primary referent tends to be the dis-tinct self) versus collectivist cultures (where the primary focused emotions in collectivist cultures, should increase

the ability of members of such cultures to process incom-referent tends to be close others and the social situation;cf. Smith et al. 1993). For example, when the referent of ing appeals relying on those emotions. Thus, we predict

that ego-focused versus other-focused emotional appealsfocus is a blocked individual goal or an abridged individ-ual right, thereby highlighting the disengagement of the should have greater influence on attitudes for members

of individualist cultures, and other-focused (vs. ego-fo-self from others, anger tends to be felt. In contrast, whenthe referent of focus is others in a situation or a collective cused) emotional appeals should have greater influence

on attitudes for members of collectivist cultures.goal that has been aided, thereby highlighting the syn-chrony between the self and others, empathy tends to be Further, we suggest that the psychological mechanism

underlying these effects is the difference in self-construalfelt.However, although this research suggests that cultural patterns, which should be reflected in the types of

thoughts generated by members of the two cultures indifferences in the relative accessibility of the two emotiontypes exist, two points should be noted. First, there is persuasion contexts. Specifically, because members of in-

dividualist versus collectivist cultures tend to differ inconsiderable overlap concerning cross-cultural percep-tions of what emotions are self-focused and other-focused their perceptions of the self, and because certain emotions

vary in the extent to which they access these perceptions,(Matsumoto et al. 1988). Second, the differential levelsof accessibility of the emotion types not only lead to the nature of thoughts evoked by the two emotional appeal

types should vary systematically for members of the twocultural differences in felt intensity, they lead to variationin the relative frequency with which the two emotion cultures. Persuasive appeals relying on ego-focused, as

opposed to other-focused, emotions should lead to moretypes are experienced across cultures. To illustrate, Kitay-ama, Markus, and Kurokawa (1991) gave college stu- favorable attitudes for members of individualist cultures

because they access the independent self (self-as-distinct-dents in the United States and Japan a number of labelsfor emotions, some indigenous to America and others from-others) , which can be measured by the number of

generated individual thoughts ( i.e., thoughts relating toindigenous to Japan, and asked the participants to rate



the self as a distinct individual) . In contrast, appeals rely- on attitudes and thoughts is examined. Thus, experiment1 relies on a 2 (emotional appeal type: ego-focused vs.ing on other-focused, as opposed to ego-focused, emo-

tions should lead to more favorable attitudes for members other-focused) 1 2 (cultural orientation: individualismvs. collectivism) between-subjects design.of collectivist cultures because they access the interdepen-

dent self (self-in-relation-to-others) and thus can be mea-sured by the number of generated collective thoughts ( i.e., Methodthoughts relating to close others) . Therefore, we suggest

Participants were undergraduates recruited from anthat the attitudes toward an appeal relying on an ego-American university (n Å 60; mean age Å 21.52, femalefocused or other-focused emotion are mediated by theÅ 75 percent, all were born and raised in the Unitednature of cognitive responses. Specifically,States, 87 percent were Caucasian) and a Chinese univer-

H1a: Members of individualist cultures will have sity (n Å 90; mean age Å 19.12, female Å 63 percent,more favorable attitudes when exposed to ego- all were born and raised in Shanghai, 100 percent werefocused versus other-focused emotional ap- native Chinese) to participate in a new brand study forpeals. $5 and 10 rmb, respectively. The beer product category,

which is relevant for participants in both cultures, wasH1b: Members of collectivist cultures will haveselected because both ego-focused and other-focusedmore favorable attitudes when exposed toemotional advertisements could be created. In the coverother-focused versus ego-focused emotionalstory, participants read the following excerpt:appeals.

A large-scale beer manufacturer is planning to introduceH2a: For members of individualist cultures, the re-a new brand of beer to the U.S. (China) market—namedlationship between ego-focused appeals andOhio Flag. We would like to know what you think aboutattitudes will be mediated by individual it before it is introduced. In the last month, the manufac-

thoughts. turer has asked consumers to taste the beer and list theattributes that make this new brand of beer unique fromH2b: For members of collectivist cultures, the rela-other beers they have tried. You will be asked to read sometionship between other-focused appeals andof their statements to give you a feel about what Ohio Flag

attitudes will be mediated by collective beer is like. Then, we will give you an example of thethoughts. potential print advertising message created for Ohio Flag.

We want to know both your feelings and thoughts aboutthe beer as well as the advertisement.EXPERIMENT 1

To enhance the realism of the cover story, participantsOverviewwere given consumer excerpts that included five one-sentence quotes, all of which favorably described the beerThe objective of experiment 1 is to examine whether

different types of emotional appeals produce asymmetric on an attribute pretested to be important across cultures(bottle design, taste of beer, number of calories, aroma,persuasion effects for members of collectivist versus indi-

vidualist cultures and to determine the extent to which and level of carbonation). For example, participants readthat one consumer felt that ‘‘the carbonation is justattitudes toward ego-focused versus other-focused emo-

tional appeals are mediated by individual versus collective right—not too fizzy, but it has zip.’’ We created posi-tively valenced appeals, both in terms of emotion andthoughts. To achieve these objectives, we use two posi-

tively valenced emotions (which varied across condi- content, for two reasons. First, since most advertisementsare created with the goal of approach versus avoidance,tions) to create advertising appeals with positively va-

lenced content (which remained constant across external validity is heightened. Second, the reliance onpositively valenced emotions in particular reduces the po-conditions) . In experiment 1, the two appeals evoked

pride (ego-focused emotion) or empathy (other-focused tential of a confound between cultural accessibility andvalence. To illustrate, Markus and Kitayama (1994) noteemotion), which were selected for two reasons. First, the

research on emotion and culture suggests that they are that anger is a negative emotion for collectivists, not onlybecause of its valence but because of a feeling of per-strong examples of ego-focused and other-focused emo-

tions (Markus and Kitayama 1994). Pride tends to in- ceived disengagement from others—which may be anemotion that is less accessible for collectivists versus indi-volve one’s internal attributes as the primary referent and

fosters independent feelings, separation from others, and vidualists.Next, participants were given the print advertisementdistinctiveness. In contrast, empathy tends to involve oth-

ers’ feelings or attributes as a primary referent, fostering in which the positively valenced emotional appeal wasmanipulated, followed by a series of questions regardingfriendly feelings, feelings of affiliation, and connected-

ness. Second, both pride and empathy are used in advertis- participants’ thoughts and attitudes toward the advertise-ment and brand. Participants were then given an unrelateding appeals in collectivist as well as individualist cultures

(Aaker, Stayman, and Hagerty 1986; Mueller 1987). The five-to-seven-minute filler task (in which they were askedto list likable and unlikable celebrities who were highlyinfluence that these two types of emotional appeals have



relevant to the product categories of automobiles and den- moving; Cronbach’s a Å .93) , as well as 37 filler emo-tions. Each message depicted the intended emotiontal care products) , followed by demographic questions.

Then, participants were asked if they drank beer (1Å yes, (F (empathy) Å 12.02, p õ .01; F (pride) Å 3.02, p õ .05) ,and no differences across cultural orientation were found2 Å no); if yes, how frequently (1 Å more than once a

week, 2 Å once a week, 3 Å once a month, 4 Å once a (Fõ 1). In addition, when American and Chinese partici-pants rated the brand name, Ohio Flag, on the pride-year, 5 Å less than once a year); and to rate their knowl-

edge of beer (1 Å extremely limited, 5 Å extensive) .3 related and empathy-related emotions, no significant ef-fects involving the depicted emotions and culture wereFinally, participants completed a set of manipulation

checks and Singelis’s (1994) Independent-Interdependent found (F’s õ 1).Scale.

Dependent VariablesAttitudes. Participants completed three attitude ques-Independent Variables tions involving the advertisement (1 Å bad, not at all

likable, unfavorable; 7 Å good, likable, favorable) andCultural Orientation. Past research on cultural orien-brand (1 Å bad, not at all likable, unfavorable; 7 Å good,tation suggests that the United States tends to be highlylikable, favorable) . In addition, participants were askedindividualistic and China tends to be highly collectivistichow much they liked the brand of beer (1 Å extremely(Hofstede 1980). Therefore, university subjects in thedislike, 7 Å extremely like) , would they try this brandUnited States and China, matched in demographic profile,of beer (1 Å definitely no, 7 Å definitely yes) , and wouldwere recruited to participate in this experiment. The origi-they buy this brand of beer the next time they saw it innal questionnaire was drafted in English, then translateda store (1Å definitely no, 7 Å definitely yes) . An averageinto Chinese. Accuracy of the Chinese version was veri-of these responses led to a three-item advertisement atti-fied via a back-translation procedure using external trans-tude index (Cronbach’s a Å .95) and a six-item brandlators (Hui and Triandis 1985).attitude index (Cronbach’s a Å .86) .

