The impact of cultural collectivism on reward allocation

12
Journal of tatonbty tai Socal Pljchotocy 1984. VoL 47, No 4. 79J-8O4 Cbpmht 1984 by the American Piycboiofmi Aaocuuon. Inc. The Impact of Cultural Collectivism on Reward Allocation Kwok Leung University of Illinois, Urbana-Champaign Michael H. Bond Chinese University of Hong Kong It is argued that the collectivism of a culture leads to different styles of reward allocation with in-group and out-group members. Two studies using Chinese and American subjects examined this issue. In the first study an out-group situation was used, in which subjects were led to believe that they worked with a partner whom they would not meet, to obtain a group reward. The collectivistic Chinese subjects were found to follow the equity norm more closely in dividing the group reward than the individualistic American subjects when pressure of social evaluation was removed. In the second study Chinese and American subjects read a scenario in which an allocator worked with either an in-group or out-group member. The allocator had either a low or high input and used either the equity or equality norm to divide a group reward. Compared with American subjects, Chinese subjects liked an allocator who divided the group reward equally with an in-group member more and regarded such an allocation as fairer. When subjects were asked to assume that they were the allocator and to hypothetically divide the reward, Chinese subjects followed the equity norm more closely than did American subjects when the recipient was an out-group member or when the subjects' input was low. However, when the subjects' input was high and the recipient was an in-group member, Chinese subjects followed the equality norm more than did American subjects. These findings are discussed in terms of the desire for maintaining group solidarity in a collectivist culture. One of the major constructs in theoretical discussions in cross-cultural psychology, so- ciology, and anthropology is that of collectiv- ism-individualism (e.g., Berger, Berger, & Kellner, 1973; Brittan, 1977). However, only recently have we had an empirical basis to evaluate the significance of this construct. In a large scale survey of beliefs and values in 40 countries, Hofstede (1980) extracted four dimensions of national culture—individual- ism was one of them. In his data the U.S. has the highest score on individualism, whereas countries of a Chinese background (Hong Kong, Singapore, and Taiwan) are on the collectivistic end of the scale. This result is consistent with Hsu's (1970) wide-ranging This article is based on &roaster'sthesis submitted by the first author to the Graduate College of the University of Illinois in partial fulfillment of the requirements for the degree. We would like to thank Professors Allan lind and Harry Triandisfortheir critical comments on earlier drafts of the manuscript We would also like to thank Leung Chung for his help in organizing the collection of some of the Chinese data. Requests for reprints should be sent to Kwok Leung, Department of Psychology, University of Illinois, Cham- paign, Illinois 61820. anthropological observations of Chinese and American societies. Although collectivism-individualism is a very important construct, there have been only a few attempts to explicate and define it (e.g., Gould & Kolb, 1964; Kluckhohn, 1952; Parsons, 1977). In general, definitions of collectivism and individualism suggest an emphasis on the social environment for the collectivist and an emphasis on the self for the individualist. These definitions are rather abstract and not focused enough to generate researchable hypotheses. A review of tie sociological literature sug- gests that such concepts as Riesman, Glazer, and Denney's (1961) tradition-direction, Par- sons and Shils's (1952) collectivity-orientation, particularism, and ascription-orientation, and Etzioni's (1975) moral involvement with an organization can be used to describe collec- tivists. Collectivists are more concerned with the consequences of their behaviors on their in-group members and are more likely to sacrifice personal interests for the attainment of collective interests (see also Bond & Wang, 1983; Jones & Bock, 1960). On the other hand, individualists can be 793

Transcript of The impact of cultural collectivism on reward allocation

Journal of tatonbty tai Socal Pljchotocy1984. VoL 47, No 4. 79J-8O4

Cbpmht 1984 by theAmerican Piycboiofmi Aaocuuon. Inc.

The Impact of Cultural Collectivism on Reward Allocation

Kwok LeungUniversity of Illinois, Urbana-Champaign

Michael H. BondChinese University of Hong Kong

It is argued that the collectivism of a culture leads to different styles of rewardallocation with in-group and out-group members. Two studies using Chinese andAmerican subjects examined this issue. In the first study an out-group situationwas used, in which subjects were led to believe that they worked with a partnerwhom they would not meet, to obtain a group reward. The collectivistic Chinesesubjects were found to follow the equity norm more closely in dividing the groupreward than the individualistic American subjects when pressure of social evaluationwas removed. In the second study Chinese and American subjects read a scenarioin which an allocator worked with either an in-group or out-group member. Theallocator had either a low or high input and used either the equity or equalitynorm to divide a group reward. Compared with American subjects, Chinesesubjects liked an allocator who divided the group reward equally with an in-groupmember more and regarded such an allocation as fairer. When subjects wereasked to assume that they were the allocator and to hypothetically divide thereward, Chinese subjects followed the equity norm more closely than didAmerican subjects when the recipient was an out-group member or when thesubjects' input was low. However, when the subjects' input was high and therecipient was an in-group member, Chinese subjects followed the equality normmore than did American subjects. These findings are discussed in terms of thedesire for maintaining group solidarity in a collectivist culture.

One of the major constructs in theoreticaldiscussions in cross-cultural psychology, so-ciology, and anthropology is that of collectiv-ism-individualism (e.g., Berger, Berger, &Kellner, 1973; Brittan, 1977). However, onlyrecently have we had an empirical basis toevaluate the significance of this construct. Ina large scale survey of beliefs and values in40 countries, Hofstede (1980) extracted fourdimensions of national culture—individual-ism was one of them. In his data the U.S.has the highest score on individualism,whereas countries of a Chinese background(Hong Kong, Singapore, and Taiwan) are onthe collectivistic end of the scale. This resultis consistent with Hsu's (1970) wide-ranging

This article is based on & roaster's thesis submitted bythe first author to the Graduate College of the Universityof Illinois in partial fulfillment of the requirements forthe degree. We would like to thank Professors Allan lindand Harry Triandis for their critical comments on earlierdrafts of the manuscript We would also like to thankLeung Chung for his help in organizing the collection ofsome of the Chinese data.

