Individualism Collectivism 1988

8/8/2019 Individualism Collectivism 1988

1/16

JournalofPersonality nd SocialPsychology Copyright1988by he AmericanPsychologicalAssociation, nc.1988, Vol.54, No, 2. 323-338 0022-3514/88/$00.75

Individualism and Collectivism: Cross-Cultural Perspectiveson Self-Ingroup RelationshipsH a r r y C . T r i a n d i s , R o b e r t B o n t e m p o , a n d M a r c e l o J . V i l l a r e a lUn iver si ty o f I l l ino is a t Urbana -C ham paign

M a s a a k i A s a i N y d i a L u c c aNihon Un iver si ty Un iver s ity o f Puer to R icoTokyo , Japan R io P ied ras , Puer to R ico

The individualism and collectivism constructs are theoretically analyzed and linked to certain hy-pothesized consequences (social behaviors, health indices). Study 1 explores the m eaning o f theseconstru cts within cu lture (in th e Unit ed States), identifying the individual-differences variable, idi o-centri sm versus alloce ntrism , that corresponds to the constructs. F actor analyses of responses toitem s related to t he constructs suggest tha t UrS. individualism is reflected in (a) Self-Reliance Wi thComp etition , (b) Low Concern for Ingroups, a nd (c) Distance from Ingroups. A higher orde r factoranalysis suggests that Subordination oflngrou p Goals to Personal Goals may be the most im portantaspect of U.S. individualism. Study 2 probe s the limits of the constructs with da ta from two collectiv-ist sample s (Japan and Puerto Rico) and on e individualist sa mple (Illinois) of students. It is shownthat responses depen d on who the ot her is (i.e., which ingroup), the context, a nd the kind o f socialbehavio r (e.g., feel similar to other, attentive t o th e views of others). Study 3 re plicate s previous workin Puerto Rico indicating hat allocentric persons perceive that they receive more and a bette r qualityof social support tha n do idiocentric persons, while the latter report being mo re lonely than theformer. Several themes, su ch as self-reliance, achievement, a nd competition, have different meaningsin the two kinds o f societies, and detailed examinations of the factor patterns show how such themesvary acr oss cultures.

This ar t icle contains two par ts . Par t 1 presents a theoreticalanalysis of the constructs of indiv id ualism (e.g . , de Tocquevil le ,1946) and collect iv ism, including a d iscussion of the probab lean teceden ts and consequen ts o f the emphases on these va lues indif ferent cu ltures . P ar t 2 presents thre e s tudies . The f ir st study,based on U .S . da ta , p rov ides a t t i tude i tems tha t can be u sedto measu re the ind iv idua l -d if f e r ences d imens ion a l locen t r i smver sus id iocen t r i sm. A l locen t r i sm ver sus id iocen t r i sm i s awi th in - cu l tu r e var iab le tha t co r r esponds to co l lec t iv i sm ver susindiv idualism at the cultural level . The second s tudy tes ts thel imi t s o f the cons t ruc ts , w i th da ta f rom Japan , Pue r to R ico ,and I l l ino is . The th i rd s tudy r ep l ica tes p r ev ious ly ob ta ined ( inI l l ino is ) f indings w i th da ta f rom Puer to R ico tha t a l locen t r icper sons r epo r t r ece iv ing more soc ial suppo r t and a be t te r qua l -i ty o f soc ia l suppo r t , and id iocen t r ic per sons r ep o r t h igher lev -els of loneliness .

Part of this work was supported by grants from the University of Illi-nois Research Bo ard (to Triandis) and N ihon University (to Asai).We wish to thank our research c ollabo rators for stimulating ideas,data, and moral suppo rt in carrying out a complex set of studies. Theyare Hec tor Betancourt (Chile), Michael Bond and Kwok Leung (HongKong), Abelard o Brenes (Costa Rica), James Georgas (Greece), GeertHofstede and John Spangenberg (The Netherlands), C. Harry Hui(Hong Kong), Sharon Lobel (Brazil), Gerardo Marin (California),J. B. E Sinha an d Jyoti Verma (India), Bernadette Setiadi (Indonesia),and Hub ert Touzard and G ermai ne de Montm ollin (France).Correspondence concerning th is article should be addressed to H arr yC. Triandis, 603 E. Daniel St., Champaign, Illinois 61820.

323

T h e o r e t i c a l A n a l y s i sCultu re is a fuzzy construct . I f we are to unde rs tand th e waycu l tu r e r e la tes to soc ia l psycho log ica l phenomena , we mus t ana-

lyze i t by de te rmin ing d imens ions o f cu l tu r a l var ia t ion . One o fthe mos t p romis ing such d imens ions i s ind iv idua l i sm-co l lec -t iv ism.

C u l tu r es d i f fe r in the ex ten t to wh ich coopera t ion , compet i -t ion , o r ind iv idua l i sm (Mead , 1967 ) a r e emphas ized . A t thepsychological level, these d if ferences are ref lected in a pers onal-i ty d imens ion labe led allocentr ism versus idiocentr ism.R el iab le ways to m easu re a l locen t r i sm in the Un i ted S ta teswere developed by Tr iandis , Leung, Vil lareal , and Clack (198 5).A l locen t r i sm was found to be pos i t ive ly co r r e la ted w i th soc ia lsuppo r t ( bo th qua n t i ty and sa t i s fac t ion w i th i t ) and wi th lowleve ls o f a l iena t ion and anomie ; id iocen t r i sm was found to bepos i tive ly co r r e la ted w i th ( a ) emphas is on ach ieve men t a nd (b )perceived loneliness.

A l loce n t r i sm - id iocen t r i sm r e flec ts a t the p sycho log ica l l evelthe d imens ion tha t has been labe led collect iv ism versus individ-ual i sm (Hofstede, 1980), cooperation versus individualism(Mead, 1967), or collateraterali ty versus individua lism (K luck -hohn & Strodtbeck, 1961) at the cultural level . As Deutsch(1949, 1962) has conceptu alize d the relat io nship of indiv id ualsand g roups , i t can take th r ee fo rms : Ind iv idua l goa ls can be pos -i t ively (cooperation) , negatively (compe ti t ion) , or no t ( in div idu-a l i sm) co r r e la ted w i th g roup goa ls .

Hof s tede (1980) found the d imens ion o f ind iv idua l i sm incro ss - cu l tur a l work va lue da ta . Th is d imens ion and Power D is -


2/16

32 4 TRIANDIS, BONTEMPO, VILLAREAL, ASAI, LUCCAtance, with which i t correlated - . 67 , accoun ted for most o f thevariance in an ecological factor analysis of value data (in whi chthe n um be r o f cultures is the N o f the cor relation coefficients).Mezei (1974), in an individual factor analysis, found that indi-vidual ism was the most im porta nt dimension in his factor anal-ys is of the Klu ckho hn and Strodtb eck (1961) values .