Emotional Appeal Type. To manipulate emotional Cognitive Responses. Two sets of independent raters,appeal type, two print advertisements were created. They two American raters for the individualist participants’were made to be as similar as possible, varying only in thoughts and two Chinese raters for the collectivist parti-the target emotion. In the pride appeal, participants read, cipants’ thoughts, categorized cognitive responses. Amer-‘‘Acing the last exam. Winning the big race. Receiving ican raters coded the thoughts in English, while Chinesedeserved recognition. Ohio Flag Beer. Celebrating life’s raters, both of whom were bilingual but born and raisedaccomplishments.’’ In the empathy appeal, participants in Shanghai, coded the thoughts in Chinese. To ensureread, ‘‘Reminiscing with old friends. Enjoying time to- that both sets of raters coded the thoughts similarly, allgether with family during the holidays. Relaxing near the were trained in cognitive response analysis and given afire with best friends. Ohio Flag beer. Celebrating the series of examples of each thought type. Both sets ofrelationships that matter most.’’ raters coded the common set of examples similarly and

To ensure that these appeals were perceived as de- were encouraged to ask the authors any questions of clari-picting the target emotion in both cultures, a pretest was fication when they coded subsequent thoughts (Brislinconducted whereby American (n Å 10) and Chinese (n 1980).Å 10) participants were asked to ‘‘please indicate how Adopted from Meyers-Levy and Peracchio (1996) andmuch the following emotions describe the advertisement Sujan, Bettman, and Baumgartner (1993), cognitive re-for Ohio Flag beer, by circling one number for each word sponses were coded in terms of individual thoughtsbelow’’ for each appeal (1 Å not at all, 7 Å very (thoughts relating to the self as a distinct individual; anystrongly) . Drawing on Frijda (1986) and Edell and Burke thought that included the use of ‘‘me’’ or ‘‘I’’ as the(1987), this list included three pride-related emotions primary subject; e.g., ‘‘I could really imagine myself try-(proud, confident, excited; Cronbach’s a Å .90) and three ing this beer’’) and collective thoughts ( thoughts relatingempathy-related emotions (warmhearted, emotional, to just close others; any thought that included family

members or friends as the primary subject; e.g., ‘‘Mymom could appreciate this brand’’) . In cases where bothindividual and collective thoughts were included in a sen-3Although the drinking age differs in the United States and China

(the drinking age in the United States is 21; no drinking age exists in tence (e.g., ‘‘My friends and I would love this beer’’) ,China) , no differences in product category knowledge or relevance coders double coded the thoughts as individual and collec-between the two samples was found. Approximately 65 percent of tive.4 Finally, product-related thoughts (e.g., ‘‘The beerthe Chinese participants and 80 percent of the American participantsdrink beer; both drink it approximately once a week (modal response) .In addition, Chinese participants’ self-reported knowledge of the beer

4Lin (1981) notes that the Chinese (vs. English) language relies oncategory is similar to that of American participants (XV collectivist Å 2.64,XV individualist Å 2.74; F õ 1). Finally, an additional set of analyses was a fewer number of pronouns because of the ability to create sentences

that consist of only the predicate, rather than the subject and predicate.run excluding American participants under the age of 21. The patternof results with this subset mirrored that found with the overall American The subject (e.g., a pronoun) may not be mentioned because it is either

impossible or unnecessary (e.g., if the pronoun ‘‘she’’ was used in asample.



seemed of high quality’’) , advertising-related thoughts F Å 6.36, p õ .01) .5 No other significant effects werefound.(e.g., ‘‘The images in the ad were really nice’’) , and

irrelevant thoughts ( thoughts that were theoreticallyAttitudes. An ANOVA on attitude toward the adver-meaningless; e.g., ‘‘Where would I buy the beer?’’) were

tisement yielded no significant main effects. Hypothesisalso coded. However, since they were not involved in1a, which suggested that members of individualist cul-the hypotheses, they are not mentioned further. Interratertures have more favorable attitudes when exposed to ap-agreement was high (95 percent agreement for Chinesepeals relying on ego-focused versus other-focused emo-coders, 91 percent for American coders) . When discrep-tions, was not supported. Further, Hypothesis 1b, whichancies existed, coders were instructed to discuss themsuggested that members of collectivist cultures have moreuntil a consensus was formed, a process that typicallyfavorable attitudes when exposed to appeals relying ontook less than a minute.other-focused versus ego-focused emotions, was not sup-ported.

However, a significant interaction was found on bothResultsattitudes toward the advertisement (F Å 17.20, põ .001)and brand (F Å 12.13, p õ .001), which yielded resultsThe hypotheses were tested based on a 2 (emotionalin the opposite direction of what was predicted. Membersappeal type) 1 2 (cultural orientation) between-subjectsof the individualist culture had more favorable advertise-analysis of variance (ANOVA). Three covariates werement attitudes when exposed to appeals relying on theused (separately) in each analysis reported below: prod-other-focused emotion (XV Å 5.08) than ego-focused emo-uct category usage, frequency of usage, and self-ratedtion (XV Å 3.87; F Å 16.73, p õ .001). Further, memberscategory knowledge. None of the covariates was signifi-of the collectivist culture had more favorable advertise-cant in the analyses, so they will not be discussed below.ment attitudes when exposed to advertising appeals rely-Unless otherwise specified, df Å 1, 147.ing on the ego-focused emotion (XV Å 4.83) than the other-focused emotion (XV Å 4.10; F Å 4.08, p õ .05) .Manipulation Checks. The results of a two-way AN-

OVA indicated that the emotional-appeal-type manipula- Analysis of brand attitudes revealed the same patternof results, with members of the individualist culture re-tion worked as expected. When asked to indicate the ex-

tent to which the emotions describe the Ohio Flag beer porting more favorable brand attitudes when they viewedthe advertising appeal featuring the other-focused versusadvertisement, participants rated the empathy appeal (XV

Å 4.93) higher than the pride appeal (XV Å 4.26) on ego-focused emotion (F Å 11.97, põ .01) , and membersof the collectivist culture exhibiting marginally more fa-empathy-related emotions (Cronbach’s a Å .87; F

Å 3.64, p õ .05) . Similarly, participants rated the pride vorable brand attitudes when viewing the advertising ap-peal featuring the ego-focused versus other-focused emo-appeal (XV Å 5.39) higher than the empathy appeal (XV

Å 4.48) on pride-related emotions (Cronbach’s a Å .93; tion (F Å 2.81, p õ .09) . This unexpected pattern ofresults is discussed below (see Fig. 1) .F Å 5.12, p õ .05) . Moreover, the advertising appeals

were generally perceived as being either ego-focused orCognitive Responses. To examine the extent to whichother-focused as intended. As expected, the pride appeal

the relationship between ego-focused appeals and atti-evoked more individual thoughts (XV Å .64) than did thetudes is mediated by the number of individual thoughtsempathy appeal (XV Å .46; F Å 2.77, p õ .05) . Thefor members of individualist cultures, four analyses wereempathy appeal evoked marginally more collectiveconducted (for both AAD and AB): (a) individualist parti-thoughts (XV Å .35) than the pride appeal (XV Å .24; Fcipants have more favorable attitudes when exposed toÅ 1.60, põ .10) . No other significant effects were found.an ego-focused versus other-focused emotional appeal;To ensure that the individualism-collectivism cultural(b) ego-focused versus other-focused emotional appealsvariable was tapped through the use of American versuslead to more individual thoughts; (c) individual thoughtsChinese participants, an interdependence-independencelead to more favorable attitudes; and (d) ego-focusedindex was created by averaging the 31 items of theversus other-focused emotions are not a significant pre-Singelis (1994) scale (e.g., ‘‘I feel it is important for medictor of attitudes in a model that includes both emotionalto act as an independent person,’’ ‘‘It is important for meappeal type and thought type (Baron and Kenny 1986).to maintain harmony within my group,’’ ‘‘My personalA conceptually similar set of analyses was conducted toidentity, independent of others, is very important to me’’;determine if the relationship between ego-focused appealsCronbach’s a Å .90) . Consistent with Hofstede (1980),

American participants (XV Å .41) received higher indepen-dent scores than did Chinese participants (XV Å 0.22;

5Note that Singelis (1994) conceptualizes independence and interde-pendence as two individual-level dimensions. However, in this research,a factor analysis of the 31 items yielded only one factor, which may

prior sentence or context, the pronoun would not need repeating). How- have been due to limited sample size or insufficient situational variabilityin the items to reflect distinct independent and interdependent dimen-ever, the relative usage of pronoun types (e.g., ‘‘I’’ vs. ‘‘we’’) is similar

across the two cultures (Comrie 1981). sions.



FIGURE 1

EXPERIMENT 1 RESULTS

and attitudes is mediated by the number of collective other-focused emotion (XV Å .33; F Å 3.27, p õ .05) .Further, the number of collective thoughts did not differthoughts for members of collectivist cultures.

As highlighted above in the failed support for Hypothe- for the two emotional appeal types for members of thecollectivist culture (F õ 1) but was higher for memberssis 1a and Hypothesis 1b, the first criterion was not sup-

ported; rather it was reversed. Further, the remaining cri- of the individualist culture when exposed to the appealevoking the other-focused (XV Å .39) versus ego-focusedteria (b– d) received no support (p’s ú .10) . Therefore,

the predicted mediation process was not supported. emotion (XV Å .07; F Å 4.98, p õ .05) .To further examine this pattern of results, an ad hocHowever, in line with the results involving attitudes, a

systematic pattern did occur with individual and collec- mediation test was conducted on the reverse relationshippredicted in Hypotheses 2a and 2b. First, we examinedtive thoughts, but in the opposite direction of what was

expected. The number of individual thoughts did not dif- the extent to which the relationship between ego-focusedappeals and attitudes is mediated by the number of indi-fer in the two emotional appeal type conditions for mem-

bers of the individualist culture (F õ 1) but was higher vidual thoughts for members of collectivist cultures. Con-sistent with analysis a , collectivist participants had morefor members of the collectivist culture when exposed to

the appeal evoking the ego-focused (XV Å .80) versus the favorable attitudes when exposed to an ego-focused ver-



sus other-focused emotional appeal (bA(AD) Å .60, p cused emotions were more persuasive for members ofindividualist cultures. In addition, the predicted mediationõ .01; bA(B) Å .41, p õ .01) . Consistent with analysis

b , ego-focused versus other-focused emotional appeals process did not occur. Instead, attitudes toward other-focused emotional appeals appear to be mediated in partled to more individual thoughts for collectivists (bA(AD)

Å .92, p õ .01; bA(B) Å .52, p õ .01) . Consistent with by collective thoughts for individualists, while attitudestoward ego-focused emotional appeals appear to be medi-analysis c , individual thoughts led to more favorable atti-

tudes for collectivists (bA(AD) Å .16, p Å .01; bA(B) ated by individual thoughts for collectivists.Thus, although the predicted differences in ability mayÅ .16, p õ .01) . Finally, in partial support of analysis

d , individual thoughts was a marginally significant pre- exist across cultures, another mechanism appears to drivethe results observed in experiment 1. In particular, otherdictor of attitudes (bA(AD) Å .72, p õ .01; bA(B) Å .37, p

õ .07) , and ego-focused emotional appeal was a signifi- antecedents that affect the processing of appeals may bemore important. One such antecedent is motivation tocant predictor of attitudes (bA(AD) Å .48, p õ .01; bA(B)

Å .34, p õ .01) . process persuasive messages (MacInnis and Jaworski1989; MacInnis, Moorman, and Jaworski 1991).Next, we examined the extent to which the relationship

between other-focused appeals and attitudes is mediatedby the number of collective thoughts for members of