Requests for reprints should be sent to Kwok Leung,Department of Psychology, University of Illinois, Cham-paign, Illinois 61820.

anthropological observations of Chinese andAmerican societies.

Although collectivism-individualism is avery important construct, there have beenonly a few attempts to explicate and defineit (e.g., Gould & Kolb, 1964; Kluckhohn,1952; Parsons, 1977). In general, definitionsof collectivism and individualism suggest anemphasis on the social environment for thecollectivist and an emphasis on the self forthe individualist. These definitions are ratherabstract and not focused enough to generateresearchable hypotheses.

A review of tie sociological literature sug-gests that such concepts as Riesman, Glazer,and Denney's (1961) tradition-direction, Par-sons and Shils's (1952) collectivity-orientation,particularism, and ascription-orientation, andEtzioni's (1975) moral involvement with anorganization can be used to describe collec-tivists. Collectivists are more concerned withthe consequences of their behaviors on theirin-group members and are more likely tosacrifice personal interests for the attainmentof collective interests (see also Bond & Wang,1983; Jones & Bock, 1960).

On the other hand, individualists can be

793

794 KWOK LEUNG AND MICHAEL H. BOND

described by such concepts as Riesman etal.'s (1961) inner-direction, Parsons and Shils's(1952) self-orientation, universalism, andachievement-orientation, and Etzioni's (1975)calculative involvement. Individualists are lessconcerned with the consequences of theirbehaviors on people in the social environment,be they of the same collectivity or differentones. Individualists are more concerned withthe relation of their behaviors to their ownneeds, interests, and goals.

After an extensive review of the literatureon individualism and collectivism, Triandis(1983) has defined individualism and collec-tivism on the basis of in-group/out-groupdifferences. He denned collectivism as,

great emphasis on (a) the views, needs, and goals, of theingroup rather than of oneself, (b) social norms and dutydefined by the ingroup rather than on pleasure, (c) beliefsshared with the ingroup rather than on beliefs thatdistinguish oneself from ingroup, and (d) great readinessto cooperate with ingroup members, (p. 9)

The corresponding definition of individualismbecomes,great emphasis on (a) one's own views, needs, and goalsrather than those of others, (b) on pleasure, fun, personalenjoyment rather than on social norms or duty as definedby others, (c) one's beliefs that are unique, and (d)maximizing one's own outcomes, (p. 10)

Empirical support for Triandis's conceptual-ization, which requires a distinction betweenin-groups (e.g., family and friends) and out-groups (e.g., unknown others), can be foundin the studies of Triandis and Vassiliou (1972)and Vassiliou and Vassiliou (1966). Tajfel's(1982) work on social categorization alsoemphasizes the importance of this distinctionbetween in-groups and out-groups.

Equity Theory and Its Relationto Collectivism

The equity theory (Adams, 1965; Homans,1961; Walster, Walster, & Berscheid, 1978)has dominated the explanation and predictionof distributive behavior since its inception.However, Deutsch (1975) has argued that theequity theory is a limited perspective becausein noneconomic social relations in whichgroup cohesiveness and mutual satisfactionare more valued, equality and need ratherthan equity will be more influential in rewardallocation.

In addition, personality may also influencethe choice of a distributive norm. Swap andRubin (1983) found that in reward allocationsituations subjects who had higher scores ontheir interpersonal orientation scale used theequality norm to a greater extent, and subjectswho had lower scores were more likely to usethe equity norm. Note that high interpersonalorientation is characterized by a higher re-sponsiveness to the interpersonal aspects ofrelationships with others, whereas low inter-personal orientation is characterized by anonresponsiveness to the interpersonal aspectsof relationships with others and a desire tomaximize own outcome.

From this analysis, it is clear that persons'socioemotional orientation can be an impor-tant factor governing their choice of distrib-utive norms. It seems reasonable to proposethat a similar relation between socioemotionalorientation and choice of distributive solutionsalso holds in moving from an individual levelto a cultural level. In cultures where inter-personal sensitivity is high, and harmony,solidarity, and cohesion more favorably val-ued, the equality solution should be preferredto the equity solution. On the other hand,the equity principle should be preferred incultures stressing productivity, competitive-ness, and self-gain.

As previously explained, however, whereasindividualists' behavior is not much influ-enced by the group membership of the targetperson, collectivists' behavior is thought tobe affected by whether the target person is anin-group or out-group member. To put itconcisely, collectivists are collectivistic onlytoward in-group members; toward out-groupmembers they become individualistic (e.g.,see Triandis & Vassiliou, 1972). It follows,then, we would expect that collectivists wouldadopt a more egalitarian allocation only within-group members.

There is some evidence suggesting thatcollectivists may even follow the equity normmore closely than individualists with out-group members. In individualistic cultures,members are socialized to treat groups astransient, and contractual engagements maybe more important Persons can join a varietyof groups, and strangers are regarded as po-tential group members. Thus, social skillsand friendliness become necessary in inter-

COULECTIV1SM AND REWARD ALLOCATION 795

acting with strangers, so that potential futureinteractions can be facilitated. Collectivists,on the other hand, are more concerned withharmonious interactions with existing groupmembers, and may view strangers as havinga low potential for becoming group members.In fact, comparisons of Americans andChinese on various personality tests havefound that American subjects endorsed com-munality and sociability more highly onthe California Psychological Inventory (CPI)(Abbott, 1972) and valued support morestrongly on Gordon's survey of interpersonalvalues (Tarwater, 1966). It is likely thatthe general concern for friendliness towardstrangers among Americans may reduce theiremphasis on equity in distributive behaviortoward strangers. Evidence consistent withthe proposed formulation has been found inthe cross-cultural studies of Leung and Bond(1982) and Bond, Leung, and Wan (1982),which found that Chinese subjects used theequality norm more than did American sub-jects in allocating with in-group members.Consistent also are the studies of Mahler,Greenberg, and Hayashi (1981) and Marin(1981), which found that in allocating without-group members, Japanese subjects usedthe equity norm to the same degree as Amer-ican subjects and that Colombian subjectsfollowed the equity norm more closely thandid American subjects, respectively. Note thatJapanese and Colombians are more collectiv-istic than Americans (Hofstede, 1980).