Given the theo retical significance of studying the relationshipof person to grou p across cul tures , these em pirical f indings fa-vor the decision to study this dim ensio n intensively. The psy-chological correlates o f individu alism-co llectivism have as yetbeen studied very little. Yet it is possible to develop attitudeitems that tap the various orientations (Hui, 1984; Triandis etal., 1985).

The distinc tion betw een the cu ltural (collectivist vs. individu-alist) and psychological (allocentric vs. idiocentric) levels an dtheir corresponding terminology needs to be emphasized. Hof-stede (1980) has show n that ecological factor analyses (in whic hthe num ber o f observat ions is based on the numb er of cultures)give different results fro m with in-cultu re analyses (in which thenumb er o f obse rva t ions a re the numb er of people respondingto a questionnaire). Yet it is possible to link the two analyses,as was done by Triandis et al. (1986). It avoids confu sion if weuse the terms collectivist cul tures and allocentric individuals.This also allows us to deter min e if allocentric persons in indi-vidual is t cul tures relate to groups different ly than do al locentr icindividuals in collectivist cultures.

An essential attribu te o f collectivist cultures is that individu-als may be induced to subordinate their personal goals to thegoals of some collective, which is usually a stable ingr oup (e.g.,family, band, t r ibe) , a nd mu ch of the behavior o f individualsmay co ncern goals that are consis tent with the goals of this in-group. In individual is t cul tures there are m any mor e ingroups l(e.g., family, coworkers, clubs, mo torcy cle gangs), and mu ch ofthe behavior o f individuals concern s goals that are consis tentwith various ingroups. In collectivist cultures the relationshipof the individual to the ingroup tends to be s table , and evenwhen the ingro up m akes highly cos tly deman ds the individualstays with it . On the othe r hand, in individualist cultur es peopleoften drop those ingroups that are inconvenient ly demandingand form new ingroups. As a result, in individualist culturesdemands by ingroups on individual contribut ions are highlysegmented, requiring contribut ions only at a certain t ime andplace or of a certain kind; in con trast, in collectivist cultures thedeman ds are diffuse.

Cultural complexi ty, as indexed by Murdock and Provost(1973), is in all probability related to individualism-collectiv-ism by a U-shaped funct ion. That is, in extremely s imple socie-ties (e.g., the Mbuti Pygmies) there is proto-individualism, inwhic h the individual is closely related to v ery few others and hasconsiderable freedom to act independent ly of others . At higherlevels o f com plexity (e.g., the R oma ns, A ztecs, Chinese) collec-tivism is very high. The individual relates to a few very impor-tant ingro ups, org anized in conc entric circles (e.g., nucle ar fam-ily, extended family, clan, city, state). A lthou gh the contro l ex-erted by these ingroups diminishes as we move from the nuclearfamily to the s tate , i t i s general ly more pronounced than thecontrol in even more com plex cul tures . In extremely complexcul tures (e.g. , mod ern industr ia l cul tures) , the nu mb er o f in-groups that one can have is muc h greater than in the col lect ivis t

cul tures . M ode m cul tures are neoindividual is t ic, characterizedby both independence from ingroups and dis tance (emotionaldetachmen t) from ingroups . Thus , o ne is able to "do on e 's ownthing" and get away with i t . A problem for one 's ingroup maynot have mu ch consequ ence for the individual.

Cultur al e lements change slowly. In societies with long tradi-tions the collectivism elements may persist although the socie-ties have become very complex (e.g., Japan). However, oneought to observe shif ts toward individual ism as complexi ty in-creases . The increase in volun tary o rganizat ions in the las t 20years in Japan (Asai 's observat ions) can be seen as an index ofthis kin d of shift.

Imp ortan t antecedents of individualism, in addi t ion to cul-tural complexity, are (a) having a frontier, (b) having substantialnum bers o f immigrants , a nd (c) having rapid social and geo-graphical mobil ity, a ll of which tend to make the control of in-grou ps less certain. T he high levels of individualism observ edby Hofs tede (1980) in the United States , Austral ia , and Canad aare consistent with this point.

I t is likely that G ross Nat ional Prod uct (G NP) is both an an -tecedent and a consequent o f individualism. Att luence impliesthe abi li ty to "d o on e 's own thing," but "d oing one 's own thin g"implies more creat ivi ty for the society, hence mo re inn ovat ionand more economic deve lopment.

A covariate o f cultural com plexity is resource availability. Insocieties that are very simple resources are very limited; inmore comp lex societ ies resources are mo re avai lable . In mod -em, industr ia l societ ies we f ind resource abu ndan ce in the for mof widely available manu factured goods and foods that mu st bedestroyed because they cannot be sold.

It follows from this analysis that individuals do not feel asat tached to any ingroup when there are numerous ingroups towhich they can be at tached, and when each ingroup providesonly a small port ion o f their material and emotiona l securi ty.As a result, social behav ior is likely to be very different in pr oto -individualist, collectivist, and neoind ividualist societies.

Conformity may occur more frequent ly in col lect ivis t cul-tures, when the nor ms are dear, and sanct ions are l ikely to beimposed for deviant behavior. However, when the n orm s are u n-clear , and sanct ions are unl ikely to be imposed, we might ob-serve anticonformity. This explains Frager's (1970) findingsthat Japanese subjects conform ed less (25%) than d id U.S. sub-jects (usual ly 33%) in Asch-type conf orm ity experimental set-t ings and showed unusual ly high levels (36%) o f ant iconform ity(giving the wrong response du ring those t r ia ls when the m ajor-ity gave the correct response). Obviously, there are few normsor sanctions in the social psychological laboratory.

As Foa and Fo a (1974) have observed, the resources that areexchanged in m odern cul tures tend to be universal is t ic , th at is,money, information, and goods . In t radi t ional cul tures , onefinds higher rates o f exchange of part icularist ic resources, thatis, love, status, and service. As cultures become complex andindividual ist ic they shif t from the t ime -consum ing part icularis -tic to universalistic exchanges. For example, paren ts give mo ne ymore frequent ly and love and individual a t tent ion to chi ldren's

Defined as a set of people with whom one shares some attribute thatcontributes t o one's positive social identity.


3/16

INDIVIDUALISM AND COLLECTIVISM 32 5needs less frequently. I f we measure d the frequencies of socialexchanges in the two types of cul tures we should observe moreparticularistic exchanges in the collectivist and more universal-istic exchanges in the individualistic cu ltures. T his is a testablehypothesis.

In collectivist cultures, cooperation is high in ingroups but isunl ikely when the other person belongs to an outgroup. Thesame phenomenon can be observed in individual is t cul tures ,but the difference between ingro up and outgro up is a t tenuated(Triandis, 1972). People in individualist cultu res are very goo dat meet ing outs iders , for ming new ingroups , an d get t ing alongwith new people.