Motivation versus Ability: Culturalindividualist cultures. Consistent with analysis a , individ-ualist participants had marginally more favorable attitudes Differences in the Effectiveness of Emotionalwhen exposed to an other-focused versus ego-focused Appealsadvertising appeal (bA(AD) Å.73, p õ .05; bA(B) Å .49, põ .07) . Consistent with analysis b , other-focused versus Previous research suggests that the emotions pride and

empathy are used in advertising appeals in both cultures.ego-focused emotional appeals led to marginally morecollective thoughts for individualists (bA(AD) Å .46, p However, the relative frequency with which they are used

in the two cultures may vary. If so, relative differencesõ .06; bA(B) Å .47, p õ .06) . Consistent with analysisc , collective thoughts led to significantly more favorable in the levels of novelty of these appeals may exist across

cultures. Appeals involving other-focused versus ego-fo-attitudes for individualists (bA(AD) Å .39, p õ .04; bA(B)

Å .36, p õ .01) . Finally, in support of analysis d , collec- cused emotions may be relatively more novel in individu-alist cultures, while appeals involving ego-focused versustive thoughts were marginally significant (bA(AD) Å .32, p

õ .09; bA(B)Å .37, põ .03) , but other-focused emotional other-focused emotions may be relatively more novel incollectivist cultures. Encountering either cosmetically orappeal was not significant (bA(AD) Å .59, p õ .12; bA(B)

Å .34, p õ .21) when predicting attitudes. substantively novel stimuli can increase motivation toprocess messages (Goodstein 1993). Novel stimuli,Together, these results provide some support for the

notion that the relationship between ego-focused emo- which are often operationalized via the use of uniqueexecutional elements within an advertisement (e.g., un-tional appeals and attitudes may be partially mediated by

individual thoughts for collectivists, while the relationship usual cinematography; Alsop 1988), automatically cap-ture attention (Berlyne 1960), thereby increasing the like-between other-focused emotional appeals and attitudes

may be mediated by collective thoughts for individualists. lihood that individuals will switch from relativelyautomatic modes of processing to more strategic modesin an attempt to resolve any ambiguity of the stimulus orDiscussionto understand its novel properties (Grunert 1996).

Perhaps more important, novel stimuli have a second-Previous research has focused on differences in therelative accessibility of cognitive categories for ego-fo- ary effect: beyond increasing elaboration at an absolute

level, novel stimuli change the nature of elaboration. Forcused and other-focused emotions across cultures due todifferences in self-construal. This variation in accessibil- example, Oliver, Robertson, and Mitchell (1993) show

that novel stimuli provoke a different kind of elaborationity implies a differential ability to recognize and experi-ence one type of emotional appeal as opposed to the (e.g., imaginal rather than analytical) . In this light, there

are two components of the novel stimulus that are criticalother, suggesting that individuals are experts at processingemotions that are relevant to their self-construal, doing to the process underlying the results found in experiment

1: (1) the amount of elaboration and (2) the nature ofso in a relatively automatic manner (Grunert 1996).Based on this premise, experiment 1 was conducted to the elaboration. That is, the novel stimulus (other-focused

emotional appeals for individualists, ego-focused emo-determine the differential levels of effectiveness of ego-focused versus other-focused emotional appeals on atti- tional appeals for collectivists) may lead to an increased

motivation to elaborate on the appeal. However, the in-tudes of members of individualist versus collectivist cul-tures. In contrast to the expected results that would follow creased elaboration is specific in nature; the thoughts elab-

orated on are novel (collective thoughts for individualists,from cultural differences in ability, however, appeals rely-ing on ego-focused versus other-focused emotions were individual thoughts for collectivists) and positively va-

lenced (because the appeal contained positively valencedmore persuasive for members of collectivist cultures,while appeals relying on other-focused versus ego-fo- content and emotions) . Thus, an increased amount of



elaboration on a positively valenced stimulus generates In contrast, self-referencing should not enhance the per-suasive effect of other-focused emotional appeals forpositively valenced thoughts, leading to more favorable

attitudes (Haugtvedt and Petty 1992). In contrast, when members of individualist cultures, which was accompa-nied by an increased number of collective thoughts inindividuals encounter a more common or typical advertis-

ing appeal, they may be less likely to spontaneously elab- experiment 1. Rather, inducing members of individualistcultures to think of relationships with close others shouldorate (Goodstein 1993), reducing the likelihood that

novel, positively valenced thoughts would arise and more enhance the persuasive effect of other-focused emotions.We call this manipulation ‘‘other-referencing,’’ which re-favorable attitudes would result.

In sum, although members of individualist cultures may fers to the process by which an individual is explicitlyencouraged to consider significant others (operationalizedhave a differential ability to process ego-focused versus

other-focused emotional appeals, they may process the as close friends or family members) when processinginformation. Previous research has shown that the refer-appeals in a relatively automatic manner due to higher

levels of familiarity with such stimuli. In contrast, when encing of a close other, about whom a well-articulatedmemory structure has been developed, results in outcomethey encounter a positively valenced appeal that is novel

(e.g., based on other-focused emotions) , they may be- patterns similar to those found with self-referencing(Bower and Gilligan 1979). Thus, other-referencingcome motivated to engage in a particular type of elabora-

tive processing, thereby generating novel ( i.e., collective) should lead to an increase in collective thoughts for mem-bers of individualist cultures and an enhanced persuasiveand positively valenced thoughts and more favorable atti-

tudes. Similarly, for members of collectivist cultures, the effect of other-focused emotions.In sum, if members of collectivist cultures elaborate onpresentation of a positively valenced appeal that is novel

may capture attention and result in the activation of rela- individual thoughts when exposed to ego-focused versusother-focused emotional appeals, explicitly encouragingtively inaccessible but available knowledge structures

representing the self-as-distinct-from-others ( i.e., individ- self-referencing should enhance the effects found in ex-periment 1. Collectivists should have more favorable atti-ual thoughts) . Increased elaboration on this novel, posi-

tively valenced stimulus should lead to more favorable tudes when viewing ego-focused emotional appeals thatinduce self-referencing (as opposed to other-referenc-attitudes. Experiment 2 was conducted to determine if

this mechanism underlies the results of experiment 1. ing), while no such effect should result for membersof individualist cultures who should automatically self-reference across conditions. Similarly, members of indi-Examining the Psychological Mechanism:vidualist cultures should have more favorable attitudesSelf- versus Other-Referencingwhen viewing other-focused emotional appeals that in-duce other-referencing versus self-referencing, while noTo examine the extent to which elaboration on novel

thoughts explains the results found in experiment 1, we such effect should result for members of collectivist cul-tures who are likely to automatically other-referenceintroduce the variable, referencing type (self- vs. other-

referencing), and examine its moderating effect on across conditions.Further, given the novel thought type mechanism, atti-thought type. Self-referencing refers to the process of

relating incoming information to aspects of the self-struc- tudes toward the emotional appeal should be mediatedby thought type. Therefore, the set of results involvingture stored in memory, such as internal traits or personal

experiences (Burnkrant and Unnava 1995; Meyers-Levy cognitive responses found in experiment 1 should be rep-licated. In addition, referencing type should moderate thisand Peracchio 1996). Manipulations such as the use of

second-person verbal information (‘‘You’’) in advertis- process such that attitudes toward an ego-focused emo-tional appeal that induce self-referencing should be medi-ing copy is enough to prompt significant self-referencing

(Debevec and Romeo 1992). This manipulation makes ated by individual thoughts for members of collectivistcultures. In contrast, attitudes toward an other-focusedthe self-structure salient, increasing the likelihood that

individuals will relate the persuasion information to their emotional appeal that induces other-referencing shouldbe mediated by collective thoughts for members of indi-self, and leads to increased elaboration and recall of the

stimulus (Bower and Gilligan 1979). vidualist cultures.Results of experiment 1 indicate that the persuasive H3a: Members of individualist cultures will have

effect of ego-focused versus other-focused emotional ap- more favorable attitudes when exposed topeals on members of collectivist cultures is accompanied other-focused emotional appeals that induceby an increase in individual thoughts. If elaboration on other-referencing, as opposed to self-referenc-these individual thoughts accounts for the increase in fa- ing.vorable attitudes of collectivist cultures when exposed

H3b: Members of collectivist cultures will haveto ego-focused emotional appeals (which are positivelymore favorable attitudes when exposed to ego-valenced), directly encouraging them to consider them-focused emotional appeals that induce self-selves as unique individuals, via self-referencing, shouldreferencing, as opposed to other-referencing.further prime these relatively inaccessible knowledge

structures and enhance the effects found in experiment 1. H4a: For members of individualist cultures, the re-



lationship between other-focused emotional focus (Chiasson, Dube, and Blondin 1996). In fact, basedon data from 15 cultures, Matsumoto (1989) found thatappeals, which induce other-referencing ver-

sus self-referencing, and attitudes will be me- people with more independent versus interdependentselves were more likely to correctly identify emotionaldiated by collective thoughts.expressions of happiness. For people with interdependent

H4b: For members of collectivist cultures, the rela- selves, positive emotional expressions tend to be used astionship between ego-focused emotional ap- public actions in the service of maintaining interpersonalpeals that induce self-referencing versus harmony and, therefore, are not regarded as particularlyother-referencing and attitudes will be medi- diagnostic of inner feelings of happiness. Peacefulnessated by individual thoughts. was chosen as the basis for the other-focused emotional

appeal because it fosters relational, social emotions ofEXPERIMENT 2 modesty, harmony, and restraint (Triandis 1994). Fur-

ther, collectivist participants report experiencing emo-Overview tions related to peacefulness more frequently than individ-ualist participants (Kitayama et al. 1991). Second, bothThe objective of experiment 2 is to assess the robustnesshappiness and peacefulness have been used in advertisingof the results of experiment 1 and examine the psychologi-appeals in collectivist as well as individualist culturescal mechanism underlying the unexpected effects. Conse-(Hong, Muderrisoglu, and Zinkhan 1987; Mueller 1987).quently, three changes were made from experiment 1. First,

the moderating effect of referencing type on the persuasiveeffect of emotional appeal type on attitudes of members of Pilot Experimentthe two cultures is examined. Thus, experiment 2 relies ona between-subjects design: 2 (emotional appeal type: ego- Overview. The focus of experiment 2 is to examinefocused vs. other-focused) 1 2 (cultural orientation: indi- the effect of self- and other-referencing on the elaborativevidualism vs. collectivism) 1 2 (referencing type: self- effects found in response to the emotional advertisingreferencing vs. other-referencing). appeals used in experiment 1. However, implicit in exper-