To conclude, the evidence available seemsto support the proposed relation betweencollectivism and distributive behavior. Thestudies reported here were designed to furtherstrengthen the theoretical analysis just pre-sented. Because the studies of Leung andBond (1982) and Bond, Leung, and Wan(1982) had already examined distributive be-havior in an in-group setting, the first studycontrasted the distributive behavior of Chineseand American subjects in an out-group allo-cation setting. An additional variable, theanonymity of the allocation, was also includedin the design because Chinese may be moresensitive to social evaluation pressure thanAmericans (Hsu, 1970), and the proposeddifferences in reward allocation betweenChinese and American subjects may be af-fected by this variable. Also, there has been

a growing interest in the self-presentationaldeterminants of social behavior (Baumeister,1982), and it seemed worthwhile to explorethis issue cross-culturally.

To sum up, this study tested the hypothesisthat Chinese subjects would allocate moreequitably than American subjects with out-group members, regardless of their perfor-mance level.

Experiment 1

Method

SubjectsSubjects were recruited from introductory psychology

courses at the University of Illinois and the ChineseUniversity of Hong Kong. Their participation was inpartial fulfillment of the requirement of the course. Atotal of 192 subjects (96 Americans and 96 Chinese)participated. Each cutural group contained an equalnumber of males and females.

Experimental DesignSubjects were run individually by an experimenter of

the same sex and cultural background. They were led tobelieve that they would work with a partner on a copyingtask and that they would receive a monetary rewarddependent on their total performance. Subjects were toldthey had done either twice as much, or half as much, asthe work done by their partners. They were then assignedto divide the monetary reward between themselves andtheir partners either in a public or in an anonymouscondition. The dependent measure was the amount ofmoney they allocated to themselves. The study employeda four-factor factorial design, with ( 2 X 2 x 2 x 2 ) (Levelsof Performance X Allocation Conditions X Sexes XCultures).

Collectivism of subjects was not measured in this studyfor two reasons. First, the focus of the study was toexamine cultural rather than individual differences. Thedifference between Chinese and Americans in collectivismhas already been reliably demonstrated by different sources(e.g., Hofstede, 1980; Hsu, 1970; Jones & Bock, 1960),and there is little need to replicate this result here.Second, a comprehensive collectivism scale was not avail-able at the time of data collection.1

ProcedureEvery attempt was made to involve the subjects in a

realistic and demanding performance with a task partnerwho remained unseen throughout the experiment. Infact, there was no partner, but rather the experimenterconducted the experiment in a way consistent with thecover story—that a partner was present in the next room.

1 Such a scale is now available from Harry Triandisand Harry Hui at the University of Illinois.

796 KWOK LEUNG AND MICHAEL H. BOND

When the subject arrived, he or she was told by theexperimenter that he or she would form a two-persongroup with another subject and that the two would workon a group task. The subject was told that his or herpartner had already arrived and had started working onthe task in another room. The subject was told to puthis or her bag and coat alongside with his or her partner's.The partner's bag and coat were put there by the exper-imenter to induce the belief that a partner actuallyexisted.

The subject was then led to a small room and wasasked to read the experimental instructions. Instructionswere presented in both a written and an oral formsimultaneously, with a tape recorder repeating what hadbeen printed on the instruction sheet. The purpose ofthe experiment was presented as being concerned withperformance on a motor task of copying words fromknown (Chinese or English) or unknown (French orJapanese) languages. Subjects were also told that moneywould be used as a reward to assess the impact ofmotivational variables on the process and that the effectof having a partner was also being assessed by assigningsome subjects to work with a partner and others to workalone.

The subject was told they would be asked to copyFrench (for Chinese subjects) or Japanese (for Americansubjects) for 20 min, and then copy Chinese (for Chinesesubjects) or English (for American subjects) words foranother 20 min. The team would supposedly be paid anamount in accordance with the number of words copiedduring these two sessions. The subject was told that heor she and his or her partner would work in differentrooms to avoid interfering with one another.

Before the subject began copying he or she was askedto draw lots to decide who would be the allocator for thegroup reward at the end of the experiment. The drawingwas rigged so that subject would always be the allocator.

The subject then began copying French or Japanesewords. After 10 min the experimenter announced thathe or she had to go next door to change the partner'scopy from French or Japanese to Chinese or English.The door's closing simulated this exchange in the nextroom.

The experimenter moved back and forth between therooms, reinforcing the impression that the partner wasworking 10 min ahead of the subject's schedule. In fact,half way through the subject's copying task, the experi-menter thanked and dismissed the partner as if he or shewere present just outside the door of the room in whichthe subject was working.

After the subject finished his or her 40 min of copying,the experimenter took his or her output to another roomfor scoring. Five min later he or she returned with the 2members' production scores and a group reward of U.S.$3.10 or its equivalent in Hong Kong currency. Thesubject was asked to take his or her share, putting theremainder in an envelope or a plastic bag. The experi-menter left the room while the subject allocated thereward.

Independent VariablesInput manipulation. The team member's relative

performance was manipulated by the production scores

presented to the subject by the experimenter. The numberof words correctly copied was said to be 235 and 118(i.e., 2:1) with the subjects half the time designated asbeing the high performer, half the time the low performer.

Anonymity manipulation. To create the impressionthat the division would be unknown to the experimenter;half of the subjects were asked to seal a white paperenvelope with their partner's share of the reward. Thenthey were to place the envelope in a sealed box througha narrow slit and leave without having to wait for theexperimenter to return. This box contained other enve-lopes supposedly being collected for distribution after theexperiment had been completed. Given that the experi-menter had been away from the room, the rewarddivision would appear to be completely private.