Social relat ions with unequa l power are more co mm on in col-lectivist cultures than in individualist cultures. In collectivistcul tures the m ost im porta nt relat ionships are vert ical (e .g., par-ent-chi ld) , whereas in individual is t cul tures the most impor-tant relat ionships are horizontal (e .g. , spouse-spouse, fr iend-friend). In collectivist cultures, interdependence is maximizedbetween parent and chi ld by frequent guidance, consul tat ion,social iz ing in which the chi ldren are included, and penetrat ioninto the child's private life. In individualist cultures, there isemotiona l detachmen t , independence, and privacy for the chi ld.Marriage between individuals fr om the two kinds o f cul tures islikely to lead to difficulties, par ticularly if a collectivist malemar ries an individualist female. She is then likely to expect thatthe spouse-spouse bond wil l be s t ronger than the parent-sonbond, w hile he is l ikely to expect the mo ther -son ( in the Indo-Euro pean col lect ivis t cul tures) or fathe r-son ( in the East Asiancol lect ivis t cul tures) to be s t ronger than the spouse-spousebond. W ho gets priori ty in fam ily relat ionships is a lways a del i-cate issue, and it gets to be very difficult to resolve when un-s tated assumptions ab out the priori ty of part icular bonds arecontradictory.The em phasis is usual ly on people more than on task in col-lectivist cultures, and the reverse happens in individualist cul-tures : For example, when meet ing a fr iend on the way to workpersons in a collectivist culture will stop and chat whereas thepersons in an individual is t cul ture migh t excuse themselves oravoid the meet ing. Relat ionships with ingroups are intens iveand interdependence is high in collectivist cultures, whereasthere is more detachmen t , dis tance, and self-rel iance in individ-ual is t cul tures . Social relat ions tend to be m ore end uring an dinvolu ntary and to occu r in large groups in col lectivis t cultures ;they a re more t emporary and volunta ry and occur in smal lgroups in individualist cultures. In collectivist cultures, one at-temp ts to "pap er over" confl ic t within the ingroup, which oftenrequires i l l ic i t behaviors that are hidden from ingroup mem-bers; in individualist cultures, one is more likely to bring con-flict to the op en, so there are mo re lawsuits.

I t i s c lear that the in ducem ent o f individuals to adopt gro upgoals resul ts in advantages to the group or to an authori tywithin the group. However, the social exchange appears fair tosome individuals because such groups often prov ide social sup-port, resources, and security. In individualistic cultures, the in-dividual has ma ny rights and few obl igat ions in relat ion to in-groups, bu t ingro ups also provide less social supp ort, resou rces,or securi ty to individuals .

These general patterns, however, are modified by individualdifferences in the extent that individuals accept ingrou p norm s.

Allocentric persons in collectivist cultures feel positive aboutaccepting ingroup norms and do not even rai s e the ques t ion ofwhe ther or not to accept them. Acceptance of ingrou p n orms i san unstated assum ption o f the cul ture that they do not chal-lenge. However, idiocentric persons in collectivist cultures feelambiva lent and even b i tt e r about acceptance of ingrou p no rms .They wonder i f this or that no rm is necessary, or i f they shouldcomp ly with i t. Thus , they chal lenge the idea that they shouldcomply. Nevertheless, s ince most people in such cul tures co m-ply, they tend to com ply too. Consequent ly, whereas al locentr icpersons in col lect ivis t cul tures may experience consis tencyam on g the behavioral, affective, and co gnitive elements o f theirsocial behavior, idiocentrics may experience discrepancies: Thebehavioral e lements may com ply to the norms, but the affectiveand cogni t ive elements usual ly ques t ion the norm .

In individualis t cul tures paral le l pheno men a m ay take place.Idiocentric persons in individual is t cul tures f ind i t comp letelynatural to "do their own thing" and to disregard the needs ofcommunit ies , family, or work group. But al locentr ic personsfeel concerne d about their commu nit ies and ingroups . Th e for-mer will show consis tency o f the behavioral , affective, and cog-nitive elements o f their social behavior, wh ich is governedmost ly by hedonis t ic and social exchange concerns ; the la t termay also be consis tent, bu t their behavior m ay be governed byingroup norms .

J . B. P . Sinha (personal communicat ion, August 25, 1986)believes that idiocentric persons in collectivist cultures yield togroup nor ms less than al locentr ic persons in individual is t cul-tures . This is a tes table hypothes is that has been supported(Bontempo, Lobel, & Triandis, in press).

People in individualistic cultures often have greater skills inentering and leaving new social groups . They make "fr iends"eas ily, but by "fr iends" they mean nonin t imate acquaintances .People in collectivist cultures have fewer skills in making new"friends ," but "fr iend " in their case implies a l i fe-long int imaterelat ionship, with ma ny obl igat ions. So the q ual i ty of thefriendships is different. This difference in quality may compli-cate our un ders tanding o f the construct o f collectivism, s incepeople in individualistic cultures are likely to a p p e a r more so-ciable , while int imacy is not a readi ly observable at t r ibute . Indeveloping measuremen ts o f the construct , i t i s imp ortan t tomake such dis t inct ions .

In collectivist cultures, the indiv idual h as few ingro ups (oftenjus t one) and frequent ly everybody else is in the o utgrou p (Tri-andis & Vassiliou, 1972); behavio r tow ard ou tgro ups can be per-ceived as highly individualistic. In individ ualist cultures, the in-d iv idua l has many ingroups and those who a re not ingroupmem bers are not necessari ly in the outgro up; behavior is oftenconform is t in order to be accepted by ingroups . These are com -plex pat terns o f social behavior that require the dis t inct ion be-tween ingroup behavior (conformi ty) and outgroup behavior(do whatever you can get away with) for collectivists. Triandisand Vass i l iou emphasized that behavior toward ingroup andoutgroup members is very different in t radi t ional Greek cul-ture, while it is much less different in the United States. In thecol lect ivis t cul tures the ingroup's influence on behavior isbroad, profoun d, and diffuse; in the individualis t i t i s narrow,superficial, a nd specific.

The sharp difference in behavior toward ingroups and out-


4/16

326 TRIANDIS, BONTEMPO, VILLAREAL, ASAI, LUCCAgroups in collectivist cultures, which does no t occur as sharplyin individualistic cultures, can confuse observers who are notexamining these cultures with the ingroup-outgroup distinc-tion in mind. In collectivist cultures people share and show har-mony within ingroups, but the total society may be character-ized by much disharmony and nonsharing, because so manyinterpersonal relationships are individual-outgroup relation-ships. For instance, if the ingroup is defined as "family andfriends and other people concerned with my welfare" (Triandis,1972), then most relationships with merchants, policemen, gov-ernment bureaucrats, and so on are outgroup relationships. Incontrast, in individualistic cultures people define the ingroup(see Triandis, 1972) as "people who are like me in social class,race, beliefs, attitudes, and values." Tha t is a huge group. Mostinterpersonal behavior occurs within that huge group.