Second, to ensure generalizability, experiment 2 relies iment 2 is also an attempt to determine whether emotionalon another advertisement type (involving a visual picture appeal types can be made malleable by crossing emotionalin the appeal) , another product category (film), and an- appeal type with referencing type. For example, whileother set of ego-focused and other-focused emotions. Film pride embedded in appeals has traditionally been treatedwas selected as the product category based on the similar as ego-focused, can an appeal that focuses on being proudcriteria outlined in experiment 1. First, both ego-focused of close others (e.g., family and friends) be characterizedand other-focused emotional advertisements could be cre- as other-focused? To address this question, a pilot experi-ated for film. Second, film tends to be a relevant product ment was conducted where experiment 1 was replicatedfor participants in both cultures.6 Third, the referencing with one important change: the referent of focus in thetype manipulation can be enhanced by highlighting photo- pride and empathy appeals was manipulated to be other-graphs of an individual versus a collective group. Finally, focused and ego-focused, respectively. If emotional ap-film tends to be a relatively more utilitarian product cate- peal type varies with a change in the referent of focus,gory than beer (Ratchford 1987, app.) . Since advertise- the pattern of results found in experiment 1 should bements for self-expressive product categories (e.g., beer) eliminated. Further, such results would provide justifica-may vary more than utilitarian product categories across tion for manipulating referencing type to alter the focuscultures, noise accounted for by possible cultural variation of the emotion depicted in experiment 2 stimuli.in product category advertising appeals may be reduced.

Method. Participants were undergraduates recruitedHappiness and peacefulness were selected as the basisfrom an American university (n Å 30; mean age Å 20.12,for the ego-focused emotion and other-focused emotionalfemale Å 60 percent, all were born and raised in theappeals, respectively, for two reasons. First, Markus andUnited States, 87 percent were Caucasian) and from aKitayama (1994) suggest they are strong examples of theChinese university (n Å 30; mean age Å 19.73, femalefocal emotion types. Happiness tends to involve one’sÅ 50 percent, all were born and raised in Shanghai, 100(vs. others’) internal feelings of pleasure as a referent ofpercent were native Chinese) to participate in a new brandstudy for $5 and 10 rmb, respectively. All aspects of theprocedure remained the same with the exception that the6As in experiment 1, participants were asked to complete a set ofreferent of focus in the pride versus empathy appealsproduct familiarity and usage questions. Approximately 89 percent ofchanged. In the pride appeal, participants read, ‘‘Yourthe Chinese participants and 84 percent of the American participants

use film. Both Chinese and American participants use film approxi- friend just aced an exam. Your sister just won a race. Theymately once a month (modal response) . In addition, Chinese partici- receive deserved recognition. Ohio Flag Beer. Celebratingpants’ self-reported knowledge of the film category was similar to that their accomplishments.’’ In the empathy appeal, partici-of American participants (XV collectivist Å 2.48, XV individualist Å 2.86; F õ 1),

pants read, ‘‘Reminiscing with old pictures. Enjoyingsuggesting that film is a relatively familiar and relevant product categoryacross cultures. time alone conjuring up memories. Relaxing near the fire,



smiling. Ohio Flag beer. Celebrating you—who matters First, participants read an advertisement for Watson ColorFilm that depicted happiness or peacefulness and con-most.’’tained three positively valenced attributes pretested to be

Results. The pilot analysis involved a 2 (emotional important across cultures (high quality of color, sharpappeal type) 1 2 (cultural orientation) between-subjects texture of photograph, inexpensive) . Then, participantsANOVA. Unless otherwise specified, df Å 1, 57. provided their thoughts and attitudes about the advertise-

The results of a two-way ANOVA indicated that the ment and the new brand. After a 15-minute unrelated filleremotional appeal manipulation worked as expected. Parti- task (same as in experiment 1) , participants completed acipants rated the empathy appeal (XV Å 5.08) higher than set of questions that included demographic information,the pride appeal (XV Å 4.04) on empathy-related emotions product usage information, manipulation checks, and(Cronbach’s a Å .87; F Å 5.00, p õ .03) . Similarly, Singelis’s (1994) independent-interdependent scale.participants rated the pride appeal (XV Å 5.92) higher thanthe empathy appeal (XV Å 4.83) on pride-related emotions Independent Variables(Cronbach’s a Å .90; F Å 3.85, p õ .05) . Moreover,the advertising appeals were generally perceived as being Cultural Orientation. Collectivism-individualism waseither ego-focused or other-focused as intended with the operationalized in the same way as in experiment 1. Simi-new manipulations. Specifically, when asked to rate the larly, the original questionnaire was drafted in English,extent to which these photographs made them focus on translated into Chinese, and back-translated using externaltheir own self (‘‘Rate the extent to which the photograph translators (Hui and Triandis 1985).made you think about yourself /your personal experi-

Emotional Appeal Type. Appeals that depicted happi-ences’’; r Å .86; Burnkrant and Unnava 1995) or othersness and peacefulness were created by selecting photo-(‘‘Rate the extent to which the photograph made yougraphs that were pretested to depict the emotions. In athink about family or friends/experiences with family orpretest, 25 participants (n Å 12 American participants, nfriends’’; r Å .83) , participants rated the empathy appealÅ 13 Chinese participants) were asked to indicate how(XV Å 4.60) higher than the pride appeal (XV Å 3.19; Fmuch the following emotions describe 20 photographs,Å 8.36, p õ .01) on the self-referencing measures. Fur-by circling one number for each word (1 Å not at all, 7ther, they rated the pride appeal (XV Å 4.90) higher thanÅ very strongly; 10 daytime photographs, 10 dusk-timethe empathy appeal (XV Å 3.33; F Å 9.44, p õ .01) onphotographs) . Drawing from Frijda (1970) and Edell andthe other-referencing measures. No other significant ef-Burke (1987), this list included three happiness-relatedfects were found.emotions (happy, cheerful, delighted; Cronbach’s aAs predicted, an ANOVA on attitude toward the adver-Å .89) , three peacefulness-related emotions (calm,tisement (Cronbach’s a Å .95) yielded no significantpeaceful, serene; Cronbach’s a Å .93) , and 20 filler emo-effects. That is, in contrast to the results in experimenttions. Of the 20 photographs, the four that most strongly1, members of the individualist culture did not have moredepicted one emotion over the other were selected (allfavorable attitudes when exposed to the empathy appealp’s õ .05) . In the happiness condition, the photographwhen it was self-referencing (XV Å 3.96) than when ex-featured a beach scene at midday (12:00 P.M.) ; in theposed to the pride appeal when it was other-referencingpeaceful condition, the photograph featured the same(XV Å 3.48; F õ 1). Further, members of the collectivistbeach at sunset (5:30 P.M.) . The photographs were equallyculture did not have more favorable attitudes when ex-well liked (F õ 1), and no cultural differences wereposed to the pride appeal when it was other-referencingfound (F õ 1). Participants also rated the brand name(XV Å 3.65) than when exposed to the empathy appeal‘‘Watson’’ on the happiness-related and peacefulness-re-when it was self-referencing (XV Å 4.08; F õ 1). Thelated emotions; it depicted equal levels of happiness andidentical pattern occurred for attitude toward the brand.peacefulness, and no cultural differences were found (F’sIn sum, these results suggest that emotional appeal typeõ 1).can be altered by varying the referent of focus and provide

This visual manipulation of emotional appeal type wassupport for the use of the self-referencing and other-refer-enhanced by two verbal manipulations. First, the bannerencing manipulations in experiment 2.at the top of each advertisement read ‘‘Capture the Happi-ness (Peacefulness) .’’ Second, the advertising copy read,

Method ‘‘The sound of the surf crashing on the beach, the happycheers of children running in the water, and the bright

Undergraduate students were recruited from an Ameri- light of the sun shining overhead,’’ or ‘‘The sound ofcan university (n Å 72; mean age Å 22.15, female Å 50 waves gently lapping onto the shore, the soft cries ofpercent, all were born and raised in the United States, 81 birds flying overhead, and the quiet beauty of the sunpercent were Caucasian) and a Chinese university (n setting over the water.’’Å 79; mean age Å 19.82, female Å 43 percent, all wereborn and raised in Shanghai, 100 percent were native Referencing Type. To encourage self- or other-refer-

encing, the photos featured an individual alone (see Fig.Chinese) to take part in a study evaluating a new productand its advertisement for $5 and 10 rmb, respectively. 2) versus several people together (who appeared to be


FIGURE 2

EXAMPLE OF STIMULI: HAPPINESS, SELF-REFERENCING, U.S. SUBJECTS



FIGURE 3

EXAMPLE OF STIMULI: PEACEFULNESS, OTHER-REFERENCING, CHINESE SUBJECTS

close friends or family; see Fig. 3) . To avoid introducing encing manipulation was enhanced through two changesin the advertising copy: (1) at the beginning of the adver-confounds, the same photographs were used for both the

Chinese and American participants. Therefore, the photo- tisement, the copy read, ‘‘Remember the day by yourself(with your family and friends) at the beach’’; and (2) atgraphs were taken so that the faces of the individuals

were not visible, making it impossible to determine their the end of the advertisement, the tag line read, ‘‘CaptureYour (The) Feelings (of Friendship) .’’ethnicity. Finally, the self-referencing versus other-refer-



The pretest sample described above rated the extent tural orientation interaction involving the self-referencingcheck (F Å 3.56, p õ .07); individualist participants (XVto which these photographs inspired self-referencing (r

Å .83) or other-referencing (r Å .87) . The four photo- Å 4.76) reported higher ratings on the self-referencingcheck than did collectivist participants (XV Å 3.73), whichgraphs selected significantly prompted self- or other-ref-

erencing as intended (all p’s õ .05) . is consistent with individualism-collectivism literature onself-construal differences (Triandis 1989).