Half of the subjects were randomly selected to dividethe reward in a way that would be known to theexperimenter, even though the experimenter was absentduring the actual division. For this group a transparentplastic envelope was used and subjects were required tohand this envelope back to the experimenter after he orshe returned to the room.

The dependent measure in the study was the percentageof the total reward the subject took for him or herself.Subjects were thoroughly debriefed at the end of theexperiment

Results

The first step in the analysis was to convertthe monetary data into percentage shares. A4 factor analysis of variance (ANOVA) wasperformed on the percentage scores. Notethat 33% for a low performer and 67% for ahigh performer were the equitable shares.

Two significant main effects were found.First, subjects in both cultures in the publiccondition took a smaller share of reward thanthose in the private condition, F{\, 176) =5.34, p < .025 (Xs = 48.36% vs. 51.21%).This finding showed that when social pressurewas removed from the allocation situation,subjects' allocation became more self-serving.

Second, subjects in both cultures who per-formed at a higher level took a larger sharethan those who performed at a lower level,i=Kl, 176) = 174.9, p < .001 (Xs = 57.92 vs.41.65). So productivity was used by bothcultural groups as a basis for the allocation.

A significant three-way interaction effectinvolving culture, sex, and performance wasalso found, F{1, 176) = 20.9, p < .001. Thepattern of this interaction is critical in eval-uating the hypothesis put forward in theintroduction (see Table 1).

It is clear that Chinese female subjectsfollowed the equity norm more closely thandid American female subjects, regardless of

COLLECTIVISM AND REWARD ALLOCATION 797

Table 1Subjects' Allocation as a Function of Culture,Sex. and Performance

Ethnicity

AmericansChinese

AmericansChinese

Performance

Low

Female

45.1337.50

Male

40.1943.76

High

53.9459.14

62.3556.26

Note Numbers refer to the percentage of money subjectsallocated to themselves.

performance level. An analysis of the Perfor-mance X Culture interaction with femalesubjects confirmed this impression, F{1,176)= 13.6, p < .01. However, a similaranalysis with male subjects showed that itwas the American subjects who followed theequity norm more closely than the Chinesesubjects, especially in the high performancecondition, F{1, 176) = 7.70, p < .01.

The means also showed that for Americansubjects, males were more equitable thanfemales, whereas for Chinese subjects, thepattern was the opposite.

Discussion

The hypothesis that in an out-group situ-ation, Chinese subjects would follow the equ-ity norm more closely than would Americansubjects was partially supported. Chinese fe-male subjects followed the equity norm moreclosely than did American subjects across thedifferent experimental conditions.

The findings that Chinese male subjectsfollowed equality more than did Americansubjects, especially in the high performanceconditions, was unexpected. One possible ex-planation for the results can be provided.Chu and Yang (1977) found that, even withstrangers, Chinese subjects might still prefera more egalitarian allocation when (a) pres-sure of social evaluation is high, and (b) ownperformance is high. Note however that theydid not find any sex differences in rewardallocation among Chinese subjects.

We may argue that the egalitarian allocationof the male Chinese subjects in the presentstudy was also due to high social evaluationpressure. The public-private manipulation inthe present study permits a reanalysis to testthis contention (see Table 2). However, caremust be taken in interpreting these resultsbecause of the post hoc nature of this analysis.

A planned comparison indicated that formales the Culture X Performance interactionwas not significant, F<\, in the privatecondition and that the interaction effect wassignificant, F{\, 176) = 9.25, p < .01, in thepublic condition. That is, American malesubjects followed the equity norm moreclosely than Chinese male subjects only inthe public conditions, and there was no dif-ference in the allocation of the Chinese andAmerican subjects in the private conditions.The means also showed that the highly egal-itarian division of the Chinese male subjectsoccurred basically in the high performanceconditions, a finding consistent with the re-sults of Chu and Yang's (1977) study.

To sum up, Chinese male subjects tend touse the equity norm toward an out-groupmember when evaluation pressure is reduced.The present laboratory study was subse-quently replicated by using a scenario format(Leung & Bond, 1983). This replication pro-vides a good test of this speculation. Evalua-tion apprehension in the scenario replicationwas virtually zero, because the questionnaireswere administered anonymously in a groupsetting. As expected, both female and maleChinese subjects used the equity norm morethan did American subjects across the differ-

Table 2Male Subjects' Allocation as a Function ofCulture, Performance, and Allocation Condition

Ethnicity

AmericansChinese

AmericansChinese

Performance

Low

Private condition

42.3444.17

Public condition

38.0443.35

High

62.7460.22

61.9652.30

798 KWOK LEUNG AND MICHAEL H. BOND

ent experimental conditions. This findingsuggests that collectivism may be more sen-sitive to social evaluation and more willingto try to maintain a favorable image in frontof significant others. The use of the equalitynorm, especially when one's performancelevel is high, conveys to the recipient and theexperimenter an image of friendliness andsolidarity and is therefore preferred by theChinese subjects in the public allocation con-ditions. This difference between collectivistsand individualists should be further exploredin future research.

Taken as a whole, the findings of thepresent study contrasted sharply with theresults of Leung and Bond (1982) and Bond,Leung, and Wan (1982), which showed thatin in-group situations Chinese subjects weremore egalitarian than American subjects. Wemay conclude that (a) when evaluation fromsignificant others (the experimenter, in thiscase) is reduced, or nonexisting, and (b) theother is an out-group member, Chinese sub-jects are more likely to follow the equitynorm than American subjects.

Another finding that is worthy of discussionis that, consistent with a large number ofprevious studies (e.g., Landau & Leventhal,1976; Sampson, 1975), American male sub-jects were more equitable than Americanfemale subjects. For Chinese subjects, how-ever, female subjects were more equitable.Recall also that Chu and Yang (1977) foundno sex difference in the distributive behaviorof Chinese college students in Taiwan. Thisdifference in findings is probably best ex-plained by a difference in the educationalopportunities in these two cultures. Universityeducation is less available in Hong Kong andTaiwan than in the U.S., and women areunderrepresented in universities. Female col-lege students in Hong Kong might be morecompetitive and independent and as a resultobserve equity to a larger extent. To evaluatethis hypothesis a more sensitive and refinedmeasure of individualism and collectivism isobviously needed.