Furthermore, the definition of the ingroup, in collectivist cul-tures, depends to some extent on the situation. While "familyand friends" is the main definition, fellow villagers, political al-lies, or the co unt ry as a whole (in time of war) become the rele-vant ingroups for particular behaviors. In Japan, for employeesof Nissan Motors "we" refers to Nissan and "they " to Toyota,but in discussions of the share of the market versus Americanautomakers they are both "we." As Zavalloni (1980) has argued,each ingroup can have a "we" and "they" aspect, since peopleare perfectly capable o f differentially describing their ingroup;for example, French subjects gave different characterizat ions to"they the French" and "we the French".

All these statements are moderated by the situation; Hui(1984) found that U.S. and Hong Kong subjects behaved as ex-pected in collectivist cultures only with respect to specific others(e.g., their parents, friends, neighbors, or coworkers). In otherwords, a person may be idiocentric in relation to specific in-groups, and allocentric in relation to other ingroups.

Collectivist societies also differ on which element of collectiv-ism they emphasize. For example, the cultures of the Far Eastemphasize ingroup harmony. One must present oneself to oth-ers as modestly as possible and avoid conflict with others at allcosts. Confrontation is highly undesirable. Thus, we find con-flict resolution taking a different form (Leung, 1987) and fewerlawyers per capita in such societies than in individualistic cul-tures. In the collectivist cultures o f the Mediterranean and LatinAmerica, the central value is respect/dignity. Preservation ofone's honor is a supreme value. An ingroup member is honor-able and dependable, does what the ingroup expects, and is un-likely to "betray" the ingroup.Shame and other mechanisms of social cont rol (e.g., religiousbeliefs) are used more widely in collectivist cultures than aremechanisms of internal control, such as guilt. The latter aremore frequently used in individualist cultures. However, in allcultures both types o f mechanisms are used; it is simply a mat-ter of degree o f use.

In some of our exploratory studies, the behavior under studywas a key element in determining whether the subject woulddo "his own thing" or conform to the ingroup demands. Forexample, American college students were more idiocentric withrespect to a choice of religion than with respect to a choice ofspouse. Thus, it may be that the best way to conceptualize theseconstructs is to think of profiles of goals of individuals and spe-

cific others that are positively or negatively intercorrelated oruncorrelated.This theoretical perspective is consistent with the one devel-oped by Greenwald (1982). Greenwald's ego-task analysis inte-grated work on ego-involvement and self-awareness. On the ba-sis of this work, Breckler and Greenwald (1986) identified tasksthat had three kinds of goals: individual, public, and collective.

These corresponded to three kinds of self-esteem that are afunction of reaching such goals. The private se/fcorresponds toself-evaluation based on reaching personal goals, the public selfis sensitive to the evaluations of significant others, and the com-munal selfcorresponds to evaluations by a reference group (in-group). Reaching internal standards is an essential facet ofMcClelland's (1961) conception of need for achievement. Sny-der's (1974) self-monitoring indicates the extent public as op-posed to private standards are used. Sherif and Cantril (1947)defined ego-involvement as concern with the goals of referencegroups. Thus, McClelland focuses on the private, Snyder on thepublic, and Sherif and Cantril on the collective self. In the pres-ent discussion we examine primarily the contrast between theprivate and collective self.

Studies of the subjective culture (Triandis, 1972, 1980) of var-ious cultural groups show differences in collectivism. Thus, thecultures o f southern I taly (Banfield, 1958), traditional Greece(Triandis & Vassiliou, 1972), and rural China (Hsu, 1971,1981,1983) tend to be collectivist, while northern and western Euro-pean and North American cultures tend to be individualistic(Inkeles, 1983; Stewart, 1966). Some aspects of collectivismmay inhibit economic development (Adelman & Morris, 1967;Triandis, 1984); however, these aspects may be related to lowlevels of social pathology, so that the evaluation of whether col-lectivism is functional for a culture is not possible, at this time.

There is a large literature that can be organized around theindividualism-collectivism dimension. Studies of morality(Shweder, 1982), religion (Bakan, 1966), work related values(Hofstede, 1980), the concept of limited good (Foster, 1965),broad value orientations (Kluckhohn & Strodtbeck, 1961),ecology and child-rearing patterns (Barry, Child, & Bacon,1959; Berry, 1979), cognitive differentiation (Witkin & Berry,1975), economic development (Adelman & Morris, 1967), mo-dernity (Inkeles & Smith, 1974; Berger, Berger, & Kellner,1973), the s tructure of constitutions of various states (Massim-ini & Calegari, 1979), and analyses of cultural patterns (Hsu,1981) have used variat ions of this dimension. Individualism isa relatively stable attribute of Americans (Inkeles, 1983). It hasbeen defended (Riesman, 1966; Waterman, 1981) and criti-cized (Hogan, 1975; Lasch, 1978; Rakoff, 1978; Sampson,1977; Smith, 1978). Various attempts have been made to definepatterns that are both individualistic and collectivist (Kanfer,1979; Rotenberg, 1977). This literature has been reviewed byTriandis (1987).

Individualism has been found to be associated with high lev-els of GNP (Adelman & Morris, 1967; Cobb, 1976; Hofstede,1980), but extreme individualism may be linked to severalforms o f social pathology, such as high crime, suicide, divorce,child abuse, emotional stress, and physical and mental illnessrates (Cobb, 1976; Naroll, 1983). Allocentric persons tend tohave happy marriages (Antill, 1983) and are more likely to re-ceive social support that acts as a buffer of life change stresses


5/16

INDIVIDUALISM AND COLLECTIVISM 32 71 0

9

8

7

6

5

.

,

,

O.

U.S. CaucasiansBangor , Penn. - demographical ly and interms of diet l ike Roseto, Penn.but individualist

Caucas ian Benedic t ine Pr ies ts

Ros eto , Penn. - c ollect ivist vi l lage in U.S.

Japa nese in Haw ai iCaucas ian Benedic t ine Monks

Japa nese in Japan

Caucas ian Trap is t Monks

Figure 1. Heart a ttack rates per 1,000 inhabitants i n selectedsamples (data from H enry an d Stephens, 1977).

(Cohen & H ober man , 1983; Cohe n & Syme, 1985). Lo w levelsof social support make a person more vulnerable to mental i ll -ness (Sarason, Sarason & Lindner, 1983) while high levels ofsocial support are likely to protect a person's health (Gottlieb,1983), m ake it mor e likely that a person will stop smoking, loseweight (Janis, 1983), and persist at a task under unfavorableconditions (Sarason, Levine, Basham , & Sarason, 1983).The potent ial u t i l ity of the s tudy of individualism-col lect iv-ism for psychology s suggested by Figure 1. The data w ere takenfrom Henry and Stephens (1977) and suggest that two majorfactors may be causally related to differences in heart attackrates. The first factor may be labeled unpleasant life events.Trappist monks, who have pledged not to talk for the rest oftheir lives and communicate only in writing, which is usuallyless emotion al, have lower rates of heart attacks t han d o Bene-dictine monks, who do speak, or Benedictine priests, who areengaged in eve ryday life activities as teachers, counselors, andso on (Hen ry & Stephens, 1977).