Finally, the results of an ANOVA with cultural orienta-Results tion as the dependent variable showed only a significantmain effect: American participants (XV Å .45) receivedThe hypotheses were tested based on a 2 (emotionalhigher independent scores than did Chinese participantsappeal type) 1 2 (cultural orientation) 1 2 (referencing(XV Å 0.02; F Å 2.92, p õ .05) . No other significanttype) between-subjects ANOVA. Three covariates wereeffects were found.used (separately) in the analysis reported below: product

category usage, frequency of usage, and self-rated cate- Attitudes. An ANOVA on the advertisement attitudegory knowledge. Only product category usage was sig- index yielded no significant main effects. However, anificant (for both AAD and AB) but did not interfere with significant interaction between emotional appeal type andthe tests of the hypotheses and will not be discussed fur- cultural orientation was found (F Å 11.83, p õ .01) .ther. Unless otherwise specified, df Å 1, 145. A planned comparison indicated that advertising appeals

relying on ego-focused (XV Å 5.06) versus other-focusedManipulation Checks. The results of the three-way (XV Å 4.10) emotional appeals were more effective forANOVA indicated that the emotional appeal worked as collectivists (F Å 8.88, p õ .01) . Further, other-focusedexpected. When asked to indicate how much the follow- (XV Å 5.00) versus ego-focused (XV Å 4.34) emotionaling emotions describe the appeal, participants rated the appeals were more effective for individualists (F Å 4.69,peacefulness appeal (XV Å 4.75) higher than the happiness põ .05) . This pattern of results suggests that the findingsappeal (XV Å 3.68) on the peacefulness-related emotions of experiment 1 relying on pride and empathy appeals(Cronbach’s a Å .93; F Å 6.90, p õ .01) . Similarly, generalize to other ego-focused and other-focused emo-they rated the happiness appeal (XV Å 5.16) higher than tional appeals.the peacefulness appeal (XV Å 3.56) on happiness-related More important, the two-way interaction was qualifiedemotions (Cronbach’s a Å .86; F Å 30.58, p õ .05) . by a significant three-way interaction (F Å 5.33, pIn addition, however, a marginally significant emotional õ .05) . The pattern of results is consistent with Hypothe-appeal type 1 cultural orientation effect occurred (F sis 3a. Members of individualist cultures had more favor-Å 3.64, p õ .06) , whereby collectivist participants (XV able attitudes when exposed to other-focused emotionalÅ 3.11) versus individualist participants (XV Å 4.00) re- appeals that induce other-referencing (XV Å 5.77) thanported lower ratings on happiness-related emotions for the self-referencing (XV Å 4.27; F Å 11.49, p õ .01) . How-peacefulness appeal. However, since equally high ratings ever, Hypothesis 3b was not supported. Members of col-were found on the happiness-related emotions for the hap- lectivist cultures did not have more favorable attitudespiness appeal across the cultures, this significant differ- when exposed to ego-focused emotional appeals that in-ence should not affect the results of the experiment. No duce self-referencing (XV Å 5.10) versus other-referencingother significant effects were found. (XV Å 5.03; F õ 1). This result will be discussed below.Suggesting that the emotional appeals were perceived In addition, members of the individualist culture hadas ego-focused or other-focused as intended, a significant more favorable attitudes when exposed to an other-fo-main effect of emotional appeal was found for both indi- cused emotional appeal that encouraged other-referencingvidual thoughts (F Å 5.72, p õ .02) and collective (XV Å 5.77) than to an ego-focused emotional appeal thatthoughts (F Å 3.51, p Å .05) . The happiness appeal encouraged self-referencing (XV Å 4.47; F Å 9.30, pled to more individual thoughts (XV Å .94) than did the õ .01) . Further, members of the collectivist culture hadpeacefulness appeal (XV Å .60) , while the peacefulness more favorable attitudes when exposed to an ego-focusedappeal led to more collective thoughts (XV Å .94) than did emotional appeal that induced self-referencing (XV Å 5.10)the happiness appeal (XV Å .40) . than to an other-focused emotional advertisement that in-Further, the manipulation check for referencing type duced other-referencing (XV Å 4.06; F Å 5.43, p õ .05) .worked as expected. Participants reported higher ratings As in experiment 1, the cultural orientation and emo-on the self-referencing checks (r Å .87) when exposed tional-appeal-type interaction on the brand attitude indexto the self-referencing appeal (XV Å 4.25) than the other- was significant (F Å 6.84, p õ .05) , and the pattern ofreferencing appeal (XV Å 3.59; F Å 3.75, p õ .05) . Simi- means paralleled those found with advertisement atti-larly, a marginally significant effect occurred for other- tudes. This two-way interaction was qualified by a mar-referencing, where participants reported higher ratings on ginally significant three-way interaction (F Å 2.77, pthe other-referencing checks (r Å .81) when exposed to õ .10) , again reflecting the same pattern of means asthe other-referencing appeal (XV Å 4.26) versus the self- found in the advertisement attitude index (see Table 1).referencing appeal (XV Å 3.71; F Å 3.30, p õ .07) . Theonly other significant effect was a referencing type 1 cul- Cognitive Responses. The same coding scheme used



TABLE 1

EXPERIMENT 2 DEPENDENT MEASURES AS A FUNCTION OF EMOTIONAL APPEAL TYPE,CULTURAL ORIENTATION, AND REFERENCING TYPE

Ego-focused emotional appeal (happiness) Other-focused emotional appeal (peacefulness)

Self-referencing Other-referencing Self-referencing Other-referencing

Collectivist Individualist Collectivist Individualist Collectivist Individualist Collectivist IndividualistDependent measures culture culture culture culture culture culture culture culture

Attitude toward the ad (AAD) 5.03 4.47 5.10 4.17 4.14 4.27 4.07 5.77Attitude toward brand (AB) 4.60 4.72 4.63 3.98 3.93 4.50 4.07 5.33Individual thoughts 1.20 1.21 .80 .42 .36 .95 .35 .74Collective thoughts .65 .05 .50 .35 .68 .15 .65 1.11Total thoughts 4.25 3.63 3.65 4.85 4.42 3.45 3.70 4.47

NOTE.—Higher means indicate more positive attitudes and greater numbers of self-related, other-related, and total thoughts. Cell sizes range from n Å 33 ton Å 40.

in experiment 1 was used here. Interrater agreement was .54, p õ .01; bA(B) Å .46, p õ .01) . Finally, in partialsupport of analysis d , collective thoughts were marginallyhigh (94 percent agreement by Chinese coders, 89 percent

by American coders); discrepancies were resolved significant (bA(AD) Å .34, p õ .07; bA(B) Å .39, p õ .02) ,while other-referencing was significant for AAD (bA(AD)through discussion.Å 1.18, p õ .01) but not for AB (bA(B) Å .46, p Å .26)Collective Thoughts. To test the extent to which novelin a model that included both emotional appeal type andthoughts mediated attitudes by members of the two cul-thought type. These results provide support for the prem-tures, two sets of mediation tests were conducted. First,ise that the relationship between other-focused emotionalwe examined the extent to which the relationship betweenappeals that induce other-referencing and attitudes mayother-focused emotional appeals and attitudes is mediatedbe mediated by collective thoughts for members of indi-by collective thoughts for members of an individualistvidualist cultures.culture, as found in experiment 1. Consistent with analysis

a , individualist participants had marginally more favor- Individual Thoughts. Two sets of mediation tests alsowere conducted with individual thoughts. First, we exam-able attitudes when exposed to an other-focused versus

ego-focused emotional appeal (bA(AD) Å .33, p õ .07; ined the extent to which the relationship between ego-focused emotional appeals and attitudes is mediated bybA(B) Å .25, p õ .10); consistent with analysis b , other-

focused versus ego-focused emotional appeals led to mar- individual thoughts by members of a collective culture,as found in experiment 1. In support of analysis a , collec-ginally more collective thoughts for individualists (bA(AD)

Å .22, p õ .06; bA(B) Å .22, p õ .06); consistent with tivist respondents had marginally more favorable attitudeswhen exposed to an ego-focused versus other-focusedanalysis c , collective thoughts led to more favorable atti-

tudes for individualists (b (A)AD Å .54, p õ .01; bA(B) emotional appeal (bA(AD) Å .96, p õ .01; bA(B) Å .61, põ .06) . Consistent with analysis b , ego-focused versusÅ .46, põ .01); and consistent with analysis d , collective

thoughts were significant (bA(AD) Å .49, p õ .01; bA(B) other-focused emotional appeals led to more individualthoughts for collectivists (bA(AD) Å .64, p õ .01; bA(B)Å .43, p õ .01) , but other-focused emotional appeals

were not significant (bA(AD) Å .22, p õ .21; b (B) Å .16, Å .64, p õ .01) . Consistent with analysis c , individualthoughts led to more favorable attitudes for collectivistsp õ .29) in a model that included both emotional appeal

type and thought type. (bA(AD) Å .62, p õ .01; bA(B) Å .50, p õ .02) . Finally,in partial support of analysis d , individual thoughts wereSecond, we tested Hypothesis 4a, which suggests that

the relationship between other-focused emotional appeals marginally significant (bA(AD) Å .43, põ .07; bA(B)Å .41,põ .08) , while ego-focused emotional appeals were mar-that induce other-referencing and attitudes is mediated by

collective thoughts for members of individualist cultures. ginally significant for AAD ( Å .68, p õ .07) but not forbA(B) ( Å .35, p õ .32) in a model that included bothIn support of analysis a , individualist respondents had

more favorable attitudes when exposed to an other-fo- emotional appeal type and thought type.Second, we examined Hypothesis 4b, which suggestscused emotional appeal that induced other-referencing

(bA(AD)Å 1.50, põ .01; bA(B)Å .83, põ .04) . Consistent that the relationship between ego-focused emotional ap-peals that induce self-referencing and attitudes is medi-with analysis b , other-focused emotional appeals that in-

duce other-referencing led to more collective thoughts for ated by individual thoughts for members of collectivistcultures. Counter to analysis a , collectivist respondentsindividualists (bA(AD) Å .96, p õ .02; bA(B) Å .96, p õ

.02) . Consistent with analysis c , collective thoughts led did not have more favorable attitudes when exposed toan ego-focused emotional appeal that induced self-refer-to more favorable attitudes for individualists (bA(AD) Å