Perception of Distributive Behavior

In addition to distributive behavior, people'sjudgment of the fairness of an allocation andthe likability of an allocator may also be

influenced by an interplay of the group mem-bership of the recipient and the performancelevel of the allocator. We have concluded thatthe collectivistic Chinese subjects adopted amore egalitarian distribution toward in-groupmembers than toward out-group members.Also, in the case of a zero-sum reward, theperformance level of the allocator becomesimportant because the use of the equalitynorm does not always lead to a larger sharefor an in-group recipient. To allocate a largershare to an in-group member, high-inputcollectivists are more likely to prefer an equalallocation, and low-input collectivists are morelikely to prefer an equitable allocation. As-suming mat the reason why collectivistschoose to divide a group reward in the aboveway is that they perceive this mode of allo-cation as fair—other persons who allocate areward in the same way should also be judgedas fair. It then seems reasonable to proposethat, compared with individualistic subjects,collectivistic subjects would be more likely tojudge an egalitarian distribution made by ahigh-input performer as fairer when the dis-tribution is made among in-group membersthan when made among out-group members.In a similar vein, collectivistic subjects wouldalso be more likely to regard an equal distri-bution among out-groups as more unfairthan individualistic subjects.

Fairness is certainly one factor that canenhance a person's likability, and a personwho observes a fairness norm should beperceived as more likable. In addition, if atarget person uses a mode of distribution thatthe observers themselves have a tendency touse, the target person should be better liked,for similarity leads to attraction (Byrne, 1971).Collectivistic subjects would therefore be likelyto judge a high-input allocator who allocatedequally among in-groups as more likable thanone who allocated equally among out-groups.Collectivistic subjects should also judgean allocator who allocated equally amongstrangers as less likable than do individualisticsubjects.

The second study to be reported was de-signed to test these hypotheses. In the firststudy out-group situations were created, andthe results were compared to the resultsobtained from in-group situations in previousstudies (Bond, Leung, & Wan, 1982; Leung

COIJJECnVISM AND REWARD ALLOCAnON 799

& Bond, 1982). To provide a clearer test ofthe proposed hypotheses, the in-group/out-group variable was manipulated in the secondstudy. A scenario format was used in thisstudy for three reasons: First, evaluation ap-prehension is minimized by a scenario formatSecond, cross-cultural equivalence can bemore easily attained with a scenario formatbecause the stimuli presented can be stan-dardized more accurately. Third, it is easierto manipulate the relationship of the recipientwith the allocator in a scenario study than ina laboratory study.

The hypotheses of Experiment 2 can thenbe stated formally as follows: (a) Comparedwith American subjects, Chinese subjects willperceive an equal allocation to in-groupmembers made by a high-input allocator tobe fairer, and the allocator to be more likablethan when an equal allocation is made toout-group members, (b) Compared withAmerican subjects, Chinese subjects will judgean equal allocation among out-group mem-bers to be more unfair and the allocator lesslikable, (c) In an in-group situation, high-input Chinese subjects will allocate moreequally than high-input American subjects,(d) Low-input Chinese subjects will allocatemore equitably than low-input Americansubjects to members of both in-groups andout-groups.

Experiment 2

Method

SubjectsSubjects were recruited from introductory psychology

classes at the University of Illinois and the ChineseUniversity of Hong Kong. Their participation partiallyfulfilled a course requirement. A total of 144 Chineseand 128. American subjects participated. Again, eachcultural group contained equal numbers of males andfemales.

ProcedureSubjects were run in small groups of 20 to 40 people

by a female experimenter of the same cultural backgroundas the subjects. Subjects were informed of the purposeof the study in general terms, namely, that it was a studyof decision-making processes. They were asked to read ascenario written in their first i « T ' » y The scenariodescribed 2 co-workers who were either friends or strangersworking on an additive task together and who received amonetary group reward for the task. The allocator of thereward was one of the 2 co-workers, whose completed

work was either two times more than the other co-worker's or half of the other co-worker's and who eitherdivided the reward according to the equity or equalityprinciple.

Dependent Measures and AnalysesSubjects rated on 7-point scales (a) the fairness of the

allocation, (b) the fairness of the allocation as perceivedby the recipient, (c) the satisfaction level of the recipient,and (d) their satisfaction level assuming they were therecipient2 Subjects then indicated how they would allocatethe reward under the assumption that they were theallocator. Subjects also rated the allocator on 8 bipolarperson-perception items: friendly-unfriendly, unfair-fair,warm-cold, uniikable-likable, stingy-generous, kind-un-kind, unhelpful-helpful, and responsible-irresponsible.

After the experiment subjects were debriefed, that is,provided with a detailed explanation of the purpose ofthe study.

In addition to some overall ANOVAS, which will beperformed on these dependent variables, some specificplanned comparisions will also be made to test theproposed hypotheses.

Results

Person Perception Items

A principal components factor analysiswas run on the eight person-perception itemsfor each culture. Only factors with eigenvaluesgreater than 1 were rotated to orthogonalstructure by using the varimax procedure.Because the resulting factor analyses fromthe two cultures were extremely similar, thedata from both cultural groups were com-bined together in an overall factor analysis.

A two-factor solution emerged. The firstfactor was labeled likability; it contained thefollowing items: friendly, warm, unlikable (—),stingy (—), kind, and helpful. The secondfactor was labeled fairness; it contained twoitems, namely, unfair (—) and responsible.

Factor scores were computed for each factorby summing the appropriate items that loadedgreater than .5 on the factor. To eliminatedifferences caused by different response setsin the 2 cultural groups, the raw scores werestandardized within each culture. A five-wayANOVA was then performed on each of thesestandardized scores as dependent variables.To provide a clearer presentation of the re-

2 Analyses of the last three items did not yield anyresults pertaining to the major hypotheses, and theseresults will not be reported.