The second factor is social cohesion and the probabi l i ty thata pers on will receive social support when unpleasant life eventsoccur. The contras t between the h ighly cohesive commu nity ofItalian immigr ants w ho settled in Roseto, Pennsylvania, and the"contro l" comm unity of Bangor, Pennsylvania, suggests thathigher probabilities of social support wh en unpl easan t lifeevents occur may be negatively related to heart attack rates.Taking the two factors together, in the man ner of a statisticalin teract ion , we might conclude that in cul tures where both un-pleasant life events are likely to oc cur and there is a low proba-bility of social support, the he art a ttack rates will be particu-larly high. Perhaps t hat is one way of explaining the largedifferences in heart attack rates between the Japanese in Japanand U.S. Caucasians and the rapid increase in the Japaneseheart at tack rate as Japan becomes m ore individual is t ic.The Ma rmo t and Sym e (1976) data are especial ly in teres t ing .Working with more than 3 ,000 Japanese Americans , they foun dthat those who were accul turated had five t imes as many heartat tacks as those who were unaccul tur ated ( that is, spoke Japa-nese at home, related with their children in the Japanese way),even thou gh the y statistically controlled for levels of cholesterol,exercise, smoking, and weight.In addi t ion to the data of Figure 1 , the He nry and Stephensvolume contains data on m any other d iseases . The a rgumen t isalso consistent with other kinds o f data. F or example, a studyby Frederichs , Chapm an, Nourjah , and Maes (1984) of cardio-vascular disease in Los Angeles shows that White and Blackage- and sex-adjusted rates of mortality from all diseases, butparticularly cardiovascular disease, are more than twice therates of Asian Americans . Some o f the d ifferences may be dueto diet an d weight control. But it is likely that som e of thedifference is due to the higher level of collectivism among theAsians. In a ny case, this is a testable hypothesis.The argument above suggests the relationships among lifeevents, culture, and disease presented in Figure 2. Unpleasantlife events (e.g., loss of loved ones) are hypo thesized t o have a neffect on disease. T heir i mpa ct is filtered by the in dividual 's psy-chobiological program. That program is a funct ion of geneticfactors (e.g., predispositions to arousal) and personality (e.g.,habitual patter ns o f behavior), which is a function o f culture(e.g., individualism with emphasis on self-reliance and compe-tition results in insecurity about coping; collectivism with em-phasis on close parent-chi ld relat ionships and social supportresults in less insecurity). Cultur e has im plications for the avail-ability of social support (e.g., close relatives are near; peopletake the time to provide social support). T he psychosoc ial stim-uli are perceived differently depending on the psychobiologicalprogram (a secure individual m ay be able to face them an d seethem as less threatening), an individual who has received ther-ap y may be able to perceive them as less threatening, and de-pending o n the level of threat there is more or less stress. Stresschanges the psychobiological pro gra m (increases the threa t lev-els that are filtered through) and is a precursor of disease thatincreases the probability of disease. Once a disease has set in,it is itself an unpleas ant life event and cha nges the psychobiolog-ical progra m (e.g., decreases optimism).Collectivism has the effect of emphasizing collective coping(Kas hima & Triandis, 1986), w hich ma kes it easier for the indi-vidual to cope with u npleasa nt life events. Males in collectivist


6/16

328

Psychosocial (~Stimuli

TRIANDIS, BONTEMPO, VILLAREAL,ASAI, LUCCA

I,L,PsychobiologicalProgramTPersonalityTS o c i a l i z a t i o ntCultureIeneticFactors

I~ Perce )tion

Therapy

Availability ofSocial Support

|

Stress

lPrecursorof DiseaseDisease

Figure 2. Adaptation of he Kagan and Levi (1974) model o illustrate the link of culture and disease.

cultures make internal attributions for failure less frequentlythan do males in individualist cultures (Chandler, Shama, &Wolf, 1983).Collectivism is associated with emphasis on harmony (atleast within the ingroup, which is the most frequent setting forsocial interaction), which reduces the stress level of everydaylife (not talking has the same effect for Trappist monks). Collec-tivism also is associated with low levels of competit ion andhence less insecurity about being able to cope. One's positionin the social order is determined by birth, age, sex, and so on,and hence one does not have to struggle to acquire a position.There is more stability in self-ingroup relationships, and thatalso reduces stress. The combinat ion of more harm ony and lessinsecurity may lead to low levels of stress. Of course, there aredisadvantages to collectivism also, such as corrupt governmentofficials, low levels of economic development, and the like. Butthese factors are more remote and less threatening to the indi-vidual and are less likely to cause high levels of stress. Thus,other things being equal (e.g. , GNP per capita), the levels ofdisease in collectivist countries should be lower than those inindividualist countr ies (see Figure 1).

Figure 2 is an adaptation o f the cybernetic model of Kaganand Levi (1974). The emphasis of the model is on feedback andinteractions. Cultures provide particular ways of viewing theenvironment (Triandis, 1972) so that the very same events (e.g.,unemployment) may be perceived differently in different cul-tures. I f the event is perceived as acceptable, normal, tempo-rary, or expected it will have less stressful consequences than if itis perceived as unacceptable, unusual, chronic, or unexpected.Stress thus is a function of the way unpleasant life events areperceived.

Data taken from Henry and Stephens (1977) suggest furtherthat those exposed to steady stress who receive social supporthave lower blood pressure levels that do no t increase with age;those exposed to variable stress events and low social support

have somewhat higher levels of blood pressure that become in-creasingly higher with age.High levels of stress may reduce immunity to disease (Ader,1981). Henry and Stephens present data that the smoking-can-cer links are stronger when people are emotionally upset thanwhen they are not. A wide range of physical disease, from infec-tions to complications in pregnancy, apparently are linked toreductions in the effectiveness of the immune system.

One may wonder if the fast pace of life in Western societies isresponsible for the higher rates of heart attacks there than incollectivist cultures, such as Japan. But that is not the case. Lev-ine and Bartlett (1984) show that the pace is faster in Japanthan in the United States, yet the heart attack rates are lower inJapan.

Some aspects of individualism may be related to an evenbroader set o f social ills. Naroll (1983) reviews data from manysocieties that suggest that when the primary group (which hecalls the moralnet) is a normative reference group that providesstrong social ties, emotional warmth, and prompt punishment;is cultural ly homogeneous, active in gossip, and very likely toshame deviant behavior; uses rituals frequently; has memorablemyths, a plausible ideology, and badges and emblems of mem-bership, then the positive social indicators outweigh the nega-tive. The data he reviews suggest that strong moralnets arelinked to low rates of homicide, suicide, crime, juvenile delin-quency, divorce, child abuse, wife beating, and drug and alcoholabuse, and to good mental health. The negative correlates, how-ever, include dissatisfaction with the family life (e.g., too manyduties), low economic development, and low GNP per capita(see also Middleton, 1963).Perhaps the major disadvantage of collectivism is in the polit-ical domain. Collectivists, particularly if they are centered onthe family as their major collective, tend toward actions thatbenefit the family rather than the broad public good. Those inpower act mostly to benefit themselves and their ingroup and


7/16

INDIVIDUALISM AND COLLECTIVISM 32 9often disregard the publ ic good. Extensive enrichment of theingr oup is usually associated w ith ineffective political behavior(Banfield, 1958).