TABLE 2

EXPERIMENT 2 TESTS OF THE MEDIATING EFFECTS OF THOUGHT TYPE ON ATTITUDES

Individualist culture Collectivist culture(collective thoughts) (individual thoughts)

Dependent variable Independent variable coefficient coefficient

Emotional appeals on AAD:AAD Emotional appeal .33/ .96**Thoughts Emotional appeal .22/ .64**AAD Thoughts .54** .62**AAD Emotional appeal, thoughts .22, .49** .68/, .43/

Emotional appeals on AB:AB Emotional appeal .25/ .61/

Thoughts Emotional appeal .22/ .64**AB Thoughts .46** .50**AB Emotional appeal, thoughts .16, .43** .35, .41/

Referencing type on AAD:AAD Referencing type 1.50** .06Thoughts Referencing type .96* .40AAD Thoughts .52** .88**AAD Referencing type, thoughts 1.18**, .34/ .44, .93**

Referencing type on AB:AB Referencing type .83* .03Thoughts Referencing type .96* .40AB Thoughts .46** .80**AB Referencing type, thoughts .46, .39* .37, .85**

NOTE.—A difference between means t-test was conducted to examine the decrease in beta coefficients in the tests of mediation (all p’s õ .01) ./p õ .10.*p õ .05.**p õ .01.

encing (bA(AD) Å .06, p õ .89; bA(B) Å .03, p õ .94) . vides further support for the finding that other-focusedversus ego-focused emotional appeals are more persua-Counter to analysis b , ego-focused emotional appeals that

induced self-referencing, as opposed to other-referencing, sive for members of an individualist culture, while ego-focused versus other-focused emotional appeals are moredid not lead to more individual thoughts for collectivists

(bA(AD) Å .40, p õ .15; bA(B) Å .40, p õ .15) . Consistent persuasive for members of a collectivist culture. Themechanism driving these effects appears to be the genera-with analysis c , individual thoughts led to more favorable

attitudes for collectivists (bA(AD) Å .88, p õ .01; bA(B) tion of and elaboration on a novel thought type resultingfrom the stimulus: more favorable attitudes resulted forÅ .80, põ .01) . Finally, in support of analysis d , individ-

ual thoughts were significant (bA(AD)Å .93, põ .01; bA(B) other-focused emotional appeals by members of individu-alist cultures when other-referencing (vs. self-referenc-Å .85, põ .01) , while self-referencing was not significanting) and collective thoughts appeared to mediate the rela-(bA(AD) Å .44, p õ .29; bA(B) Å .37, p õ .39) whentionship between advertising appeals and attitudes.predicting attitudes in a model that included both emo-

However, no support was found for Hypothesis 3b,tional appeal type and thought type.which suggested that attitudes will be more favorable forIn sum, although the results provide support for theego-focused emotional appeals when members of collec-premise that the relationship between ego-focused emo-tivist cultures are induced to self-reference, or Hypothesistional appeals and attitudes is mediated by individual4b, which suggested that the relationship between ego-thoughts for collectivists, they do not provide support forfocused emotional appeals that induce self-referencingHypothesis 4b (see Table 2). Although the results indi-and attitudes is mediated by individual thoughts for mem-cate that individual thoughts lead to more favorable atti-bers of collectivist cultures. An explanation for these re-tudes for collectivists, the missing link is the ability ofsults may lie in the varying meaning of the ‘‘self ’’ acrossself-referencing to generate those individual thoughts, acultures, and thus what it means for members of collectiv-finding that is discussed below.ist cultures to ‘‘self-reference.’’

Although individualists consider the self to be whoDiscussionthey are (i.e., their internal feelings and attitudes, distinct

The objective of experiment 2 was to assess the ro- from those of others) , collectivists are more likely tobustness and generalizability of the results found in exper- consider the self to include their family and close friendsiment 1 and to examine the mechanism underlying these (i.e., in-group’s feelings and attitudes; Triandis 1989).

Consequently, when induced to self-reference, individual-effects. The pattern of results found in experiment 2 pro-



ists may think of themselves as unique and distinct from generate thoughts organized around ‘‘we’’ and ‘‘us’’when asked to self-reference.others, while collectivists may think of themselves in rela-

tion to close others. To examine this possibility, we draw on Triandis (1989)who distinguishes among ‘‘allocentric’’ thoughts (those re-In support, when members of the two cultures were

asked to free-associate to the word ‘‘me,’’ the most fre- lating to the interaction or interrelationship of one’s ownself and close others), and ‘‘idiocentric’’ thoughts (thosequent association for American participants was ‘‘I, per-

son, individual,’’ but for Korean participants it was ‘‘fam- relating to only the self as an individual) and ‘‘group’’thoughts (or collective thoughts; those relating to only closeily, love’’ (Maday and Szalay 1976). Further, when asked

to list important personal goals, Indian participants in- others).7 Then, we examine the possibility that ego-focusedversus other-focused emotions evoke more idiocentricclude their own good health and the good health of their

families, while American participants are less likely to thoughts for members of collectivist cultures (but not indi-vidualists), while self-referencing versus other-referencingmention the latter (Roland 1984). Finally, Sampson

(1988) argues that ‘‘although all cultures draw a line evokes more allocentric thoughts for members of collectiv-ist cultures (but not individualist cultures).between a region defined as belonging intrinsically to the

self and a region defined as extrinsic or outside the self The two sets of raters recategorized participants’ thoughtsbased on the new coding scheme. Operationally, the distinc-and hence belonging to the ‘nonself /other,’ where that

line is drawn varies extensively. There are cultures in tion among the three codes was the subject of the thought(McGuire and McGuire 1986). If the subject was the selfwhich firmly drawn boundaries marking sharp self /non-

self separations define the culture’s indigenous psychol- (i.e., any thought that mentioned only the distinct self orused ‘‘me’’ or ‘‘I’’), the thought was coded ‘‘idiocentric.’’ogy, and others in which such firm boundaries do not

exist. According to the indigenous psychology currently If the subject was close others (i.e., any thought that men-tioned only family members or friends or used ‘‘they’’ ordominant in the United States, for example, there exists

a region intrinsic to the person and a region of ‘other.’ ‘‘them’’), the thought was coded ‘‘collective.’’ If the sub-ject was both the self and others (i.e., any thought that. . . At the other extreme, we find cultures in which the

self /nonself boundary is less sharply drawn and more mentioned family members or friends or used ‘‘us,’’ ‘‘we,’’‘‘ours’’), the thought was coded ‘‘allocentric.’’ See Tablefluid’’ (p. 15).

This stream of research suggests that the self-structure 3 for examples of thoughts. Interrater agreement was high(91 percent agreement by Chinese coders, 85 percent byis more organized around ‘‘we,’’ ‘‘our,’’ and ‘‘us’’ in

collectivist versus individualist cultures. If true, self- and American coders), and discrepancies were resolved throughdiscussion.other-referencing manipulations may have similar effects

for members of collectivist cultures, but different effects The results of an ANOVA on idiocentric thoughtsyielded a significant main effect for cultural orientationfor members of individualist cultures. Indeed, this expla-

nation is consistent with the asymmetric pattern of results (F Å 14.80, p õ .01) , where more idiocentric thoughtswere generated by members of individualist (XV Å .79)whereby self- and other-referencing had similar effects

on the attitudes and thoughts for members of collectivist than collectivist cultures (XV Å .34) . In addition, a signifi-cant emotional appeal type 1 cultural orientation interac-cultures. For example, participants in the collectivist (but

not individualist) culture increased their individual tion was found (FÅ 9.59, p õ .01) , where appeals thatwere ego-focused (XV Å .58) versus other-focused (XVthoughts when they viewed ego-focused versus other-fo-

cused emotional advertisements, but the same pattern did Å .10) generated more idiocentric thoughts for membersof collectivist cultures but not for members of individual-not result when they were induced to self-reference, rather

than other-reference. An additional reanalysis of the ist cultures (F õ 1). No other effects were significant.The results of an ANOVA on allocentric thoughts alsothought protocols was conducted to examine why this

pattern of results may have occurred. yielded a significant main effect for cultural orientation(F Å 14.31, põ .01) , whereby more allocentric thoughtswere generated by members of collectivist (XV Å .79) thanExamining the Distinction between Idiocentricindividualist cultures (XV Å .34) . This main effect wasand Allocentric Thoughtsqualified by a significant referencing type 1 cultural ori-entation interaction (F Å 4.35, p õ .05) , whereby self-The research reviewed above suggests that the emo-

tional appeal type and referencing type manipulations do referencing (XV Å .59) versus other-referencing (XV Å .35)not tap the same underlying construct for collectivists.Although ego-focused emotional appeals may be a rela-tively pure way to highlight the self to the exclusion 7Conceptually, the main difference between (1) individual and idio-of others (self-as-distinct-from-others) , self-referencing centric thoughts and (2) collective and group thoughts is that idiocentric

and group thoughts are mutually exclusive. In this research, thoughtsmay tap the larger self-structure that for collectivists con-that pertained to the interrelationship between the self and close otherssists of elements linking the self to important others (self-were coded as both individual and collective. In the new coding scheme,in-relation-to-others) . Thus, collectivists may generate such thoughts are coded as allocentric. To avoid confusion, we rely on

individual thoughts organized around ‘‘me’’ and ‘‘I’’ Triandis’s terminology, now calling individual thoughts ‘‘idiocentric’’and add the new thought type ‘‘allocentric.’’when exposed to an ego-focused emotional appeal but



TABLE 3

EXAMPLES OF THOUGHT-CODING SCHEME

Individual/idiocentric thoughts Collective/group thoughts Allocentric thoughts

Individualist respondents ‘‘I remembered how much I love ‘‘The pictures of the families ‘‘I think of memories with old friendsthe ocean and how I am and kids playing in the at the beach.’’always saying I’ll go more water were great.’’often but never do.’’

‘‘I imagined the sound of waves, ‘‘Contemplating the cheers ‘‘I wonder when we are going tobirds, and the smell of the of children, how much fun play volleyball again—at theocean and wished I was there they were having.’’ beach.’’right now.’’

Collectivist respondents ‘‘There is very beautiful scenery, ‘‘It is necessary and ‘‘My initial feeling was the times II hope I can have a chance to wonderful to take the time went to the beach with mybe at a spot like this.’’ relaxing when with the boyfriend and enjoyed the

one you love.’’ peaceful atmosphere of theocean.’’