800 KWOK LEUNG AND MICHAEL H. BOND

suits, only findings relevant to the hypothesesare reported below.

Likability. A highly significant interactioneffect was found between performance andstyle of allocation, F{\, 240) = 278.0. p <.001. Generous allocators, that is, those whoadopted a norm that gave the recipient alarger share, were better liked, and self-servingallocators, who allocated equally when theirperformance was low, were least liked.

A marginally significant three-way inter-action effect involving culture, relation, andstyle of allocation was found, F(l, 240) =3.23, p < .08. In an analysis for simpleinteraction effects, it was found that in thefriend condition, the interaction effect betweenculture and allocation style was significant,F[lt 240) = 6.05, p < .05, whereas in thestranger condition this interaction was non-significant, F < 1. Thus, American subjectsliked an allocator who allocated equitablywith a friend more, whereas Chinese subjectsliked an allocator who allocated equally witha friend more (see Table 3).

The above effect seemed to be unaffectedby the sex of subjects, as when the sexvariable was added, the four-way interactionwas not significant, F < 2.5, ns.

Fairness of the allocator. It was foundthat allocators who followed the equity normwere rated as fairer than those who followedthe equality norm, F\\, 240) = 101.8, p <.001. In addition, a two-way interaction effectwas found between performance and style of

Table 3Likability Ratings as a Function of Culture,Relation, and Allocation Style

Table 4Ratings of Fairness of Allocation as a Function ofCulture, Sex. and Allocation Style

Ethnicity

AmericansChinese

AmericansChinese

Allocation

Equality

Friends

.281.25

Strangers

- .47- .12

style

Equity

1.24-.61

-1.05-.51

Ethnicity

AmericansChinese

AmericansChinese

Sex of subjects

Female

Equality

- .40-.81

Equity

.40

.74

Male

- .60- .52

.60

.59

Note. Friends and strangers denote the relation of the re-cipient with the allocator. Numbers refer to standardizedlikability ratings. The larger the number, the higher is thelikability judgment

Note. The larger the number, the higher is the fairnessjudgment

allocation. F\l, 240) = 86.3, p < .001. Whenthe equality norm was used, a high-inputallocator was rated as fairer than a low-inputallocator. When the equity norm was used,however, the low-input allocator was rated asfairer than the high-input allocator.

Fairness of the Allocation

A five-way ANOVA revealed that an equi-table allocation was rated as fairer than anequal allocation, F{\, 240) = 170.2, p< .001.In addition, A two-way interaction effect wasfound between performance and style of al-location, i=Xl, 240) = 39.5, p < .001. Whenthe equality norm was used, the allocationby the high-input allocator was rated as fairerthan that of a low-input allocator. When theequity norm was used, however, the allocationby the low-input allocator was rated as fairerthan that of the high-input allocator.

A significant three-way interaction effectamong culture, sex, and style of allocation,F[\, 240) = 5.4, /><.025, suggested thatAmerican male subjects regarded equitableallocations as fairer and equal allocations asless fair, than did American female subjects.This relationship was the opposite for Chinesesubjects (see Table 4).

Another three-way interaction among cul-ture, relation, and style of allocation was alsofound, fU, 240) = 4.64, p < .05. Chinesesubjects regarded an equal allocation betweenfriends as fairer than when it was betweenstrangers, and American subjects regarded anequitable allocation between friends as fairer

COLLECTIVISM AND REWARD ALLOCATION 801

than when it was between strangers (seeTable 5).

To test the hypothesis that Chinese subjectswould judge an equal allocation to an in-group co-worker as fairer when the allocatorwas a high performer than when he or shewas a low performer, two a priori contrasts3

were performed. The four-way interactioninvolving culture, relation, allocation style,and performance was not significant, F < 1.5,ns. However, it was found that Chinese sub-jects rated an equal allocation between friendsas fairer only when the allocator was a highperformer, f\\, 240) = 5.89, p < .05 (Xs =-.02 vs. -.62), but not when the allocatorwas a low performer, F > 1 (Xs = — 1.00 vs.-1.02). American subjects did not show thispattern, F < 1, ns.

Allocation of the Reward

The first step in the analysis was to convertthe monetary data into percentage shares.Note that 33.33% for a low performer and66.67% for a high performer were the equi-table shares.

The critical three-way interaction involvingculture, relation, and performance just fellshort of significance, F\l, 240) = 2.6, p <.11. Nevertheless, two a priori contrasts wereperformed to test hypotheses 3 and 4. First,it was hypothesized that high-input Chinesesubjects would allocate more equally withfriends, and more equitably with strangers,than would American subjects. The interac-tion effect pertaining to this hypothesis wasin the predicted direction and significant,F{\, 240) = 14.8, p< .01 (see Table 6).

Table 5Ratings of Fairness of Allocation as a Function ofCulture, Relation, and Allocation Style

Table 6Subjects' Allocation as a Function of Culture,Performance, and Relation

Ethnicity

AmericansChinese

AmericansChinese

Friends

Equality

- .57-.51

Equity

.57

.58

Relation

Strangers

- .43- .82

.43

.75

Ethnicity

AmericansChinese

AmericansChinese

Relation

Friends

Low performance

37.0034.25

High performance

58.5056.00

Strangers

36.5034.17

57.0060.17

Note Numbers refer to the percentage of money subjectsallocated to themselves.

Second, it was hypothesized that low-inputChinese subjects would allocate more equi-tably than American subjects, regardless ofwhether the recipient is a friend or a stranger.This hypothesis was also confirmed, F{\,240) = 11.9, p < .01 (see Table 6). Finally,no significant sex differences were found.