In order to reach a bet ter unders tanding of the m eaning o f theindividual ism-col lect ivism construct , Hui and Triandis (1986)asked social scient is ts from every cont inent to respond to aquestio nnaire "as if " th ey were individualists and collectivists.There w as considerable agreement amo ng the 49 psychologistsand anthropologis ts who responded. Th e ma jor themes associ-ated with the con ceptio ns that social scientists have abo ut theseconcepts were used by Hui (1984) in construct ing a scale thatmeasu res individua lism and collectivism (INtX~OL). Triandis etal. (1985) used that scale together with other scales and foundthat a l locentr ic individuals in the U nited States are low on ano-mie, alienation, and loneliness; perceive that they receive moresocial support , o f better qual i ty; and value mos t cooperat ion,equal i ty and honesty. Idiocentric individuals in the UnitedStates showed the opposi te pat tern and valued a comfortablelife, competition, pleasure, and social recognition.

In evaluat ing the individual ism-col lect ivism constructs, w emus t take into a ccou nt the complexi ty o f the suggested relat ion-ships . Neoindividual ism is found mo st ly in no rth and west Eu-rope, No rth America, and those cul tures that were heavi ly in-f luenced by northern Europe. Col lect ivism is found in most o fthe othe r regions of the world. Proto-in dividu alism is still fou ndin many regions that are undeveloped (e.g., the Arctic, deserts,jungles). Neoindividualism seems to favor long-lasting, demo-cratic regimes and is reciprocally linked to attluence. However,there is a s t rong suspicion that some form s of social pathology(e.g., high crime rates) are also linked to extreme neoindividu-alism (NaroU, 1983). Collectivism may be associated with lowerlevels of st ress , but i f the ec ono my o f the co untry canno t sup-port mod ern medicine the heal th advantages of low stress levelsare lost, and dea ths o ccu r early. Ex trem e collectivism (e.g., thecul tural revolut ion in China) seems undesirable by any cri te-rion. Clearly, the picture is very complex, and it will take de-cades of research to unravel i t .

The first step in the long jou rne y suggested by this researchprog ram is to measure the constructs .

As mentio ned earlier, H ui (1984) developed proc edur es basedon at t i tude scales to measure the construct . Hui 's bes t i temsand addi t ional i tems that conformed to the theoret ical under-s tanding o f the construct , as well as i tems suggested by researchcollaborators from several countries (see following), were thenused in s tudies o f convergent and discriminant val idi ty (Tri-andis et a l ., 1985) and for cross-cul tural compa risons (Triandiset al., 1986). The latter study tested mostly student samplesfrom Chile , Costa Rica, I l l inois, Cal i fornia , Ho ng Kong, Ind o-nes ia , India , Greece, The Netherlands , and France. In mostcases, 100 students o f each sex responde d to 21 attitude item sand 16 rat ings of how m uch at tent ion they paid to the views ofvarious ingroup members , as wel l as two scenarios that mea-sured concern for others . A factor analys is based on an N of1,484 obtained four factors: (1) Self-Reliance With Hedonism(35% o f comm on variance; most ly i tems such as those in Table1, Factor 1, but without the co mpeti t ion i tems and w ith the he-donism i tems), (2) Separat ion F rom Ingro up (19% o f variance;items 15 and 22 in Table 1, Factor 2), (3) Family Integ rity (12%of variance; items reflecting close pare nt-s ubje ct relationships),

and (4) Interdependence W ith Sociabi l i ty (11% of variance;items 17 and 18 of Table 1).

A discriminant funct ion analys is a t temp ted to get the maxi-mu m di sc r imina t ion among the na t ional s amples . The m os tdiscriminat ing set o f items was Fam ily Integri ty. The locat ionof the nat ional samples on th at dim ension w as similar to thelocat ion of these countries on H ofs tedc 's (1980) individ ual ism-col lect ivism factor (rank-order correlat ion of .73, p < .05). TheUnited States was low on Fam ily Integri ty, high on Interdepen-dence With Sociabi li ty, high on Separat ion Fr om Ingroups , andhigh on Self-Reliance With Hedonism. If we assume that theUnited States is a prototypical individualist culture, this maybe the profi le o f individualism.

T h e S t u d ie sStu dy 1: Measu ring Idiocentrism in the United StatesThe present s tudy was des igned to probe the individual ismconstruct more deeply and more appropriately for the United

States by at tempting to operat ional ize i t. We used the themesextracted from the Hui and Triandis (1986) s tudy to constructitems. Specifically, the first facet was "sacrifice" (subo rdina tionof personal to ingroup goals; ingroup regulat ion of behavior)versus "hedonism" (doing what is sat is fying in terms of per-sonal goals ; behavior regulated by the individual) . T he secondwas "extens ion of self to ingrou p" versus "self as dis t inct en t i tyf rom in grou p" which a l so inc luded themes of in te rdependenceversus self-sufficiency. The th ird was " con cer n" fo r ingr oup (in-group is center of psychological f ie ld, ing roup har mo ny is im-portant , com mo n fate is sal ient) versus self-rel iance and emo -t ional dis tance from the ingroup.Method

Subjects. Three hun dred subjects from the psychology subject poolparticipated for course credit. Since the questionnaires w ere counter-balanced and the y required mo re time than was available to the subjects,the Ns for the various parts o f he questionnaire are not equal. They arementioned in the results section.Instruments. One questionnaire consisted of 158 items. It included63 items from the INDCOL Hui, 1984) and the above-mentioned 95theoretically constructed items. The 95 items must be considered asU.S. emics (culture-specific ideas). A second questionnaire included145 items that provided additional measurements of the tendency to-ward alloccntrism or idiocentrism, previously used and validated byTriandis et al. 098 5) and 13 new items tha t subsequent discussion sug-gested. The 145 items include d five scenarios, in which ind ividuals wereasked to imagine that they were in a particular social situation an d m ustdecide what to do. For example, in one of the scenarios the individualwanted to take a long trip, w hich various ingroups would find inconve-nient. T he subjects were asked to judge the extent they w ere likely totake the views of parents, spouses, close relations, clo se friends, acquain-tances, neighbors, and cow orkers into ac coun t in reaching a decisionabou t taking such a trip.A third questionnaire was an adaptation o f the Kluckhohn andStrodtbeck (196 l) instruments, and consisted of nine situations whichcould be responded to by (a) consulting an authority (lineal orienta-tion), (b) consulting peers (collateral orientation), or (c) using the bestinformation th at one has and deciding on on e's own (individualism).