‘‘The picture reminds me of ‘‘My family would like me to ‘‘It reminds me about those dayssome very memorable scenes take a leisure time to being together with my friends,in my life.’’ relieve myself from hard makes me recall with nostalgia.’’

work.’’

generated more allocentric thoughts for members of col- emotional appeals highlight the limited self-structure be-cause the referent of focus is self-as-distinct-from-others.lectivist but not for individualist cultures (F õ 1). Only

one other effect was significant: emotional appeal type Consistent with this interpretation, Kitayama et al. (1991)suggest that across cultures, ego-focused emotions tend1 referencing type (F Å 6.42, p õ .01) , whereby ego-

focused (vs. other-focused) emotional appeals generated to highlight those unique aspects of the self not contingenton others, such as one’s individual desires, motives, abili-more allocentric thoughts in conditions of self-referencing

but not other-referencing. Though this effect was not pre- ties, rights, or internal attributes (e.g., individual accom-plishment, which invokes pride, or the abridgment of andicted, it is consistent with the pattern found above, driven

by collectivist participants. individual right, which invokes anger) . In this light, self-referencing may keep the self /nonself boundary blurredIn sum, although the original cognitive response results

suggest that the referencing type manipulations had iden- for members of collectivist cultures, while ego-focusedemotional appeals may sharpen it.tical effects for members of collectivist cultures, this re-

coding of the thoughts indicates otherwise. Though mem-bers of collectivist cultures do not generate purelyindividual thoughts in response to the self-referencing GENERAL DISCUSSIONmanipulation, self-referencing does encourage them togenerate more thoughts of which they are the subject. The objective of this research was to examine the per-

suasive effect of emotional appeal types across culturesHowever, these thoughts are allocentric rather than indi-vidual in nature. Thus, self-referencing may affect collec- and explore why such appeals vary in their effectiveness

across cultures. In a set of experiments, we found thattivists in the intended direction but not in the same wayit affects individualists. In hindsight, this finding may not appeals relying on other-focused emotions (e.g., empathy,

peacefulness) versus ego-focused emotions (e.g., pride,be surprising. Triandis (1993) suggests the most im-portant cognitions are sentences that include ‘‘I,’’ ‘‘me,’’ happiness) led to more favorable attitudes for members

of the individualist culture (United States) , while appealsand ‘‘mine’’ for those with an independent self, but sen-tences that include ‘‘us,’’ ‘‘we,’’ and ‘‘ours’’ for those relying on ego-focused emotions, as opposed to other-

focused emotions, led to more favorable attitudes forwith an interdependent self. What is surprising, however,is that ego-focused emotional appeals lead to the produc- members of the collectivist culture (China) . Further, col-

lective thoughts mediated attitudes toward other-focusedtion of more individual thoughts for members of the col-lectivist culture, even though self-referencing did not. versus ego-focused emotional appeals for individualists,

while individual thoughts mediated attitudes toward ego-What is the difference between these two manipulationsfor collectivist participants? focused versus other-focused emotional appeals for col-

lectivists. In experiment 2, we investigated why theseOne difference may lie in the primary referent of focusinvolved in self-referencing relative to ego-focused emo- effects occur and found support for the notion that the

novel types of thoughts generated by the persuasion ap-tional appeals. Although the self-referencing manipula-tion highlights the larger self-structure because the refer- peals mediated attitudes, thereby driving the attitudinal

and cognitive responses results.ent of focus is self-in-relation-to-others, ego-focused



ing of specific emotions as it relates to processing andMotivation versus Ability to Processattitude formation. Previous research has suggested thatEmotional Advertising Appealsemotional responses, compared to the recognition of emo-

Previous research has focused on differential levels of tion, result in more positive attitudes toward advertise-accessibility of ego-focused versus other-focused emo- ments, as well as higher levels of advertisement and brandtions for members of collectivist versus individualist cul- recall (Stout and Leckenby 1986). However, future re-tures. Although the original premise of this research was search addressing the felt versus depicted emotion distinc-that the differences in accessibility should lead to relative tion and measuring both variables is needed to gain adifferences in the ability to recognize, experience, and more complete understanding of the role of emotions inthus be persuaded by ego-focused versus other-focused persuasion across cultures.emotional appeals in the two cultures, instead the pattern Further, the antecedents of these emotion types mayof results for both experiments suggest that a motivation be examined. For example, the frequency of specific emo-process was operating. That is, the relative novelty of tions may vary across cultures; because of the extent tothe appeals in the two cultures induced higher levels of which they are socially embedded and therefore rein-motivation to process the message. forced in a given culture, some may be repressed while

Might the current experimental situation be responsible others are encouraged. Alternatively, the frequency offor the motivationally driven results observed? The self- specific emotions may vary because some are biologicallystructure has been conceptualized as a dynamic entity, diminished, stemming from more basic, hardwired differ-such that different aspects of the self will be salient in ences across cultures. Insight into the antecedents of emo-different situations. Thus, while the interdependent self tions would allow researchers to understand the extent totends to be most dominant in collectivist cultures, mem- which certain types of emotional appeals may becomebers of such cultures can and do sample aspects of the more prevalent over time—and perhaps less persua-independent self under certain situations (Triandis 1989). sive—as cultural norms change.Encountering an ego-focused emotional appeal may beone way to prompt members of a collectivist culture toaccess aspects of the independent self. Importantly, how- Broadening the View of Emotionalever, such appeals may be most relevant to them when Advertising Appealsviewed in private rather than in the presence of others,when elements of the self-as-distinct-from-others may be The role of emotion in persuasion appeals has receivedaccessed without social risk. Similarly, other-focused ap- growing interest in consumer behavior research. Much ofpeals may be most effective when experienced privately this work has focused on developing a typology of emo-for members of an individualist culture, when elements tional responses that varies in valence and arousal. Forof the self-in-relation-to-others may be accessed without example, one emotional response is characterized byany public diminishment of the self as a unique individual. highly arousing positive emotions, another representsThis possibility suggests that the effectiveness of such milder positive emotions, and the third represents nega-appeals may vary by media type. The appeals in the cur- tive emotions (termed ‘‘warm,’’ ‘‘upbeat,’’ and ‘‘nega-rent research were print-based and viewed individually, tive’’ in Edell and Burke [1987]; Burke and Edell [1989];which may have heightened the degree to which these see also Batra and Ray [1986] for a related typology).less dominant aspects of the self were sampled. Future In this article, we suggest another important dimensionresearch may examine the effects of media type or view- that typifies emotional appeals: referent of focus. Impor-ing situations that are more public. tantly this dimension broadens the dominant Western

view of emotions beyond the typical three-factor structureand a focus on valence and arousal to examine how self-Distinction between Felt and Depictedconstrual can affect the persuasive effect of emotionalEmotionsappeals. However, future research is still needed to exam-

Throughout this research, we focused on the recogni- ine the extent to which referent of focus can be appliedtion of emotions depicted in advertisements and the differ- to the basic three-factor structure of emotions found inential effects such emotional depictions have on members persuasion appeals and, more broadly, the extent to whichof individualist versus collectivist cultures. Since depicted the three-factor structure of emotional appeals is general-and felt emotions are not always identical and the recogni- izable across cultures in order to better understand thetion of an emotion depicted in an appeal does not neces- extent to which persuasion appeals are differentially ef-sarily lead to the experience of an emotion (Stout, Homer, fective across cultures, and the specific role of emotionand Liu 1990), one limitation of the current work is the in appeals across cultures.absence of measured felt emotional responses to the dif-ferent advertising appeals. As a result, we offer little in-sight into the possible motivational effect of felt emo- [Received February 1997. Revised May 1998. Robert

E. Burnkrant served as editor, and Deborah MacInnistional responses on the target advertisement, nor the linkbetween recognition of an emotional appeal and the feel- served as associate editor for this article.]



(1986), The Emotions, New York: Cambridge Univer-REFERENCESsity Press.

and Batja Mesquita (1994), ‘‘The Social Roles andAaker, David A., Douglas M. Stayman, and Michael R. HagertyFunctions of Emotions,’’ in Emotion and Culture: Empiri-(1986), ‘‘Warmth in Advertising: Measurement, Impact,cal Studies of Mutual Influence, ed. Shinobu Kitayama andand Sequence Effects,’’ Journal of Consumer Research,Hazel Markus, Washington, DC: American Psychological12 (March), 365–381.Association, 51–88.Aaker, Jennifer and Durairaj Maheswaran (1997), ‘‘The Impact

Goodstein, Ronald C. (1993), ‘‘Category-Based Applicationsof Cultural Orientation on Persuasion,’’ Journal of Con-and Extensions in Advertising: Motivating More Extensivesumer Research, 24 (December) , 315–328.Ad Processing,’’ Journal of Consumer Research, 20Alsop, Ronald (1988), ‘‘Surreal Ads Startle—but Do They(June), 87–99.Sell?’’ Wall Street Journal (October 20).

Grunert, Klaus G. (1996), ‘‘Automatic and Strategic ProcessesArnold, Magda B. (1960), Emotion and Personality, Vol. 1,in Advertising Effects,’’ Journal of Marketing, 60, (Octo-New York: Columbia University Press.ber) , 88–101.Baron, Reuben M. and David A. Kenny (1986), ‘‘The Modera-

Han, Sang-Pil and Sharon Shavitt (1994), ‘‘Persuasion andtor-Mediator Variable Distinction in Social PsychologicalCulture: Advertising Appeals in Individualistic and Collec-Research: Conceptual, Strategic, and Statistical Considera-tivistic Societies,’’ Journal of Experimental Social Psy-tions,’’ Journal of Personality and Social Psychology, 51chology, 30 (July) , 326–350.(December) , 1173–1182.