Discussion

Performance and Style of Allocation

It was found that performance level inter-acted with style of allocation on judgmentsof (a) likability of the allocator, (b) fairnessof the allocator, and (c) fairness of the allo-cation. A generous allocator (i.e., one whoused a norm that gave a larger share to therecipient) was better liked. This finding sug-gests that distributive norms interact withperformance to determine likability ratings.Observers liked high-input-equality and low-input-equity allocators most. In fact, thispattern of results is consistent with findingsobtained by Feather and O'Driscoll (1980)

3 These a priori tests are legitimate, as Keppel (1973)has recommended that "These planned comparisions canbe made whether the omnibus F is significant or not. . . With planned comparisions, the experimenter isinterested not in the simultaneous test of all comparisionsbetween pairs of means but in the ones which haveexperimental relevance" (p. 90, italics original). Ourreasoning is in fact a priori, and the use of such tests isjustified. The subsequent use of such tests is also justifiedon this basis.

802 KWOK LEUNG AND MICHAEL H. BOND

with Australian subjects, and by Kahn,Lamm, and Nelson (1977) with Americanand German subjects.

The results of the fairness ratings of theallocator and the allocation are identical.Again, generous allocators and allocationsthat resulted in a larger share for the recipientwere regarded as fairer. Furthermore, regard-less of the performance level of the allocator,equitable allocators and allocations were re-garded as fairer than equal allocators andallocations.

From the results we can conclude that thelikability judgments of the allocator are ba-sically determined by his or her generosity,whereas the fairness judgments of the allocatorand the allocation are determined by thedistributive norm adopted, with equity alwaysjudged as fairer than equality. However, ge-nerosity can also enhance the fairness judg-ments of the allocation and allocator.

Collectivism and Perception ofDistributive Behavior

Consistent with hypothesis 1, Chinese sub-jects liked an allocator who allocated equallywith a friend more than one who allocatedequitably, whereas American subjects likedallocators who allocated equitably with afriend more than those who allocated equally.

The findings of the fairness judgments areconsistent with the likability judgments.Chinese subjects regarded an equal allocationbetween friends as fairer than between strang-ers, whereas American subjects regarded anequal allocation between strangers as fairerthan between friends. Symmetrically, Chinesesubjects regarded an equitable allocation asfairer between strangers than between friends,whereas American subjects regarded an eq-uitable allocation between friends as fairerthan between strangers.

Taken as a whole, these results suggest aheightened concern for in-group harmony forthe collectivistic Chinese subjects. With out-group members, however, the concern of thecollectivist is less oriented toward enhancingharmony and more toward fairness. Equitybecomes an appropriate norm to use, as it isregarded by both cultural groups as fair. Thatis probably why Chinese subjects preferred

an equitable allocation with a stranger inboth their perceptions and behavior.

It is interesting to note that Americansubjects regarded an equitable allocation be-tween friends as fairer than between strangers,and furthermore, they regarded an equitableallocator allocating with a friend as morelikable than an equitable allocator allocatingwith a stranger. At this point we can onlyprovide a speculation for this pattern. InAmerican culture, equity is a very importantvalue, and people observing this norm areapproved of and regarded as fair. In socialinteractions with friends, however, there ispressure to avoid conflicts and to maintain asmooth relationship, and the equity norm isinconsistent with these ends (e.g., Lerner,1974; Leventhal, Michaels, & Sanford, 1972).If a person runs the risk of dissolution offriendship by observing the equity norm inorder to be fair, American observers mayperceive the allocation as extremely fair, andthe allocator as more likable because fairnessis an important value. In fact, the fairnessratings of the allocator also showed this pat-tern in the American subjects but not at asignificant level. In Chinese culture, however,the desire to maintain harmony with in-group members probably leads Chinese toregard an equitable allocation with friends asless fair than one with strangers.

General Discussion

The relationship between collectivism andsocial behavior can best be understood by thefollowing themes emphasized by collectivists:(a) a heightened distinction between in-groupsand out-groups, (2) a concern for harmonyin in-group situations, and equity in out-group situations, and (c) a willingness tosacrifice for in-group members. The in-group/out-group distinction and the concern forharmony in in-groups have been discussedextensively by Triandis (1972; 1983), Hofstede(1980), and Hsu (1970). The coUectivist'swillingness to sacrifice for in-group membersand its impact on social behavior have beendiscussed by Triandis (1972) but had notbeen studied systematically until the presentstudy. Specifically, it was shown that thecollectivistic Chinese subjects would be more

COLLECTIVISM AND REWUUD ALLOCATION 803

likely to use the equality norm when theirperformance was high and when the recipientwas a Mend. In doing so, the recipient re-ceived, at the expense of the allocators' shares,a larger share of the reward than what he orshe would have received if the equity normwas used. On the other hand, if their perfor-mance was low, they would be more likely tochoose the equity norm to give the recipienta larger share at the expense of their ownshares.

How then does collectivism translate intosocial behaviors? To answer this question, alarge sample of behaviors and contexts mustbe sampled and analysed. However, so farvery few social behaviors have been examinedwithin a collectivism-individualism frame-work (for exceptions, see Bond & Forgas,1983; Bond, Wan, Leung, & Giocalone, 1983).Triandis's (1978) work on a universal classi-fication of social behaviors may provide asystematic framework for examining collec-tivism and its relation to social behaviors. Tosay the least, we are still a long way from athorough understanding of the relation be-tween collectivism and social behaviors. Thepresent study must be viewed as a startingpoint in that direction, as it examined onlyone class of social behavior. Nevertheless, theresults are entirely consistent with the notionthat collectivists are oriented toward harmonyin well-defined in-groups and behave in asharply different way toward out-group mem-bers.

To sum up, this project suggested andprovided some support for a way to concep-tualize the impact of collectivism on socialbehavior. Further research should be directedat other types of social behavior to evaluatethe usefulness of this conceptualization.

ReferencesAbbott, K_ A. (1972). Harmony and individualism. Taipei;

Oriental Cultural Service.Adams, J. S. (1965). Inequity in social exchange. In L.

Berkowitz (Ed.), Advances in experimental social psy-chology (Vol. 2. pp. 267-299). New York: AcademicPress.

Baumeister, R. F. (1982). A self-presentational view ofsocial phenomena. Psychological Bulletin, 91, 3-25.

Berger, P., Berger, B., & Kellner, H. (1973). The homelessmind. New York: Random House.