8/16

330 TRIANDIS, BONTEMPO, VILLAREAL, ASAI, LUCCAResults

A note on the naming offactors. In this and the other studiesdiscussed in this article, we used the most accurately descriptivelabel for each factor tha t we were able to devise. For example,if the items that had high loadings referred only to family, weused "family," but if they also included items that referred toother groups (e.g., coworkers) we used the more abstract termingroup. We examined the factors of different analyses veryclosely, and if most of the highly loaded items were the same b utsome items that h ad high loadings on one factor did no t have ahigh loading on another, we at tempte d to use slightly differentnames for the two factors. Thu s, Su bordi nati on of Own Goalsto Goals o f Others is not the same factor as Subordin ation ofPersonal Goals to In group Goals, suggesting that man y of theitems are the same but perhaps as man y as a quarter of the high-loading items of one factor are no t the same as the items o f theother factor. The reader should look at the labels as suggestiveof the content, n ot as if the factor is a reality (such as an individ-ual's weight or height). Factors are summar ies of observationscompara ble to the average of a dist ribution. Just like the state-ment "the average number of children in this sample is 1.8"does not impl y that there are any families with that m any chil-dren, so the existence of the exact factor in a given sample isonly an appr oxima tion to the underlyin g reality.The structure of idiocentrism in the United States. We hadtoo ma ny items (95 new items and 63 INDCOL tems = 158) tobe able to obta in a stable factor analytic solution, since stabilityrequires more subjects than items (how many more is contro-versial; see, for example, Thu rsto ne, 1947; J6reskog & S6rbom,1979; Harman, 1967). Therefore we devised a procedure foreliminatin g items. First, we noted the communalities of theitems in the factor analysis of the 158 158 correlation matrix,based on 188 observations per variable. The communalitieswere the R2s, which means they reflected the extent each itemwas related to all other items. Since we were dealing mostly withidiocentric items, the com munali ty is a measure of the extentthe ite m dealt with the in dividu alism construct. Second, wemade sure that items that h ad high loadings on each of the threefactors obta ined in this factor analysis were selected for the nex tanalysis. Thus, we selected 20 items from each of the three fac-tors. These items were those with the highest communa lities.

These 60 item s were then subjected to a new factor analysis.This time the 60 60 matrix of correlations, based on 188 ob-servations per variable, was subjected to a principal-axis factoranalysis, with R 2 as com munali ties, and the factors were rotate dby the varimax method. This factor analytic procedure wasused in all the analyses to be reported in this article, unless oth-erwise noted. Table 1 presents the wording of the items thatloaded highly on the three factors identified by that second anal-ysis, as well as the percentage o f total varian ce accoun ted for byeach factor.

Inspection of Table 1 suggests that the first and third factors(Self-Reliance With Com petit ion and Distance Fr om Ingroups,respectively) are A meric an emics, whereas the second (Conce rnfor Ingroup) is eric (because the it ems are those tha t H ui devel-oped using Hong Kong Chinese and U.S. subjects). Further-more, we see that the items we selected fall into two of the three"theoretically" dete rmine d patterns: self-sufficiency and dis-

Table 1Wording of Three Factors of Study 1Factor 1: Self-RelianceWith Competition (35.2% of total variance)

1. If the group is slowing me down, it is better to leave it and workalone.2. To be superior a man must stand alone.3. Winning s everything.4. Only those who depend on themselves get ahead in life.5. If you want something done right, you've got to do it yourself.6. What happens to me is my own doing.7. I feel winning s important in both work and games.8. Success s the most important thing in life.9. It annoys me when other people perform better than I do.10. Doingyour best isn't enough; it is important to win.11. In most cases, to cooperate with someone whose ability is lowerthan oneself s not as desirable as doing the thing on one 's own.12. In the long run the only person you can count on is yourself.

Factor 2: Hui's INDCOLScaleJConcern for Ingroup(13.7% of total variance)13. It is foolish to try to preserve resources for future generations.(Reversed)14. People should not be expected to do anything or the communityunless they are paid for it. (Reversed)15. Even if a child won the Nobel Prize the parents should not feelhonored in any way. (Reversed)16. I would not let my parents use my car (ifI had one), no matterwhether they are good drivers or not. (Reversed)17. I would help within my means ira relative told me that s(he) isin financial difficulty.18. I like to live close to my friends.19. The motto "sharing is both blessing and calamity" is still appli-cable even if one's friend is clumsy, dumb, and causing a lotof trouble.20. When my colleagues ell me personal things about themselves,we are drawn closer together.21. I would not share my ideas and newly acquired knowledge withmy parents. (Reversed)22. Children should not feel honored even if the father were highlypraised and given an award by a government officialfor hiscontributions and service o the community. (Reversed)

Factor 3: Distance From Ingroups (12.0% of total variance)23. I am not to blame if one of my family members fails.24. My happiness is unrelated to the well-beingof my coworkers.25. My parents' opinions are not important in my choice of aspouse.26. I am not to blame when one of my close friends fails.27. My coworkers' opinions are not important in my choice of aspouse.28. When a close friend of mine is successful, t does not really make

me look better.29. One need not worry about what the neighbors say about whomone should marry.Note. INDCOL Individualism-Collectivism.

tance from ingroup. The h edonism theme, however, which wehad incorporated in our item generation, did no t emerge as aseparate factor. Fina lly, compet itio n was present in the first fac-tor, suggesting (consistent with Hsu's, 1983, view) that at leastin the Uni ted States, self-reliant indiv idual ism s linked to com-petition.

One might expect that people who are allocentric will seethemselves as more similar to their ingroups than people who


9/16

INDIVIDUALISM AND COLLECTIVISM 331are idiocentric. To test that hypothesis, we asked the subjects tojudge their sim ilarity to various ingroups on various topics, forexample, "How similar are your views to the views of yourfr iends concerning the kind of education you should have?" Thesimilarity judg men ts were sufficiently homo geneous after ite m-total correlations to allow construction of a Similarity to In-group s scale (no t show n in Table 1, to preserve space).