Haugtvedt, Curtis and Richard Petty (1992), ‘‘Personality andBatra, Rajeev and Michael Ray (1986), ‘‘Affective ResponsesPersuasion: Need for Cognition Moderates the PersistenceMediating Acceptance of Advertising,’’ Journal of Con-and Resistance of Attitude Changes,’’ Journal of Personal-sumer Research, 13 (September) , 234–249.ity and Social Psychology, 63 (August) , 308–319.Berlyne, D. E. (1960), Conflict, Arousal and Curiosity, New

Hofstede, Geert (1980), Culture’s Consequences, BeverlyYork: McGraw-Hill.Hills, CA: Sage.Bower, Gordon H. and Stephen G. Gilligan (1979), ‘‘Remem-

Holbrook, Morris B. and Rajeev Batra (1987), ‘‘Assessing thebering Information Related to One’s Self,’’ Journal of Re-Role of Emotions as Mediators of Consumer Responses tosearch in Personality, 13 (December) , 420–432.Advertising,’’ Journal of Consumer Research, 14 (Decem-Brislin, Richard W. (1980), ‘‘Translation and Content Analysisber) 404–420.of Oral and Written Materials,’’ in Handbook of Cross-

and Elizabeth C. Hirschman (1982), ‘‘The ExperientialCultural Psychology, ed. Harry C. Triandis and John W.Aspects of Consumption: Consumer Fantasies, FeelingsBerry, Boston: Allyn & Bacon.and Fun,’’ Journal of Consumer Research, 9 (September) ,Burke, Marian Chapman and Julie A. Edell (1989), ‘‘The Im-132–140.pact of Feelings on Ad-Based Affect and Cognition,’’

Hong, Jae W., Aydin Muderrisoglu, and George M. ZinkhanJournal of Marketing Research, 27 (February) , 69–83.(1987), ‘‘Cultural Differences and Advertising Expres-Burnkrant, Robert E. and H. Rao Unnava (1995), ‘‘Effects ofsion: A Comparative Content Analysis of Japanese andSelf-Referencing on Persuasion,’’ Journal of ConsumerU.S. Magazine Advertising,’’ Journal of Advertising, 16Research, 22 (June), 17–26.(January), 55–68.Chiasson, Nicole, Lisa Dube, and Jean-Pierre Blondin (1996),

Hui, Harry C. and Harry C. Triandis (1985), ‘‘Measurement‘‘Happiness: A Look into the Folk Psychology of Fourin Cross-Cultural Psychology: A Review and ComparisonCultural Groups,’’ Journal of Cross-Cultural Psychology,of Strategies,’’ Journal of Cross Cultural Psychology, 1627 (November) , 673–691. (June), 131–152.Comrie, Bernard (1981), Language Universals and Linguistic Kamp, Edward and Deborah MacInnis (1995), ‘‘Characteristics

Typology, Chicago: University of Chicago Press. of Portrayed Emotions in Commercials: When Does WhatDarwin, Charles (1896), The Expression of Emotions in Man Is Shown in Ads Affect Viewers?’’ Journal of Advertising

and Animals, New York: Philosophical Library. Research, 35 (November) , 19–28.Debevec, Kathleen and Jean B. Romeo (1992), ‘‘Self-Referent Kitayama, Shinobu and Hazel Rose Markus (1995), ‘‘Culture,

Processing in Perceptions of Verbal and Visual Commer- Self and Emotion: A Cultural Perspective to ‘Self-Con-cial Information,’’ Journal of Consumer Psychology, 1 scious’ Emotions,’’ in Shame, Guilt, Embarrassment and(January), 83–102. Pride: Empirical Studies of Self-Conscious Emotions, ed.

Edell, Julie A. and Marian C. Burke (1987), ‘‘The Power of June Price Tangney and Kurt W. Fisher, New York, Guil-Feelings in Understanding Advertising Effects,’’ Journal ford.of Consumer Research, 14 (December) , 421–433. , Hazel Rose Markus, and Masaru Kurokawa (1991),

Ellsworth, Phoebe C. (1994), ‘‘Sense, Culture, and Sensibility: ‘‘Culture, Self and Emotion: The Structure and FrequencyEmpirical Studies of Mutual Influence,’’ in Emotion and of Emotional Experience,’’ paper presented at the SocietyCulture: Empirical Studies of Mutual Influence, ed. Shi- for Psychological Anthropology, Chicago.nobu Kitayama and Hazel Markus, Washington, DC: Lazarus, Richard S. (1982), ‘‘Thoughts on the Relations be-American Psychological Association, 23–50. tween Emotion and Cognition,’’ American Psychologist,

Friestad, Marian and Esther Thorson (1993), ‘‘Remembering 37 (September) , 1019–1024.Ads: The Effects of Encoding Strategies, Retrieval Cues Lin, Henry (1981), Essential Grammar for Modern Chinese,and Emotional Response,’’ Journal of Consumer Psychol- Boston: Cheng & Tsui.ogy, 2 (January), 1–23. MacInnis, Deborah and Bernard Jaworski (1989), ‘‘Informa-

Frijda, Nico (1970), ‘‘Emotion and Recognition of Emotion,’’ tion Processing from Advertisements: Toward an Integ-in Feelings and Emotions: The Loyola Symposium, ed. rative Framework,’’ Journal of Marketing, 53 (October) ,

1–23.Magda Arnold, New York: Academic Press, 241–250.



(1991), ‘‘Consumer Responses to Advertising: The Ef-, Christine Moorman, and Bernard Jaworski (1991),fects of Ad Content, Emotions and Attitude toward the Ad‘‘Enhancing and Measuring Consumers’ Motivation, Op-on Viewing Time,’’ Journal of Consumer Research, 17portunity and Ability to Process Brand Information from(March), 440–453.Ads,’’ Journal of Marketing, 55 (October) , 32–53.

Osgood, Charles E., William H. May, and Murray S. MironMaday, Bela C. and Lorand B. Szalay (1976), ‘‘Psychological(1975), Cross-Cultural Universals of Affective Meaning,Correlates of Family Socialization in the United States andChicago: University of Illinois Press.Korea,’’ in Psychological Anthropology, ed. T. Williams,

Ratchford, Brian T (1987), ‘‘New Insights about the FCBThe Hague: Mouton.Grid,’’ Journal of Advertising Research, (August) , 24–Markus, Hazel and Shinobu Kitayama (1991), ‘‘Culture and38.the Self: Implications for Cognition, Emotion and Motiva-

Roland, Alan (1984), ‘‘Psychoanalysis in Civilizational Per-tion,’’ Psychological Review, 98 (April) , 224–253.spective: The Self in India, Japan, and America,’’ Psycho-and Shinobu Kitayama (1994), ‘‘The Cultural Con-analytic Review, 71 (December) , 569–590.struction of Self and Emotion: Implications for Social Be-

Sampson, Edward E. (1988), ‘‘The Debate on Individualism:havior,’’ in Emotion and Culture: Empirical Studies ofIndigenous Psychologies of the Individual and Their RoleMutual Influence, ed. Shinobu Kitayama and Hazel Mar-in Personal and Societal Functioning,’’ American Psychol-kus, Washington, DC: American Psychological Associa-ogist, 43 (January), 15–22.tion, chap. 4.

Singelis, Ted (1994), ‘‘The Measurement of Independent and, Richard L. Moreland, and Jeanne Smith (1985), ‘‘RoleInterdependent Self-Construals,’’ Personality and Socialof the Self-Concept in the Perception of Others,’’ JournalPsychology Bulletin, 20 (October) , 580–591.of Personality and Social Psychology, 42 (December) , 38–

Smith, Craig, Kelly Haynes, Richard Lazarus, and Lois Pope50.(1993), ‘‘In Search of the Hot Cognitions: Attributions,Matsumoto, David (1989), ‘‘Cultural Influences on the Percep-Appraisals, and Their Relation to Emotion,’’ Journal oftion of Emotion,’’ Journal of Cross Cultural Psychology,Personality and Social Psychology, 65 (November) , 916–20 (March), 92–105.929., T. Kudoh, K. Scherer, and H. Wallbott (1988), ‘‘Ante-

Stout, Patricia A., Pamela M. Homer, and Scott S. Liu (1990),cedents and Reactions to Emotions in the United States‘‘Does What We See Influence How We Feel? Felt Emo-and Japan,’’ Journal of Cross Cultural Psychology, 19tions versus Depicted Emotions in Television Commer-(September) , 267–287.cials,’’ in Emotion in Advertising: Theoretical and Practi-McGuire, William J. and Claire V. McGuire (1986), ‘‘Differ-cal Explorations, ed. Stuart Agres et al., New York:ences in Conceptualizing Self versus ConceptualizingQuorum, 195–210.Other People as Manifested in Contrasting Verb Types

and John D. Leckenby (1986), ‘‘Measuring EmotionalUsed in Natural Speech,’’ Journal of Personality and So-Response to Advertising,’’ Journal of Advertising, 15 (Jan-cial Psychology, 51 (December) , 1135–1143.uary) , 35–42.Meyers-Levy, Joan and Laura A. Peracchio (1996), ‘‘Modera-

Sujan, Mita, James R. Bettman, and Hans Baumgartner (1993),tors of the Impact of Self-Reference on Persuasion,’’ Jour-‘‘Influencing Consumer Judgments Using Autobiographi-nal of Consumer Research, 22 (March), 408–423.cal Memories: A Self-Referencing Perspective,’’ JournalMorris, Michael and Kaiping Peng (1994), ‘‘Culture andof Marketing Research, 30 (November) , 422–36.Cause: American and Chinese Attributions for Social and

Triandis, Harry C. (1989), ‘‘The Self and Social Behavior inPhysical Events,’’ Journal of Personality and Social Psy-Differing Cultural Contexts,’’ Psychological Review, 96,chology, 67 (December) , 949–971.(July) , 506–552.Mueller, Barbara (1987), ‘‘Reflections of Culture: An Analysis

(1993), ‘‘Collectivism, Individualism as Cultural Syn-of Japanese and American Advertising Appeals,’’ Journaldromes,’’ Cross Cultural Research, 27 (August) , 166–of Advertising Research, (June/July) , 51–59.180.Oliver, Richard L., Thomas S. Robertson, and Deborah J.

(1994), ‘‘Major Cultural Syndromes and Emotion,’’ inMitchell (1993), ‘‘Imaging and Analyzing in ResponseEmotion and Culture: Empirical Studies of Mutual Influ-to New Product Advertising,’’ Journal of Advertising, 22ence, ed. Shinobu Kitayama and Hazel Markus, Washing-(December) , 35–49.

Olney, Thomas J., Morris B. Holbrook, and Rajeev Batra ton, DC: American Psychological Association, chap. 9.


Documents

The Influence of Emotional Appeals across Cultures - CiteSeerX