Bond, M. H., & Forgas, J. (1983). Linking person per-

ception to behavior intention across cultures. Manu-script submitted for publication, rrriiyw University ofHong Kong.

Bond, M. H., Leung, K., & Wan, K. C. (1982). Howdoes cultural collectivism operate? The impact of taskand maintenance contributions on reward distribution.Journal of Cross-cultural Psychology, 13, 186-200.

Bond, M. H., Wan, K. C, Leung, K., & Giocalone, R.(1983). Cultural collectivism and reactions to verbalinsult. Manuscript submitted for publication, ChineseUniversity of Hong Kong.

Bond, M. H., & Wang, S. H. (1983). Aggressive behaviorin Chinese society: The problem of maintaining orderand harmony. In A. P. Goldstein & M. Segall (Eds.),Global perspectives on aggression (pp. 58-74). NewYork: Pergamon.

Brittan, A- (1977). The privatized world. London: Rou-tledge & Kegan Paul.

Byrne, D. (1971). The attraction paradigm. New York:Academic Press.

Chu, J. L., & Yang, K. S. (1977). The effects of relativeperformance and individual modernity on distributivebehavior among Chinese students. Bulletin of the In-stitute of Ethnology, Academia Sinica, 41, 79-95. (inChinese)

Deutsch, M. (1975). Equity, equality, and need: Whatdetermines which value will be used as the basis ofdistributive justice? Journal of Social Issues, 31, 137-149.

Etzioni, A. (1975). A comparative analysis of complexorganizations- On power, involvement, and their corre-lates. New York: Free Press.

Feather, N. T, & O'DriscoU, M. P. (1980). Observers'reactions to an equal or equitable allocator in relationto allocator input, causal attributions, and value im-portance. European Journal of Social Psychology, 10.107-129.

Gould, J., & Kolb, G. A. (1964). A dictionary of thesocial sciences. Glencoe, 111: Free press.

Hofstede, G. (1980). Culture's consequences: Internationaldifferences in work-related values. Beverly Hills, CA:Sage.

Homans, G. C. (1961). Social behavior: Its elementaryforms. New \orfc Harcourt, Brace, & World.

Hsu, F. L. K. (1970). Americans and Chinese. New York:Natural History Press.

Jones, L. V, & Bock, R. D. (1960). Multiple discriminantanalysis applied to "Ways to Live" ratings from sixcultural groups. Sociometry. 23, 162-176.

Kahn, A., Lamm, H., & Nelson, R. E (1977). Preferencesfor an equal or equitable allocate*. Journal of Personalityand Social Psychology. 35, 837-844.

Keppel, K. (1973). Design and analysis: A researcher'shandbook. Englewood Clifis, NJ: Prentice-Hall.

Kluckhohn, C. (1952). Values and value-orientatin in thetheory of action: An exploration in definition andclassification. In T. Parsons & E. A. Shils (Eds.),Toward a general theory of action (pp. 388-433),Cambridge, MA: Harvard University Press.

Landau, S. B., & Leventhal, G. S. (1976). A simulationstudy of administrators' behavior toward employeeswho receive job offers. Journal of Applied SocialPsychology. 6, 291-306.

804 KWOK LEUNG AND MICHAEL H. BOND

Lemer, M J (1974) The justice motive "Equity" and"parity" among children. Journal of Personality andSocial Psychology. 29, 539-550.

Leung, K, & Bond, M H (1982) How Chinese andAmericans reward task-related contributions. A pre-liminary study Psychologm. 25, 32-39

Leung, K , & Bond, M H. (1983). Evaluating the use ofscenario studies for cross-cultural comparisons. A caseof distributive behavior Unpublished manuscript Uni-versity of Illinois

Leventhal, G. S, Michaels, J W., & Sanford, C. (1972)Inequity and interpersonal conflict' Reward allocationand secrecy about reward as methods of preventingconflict Journal of Personality and Social Psychology,23, 88-102.

Mahler, I , Greenberg, L., & Hayashi, H (1981). Acomparative study of rules of justice Japanese versusAmericans Psychologia, 24, 1-8

Mann, G (1981) Perceiving justice across culturesEquity vs. equality in Colombia and in the UnitedStates International Journal of Psychology, 16, 153-159

Parsons, T (1977) The evolution of societies EnglewoodCliffs, NJ Prentice-Hall.

Parsons, T., & Shils, E A (1952). Toward a generaltheory of action Cambridge, MA Harvard UniversityPress

Riesman, D., Glazer, N , & Denney, R. (1961) Thelonely crowd A study of the changing American char-acter New Haven, CT: Yale University Press.

Sampson, E E (1975) On justice as equality Journal ofSocial Issues, 31, 45-64.

Swap, W G, & Rubin, J Z (1983) Measurement ofinterpersonal orientation. Journal of Personality andSocial Psychology, 44, 208-219.

Tajfel, H. (1982). Social identity and mtergroup relationsCambridge, England: Cambridge University Press.

Tanvater, J. T (1966). Chinese and American students'interpersonal values: A cross-cultural comparisonJournal of College Student Personnel, 7, 351-354.

Tnandis, H. C (1972). The analysis of subjective cultureNew York Wiley.

Triandis, H C (1978). Some universals of social behaviorPersonality and Social Psychology Bulletin, 4, 1-16

Tnandis, H. C. (1983). Collectivism vs. individualism: Areconceptualization. Unpublished manuscript, Univer-sity of Illinois

Triandis, H G, & Vassiliou, V. (1972) A comparativeanalysis of subjective culture. In H C. Tnandis (Ed.),The analysis of subjective culture (pp. 299-335). NewYork- Wiley.

Vassiliou, G , & Vassiiiou, V. (1966). Social values aspsychodynamic vanables. Preliminary explorations ofthe semantics of philotimo Acta Neurologica el Psy-chologica Hellemka, 5. 121-135.

Walster, E., Walster, G. W, & Berscheid, E. (1978).Equity Theory and research Boston: Allyn & Bacon.

Received September 1, 1983Revision received February 15, 1984 •