We also wondered if people who are allocentr ic might admitthat they are payin g a lot of attention to the views of their in-group s about m an y social issues. A gain, we found th at the at-tention items did fo rm a scale: Attention to the Views of In-group s (not shown in Table 1, to preserve space).Kluckhohn-Strodtbeck items. The lineal, collateral, and in-dividualistic judgments to the Kluck hohn -Strodtbe ck (1961)situations were given in thre e ways: (a) Wha t does the sub jectthink is the best way and (b) second best way, and (c) what doesthe subject think "others in the com mun ity" think is the bestway to deal with the particular situ ation (e.g., consult an au thor-ity, ask one's peers, decide on one's ow n). Factor analyses of thenine s cores (a, b, and c for the lineal, collateral, and individualis-tic response of each item) showed that there were four factors,the first of which was Individualism -Collaterality, and this ac-counted for most (35%) of the comm on variance.Second-order actor analyses: Wha t is the "bes t" measure ofidiocentrism in the United States. The highest loading itemsfrom each of the above-mentioned factor analyses weresummed and entered into a second-order factor analysis, inwhich the factor scores from the first round of factor analysesbeca me the new variables o f the second-order factor analysis.However, since this required subjects who had answered allquestionnaires, and most of our subjects had no t f inished an-swering all questionnaires, the N of this analysis is only 34.However, since factors were entered in the analysis we had o nly14 variables, and the 14 variable/34 number of observationsratio is sufficient for stability. We entere d fac tor scores for thethree factors of Table 1 and also the Similarity to Ingrou ps andAttention to the Views of Ingroups scales, the Trip, Lottery, In-vestment, Help, and Feeling Honored if an Ingroup MemberWins the Nobel Prize scenarios, and the four items (what sub-ject think s is best and second best, for the individualist and col-lateral answers) of the Individ ualism-C ollaterality factor fromthe Kluckhohn-Strodtbeck instrument. The Cronbach alphareliabilities of these 14 variables ranged fro m .62 (lottery sce-nario) to .92 (attention to the views of others), w ith a medi anof.83.

The first factor of the second-order factor analysis may becalled Subordination o f Own Go als to the Goals o f Others. I taccounted for 37% of the total variance or 47% of the com monvariance and had high loadings on all of the scenario factors(except Feeling Honored if an Ingroup M ember W ins the NobelPrize) and the A ttention to the Views o f Others factor.The second factor may be called D istance From Ingroups. I taccounted for 20% o f the total variance and 25% of the comm onvariance. Scales such as Self-Reliance With Competition andEmotional Detachment had high loadings on this factor, whilethe Feel H onored if Ingroup M ember Wins N obel Prize sce-nario had a low loading.The th ird factor may be called Individualism versus Collater-ality, since it had positive and negative loadings on the scales,

which ind icated that the first and second choices were collateral.I t accounted for 17% of the total and 18% of the co mm on v ari-ance.

The fourth factor accounted for 10% of the total variance andhad a .81 loading on the Concern for Ingroup factor (the HuiINDCOL scale items) and a .55 loading on the Lending Mon eyto Ing roup Mem bers scenario factor. Given the disparity in theloadings of this doublet factor, it may b e called simpl y Concernfor Ingroup.Discussion

These data suggest that U.S. idiocentrism is a multifacetedconcept. The ingredients include more concern for one's owngoals than the ing roup's goals, less attention to the views of in-groups, self-reliance with competition, detachment from in-groups, deciding on one 's own rather than asking for the viewsof others, and less general concern for the ing roup. Of course,these particular themes emerged in U.S. data. It is not knownif the same themes or som e other combination of ideas wouldemerge in other cultures.

Furthe rmore, wh en ingroup and the individual are in conflictthere is a general trend fo r subjects in individualistic cultures toanticipate that people will do their own thing rather than whatis expected by the ingroup, an d the perceived cause of that be-havior is hedonism.These results are in agreement with the major themes ofAm erican indiv idualism identified by Bellah, Madsen, Sullivan,Swindler, and Tip ton (1985). We did an inform al conten t analy-sis of their book and extracted themes. We found that thesethemes were mostly present in our factors. Of course, theymerged in different mixtures. D istance fro m ingroups (our Fac-tor 3 in Study 1) is linked to their most freque nt theme. Con cernfor ingroup (our Factor 2) was not strongly present in theirwork. But the correspondence is good enough, given the verydifferent method s we have used, that it seem s that we are look-ing at a similar content area.So how should idiocentr ism be measured in the UnitedStates? T he results o f Study 1 suggest that the item s in T able1 and the scenarios dealing with Trip, Lottery, and Help (seeTriandis et al. , 1985, for exact wording) seem to be the bestmeasures.

Stu dy 2: Probing Collectivism a nd Individualism:Japan, Puerto Rico, an d ll l inois

The previous study and our theoretical analyses suggestedthat people will relate to group s differently in collectivist socie-ties, such as Japan an d Pu erto R ico (see Hofstede, 1980), thanin individu alist societies such as the Unit ed States (Illinois). Ex-actly what does this imply? We wondered if this implies thatpeople in collectivist cultures perceive themselves as being m oresimilar to their ingroups, as paying more attentio n to the viewsof their ingroups, and as being more concerned with the needsof their ingroups than do people in individualist cultures. Dopeople in collectivist cultures indicate mo re willingness to sub-ordinate their personal needs to the needs ofing roups?

To explore these ideas and essentially find the limits of thecollectivism and individualism constructs, we undertook a


10/16

33 2 TRIANDIS, BONTEMPO, VILLAREAL, ASAI, LUCCATable 2Percentage of Illinois and Japanese Samples Conforming tothe Ingroup's Wishes or Evading a Confrontation inResponse to Conflict Resolution Problems

Since we had a total of 145 responses on 5-point scales, we assumed thatthe mean of the 145 256 Japanese (N = 37,120) responses should bethe same as the mean of the 13,195 Illinois responses. We did mostanalyses with both the original scores and the standardized scores.Conforming Evading

When the other person is Japan Illinois Japan IllinoisParent 15*** 44 19 26Close friend 12*** 31 34* 24Close relative of owngeneration 4** 16 42 40Coworker 9* 18 39 36Neighbor 5 8 45 47Person hardly known beforethe incident 2** 13 37*** 22Person from another country 1"* 12 56 63*p < .05. **p < .01. ***p < .001.

study with students from Japan, P uerto Rico, and Illinois. Asai,in Japan, was also able to get data from an older sample (age40+) consisting of parents a nd relatives of the students he sam-pied.Method

Ninety-one Illinois, 97 Puerto Rican, and 150 Japanese students con-stituted the main samples. In addition, 106 older Japanese were studied,so that the total Japanese sample is 256. The samples were approxi-mately equally male-female.These samples responded to a questionnaire that measured allocen-tric tendencies in Illinois (Triandis et al., 1985), after double translation(Brislin, 1980) into Japanese and Spanish. The Japanese instrumentincluded 144 items. Forty-eight of these required the subject to rate ona 5-point scale his or her perception of similarity to various ingroups(parents, relatives,close friends, coworkers, persons of the same nation-ality as self) with respect to several topics (for example, views on reli-gion, tastes in music, political views). Forty-two items required the sub-ject to rate the amount of attention he or she pays to the views of theingroups mentioned, concerning various topics (for example, how tovote, where to shop, how to choose friends). Forty-seven tems requiredthe subject to indicate how he or she would behave in various situations.Previous work had shown these situations to reflect Concern for In-group, Distance of Self from Ingroup, and Subordination of Own Goalsto Ingroup Goals (Triandis et al., 1985).Finally, the Japanese and Illinois instruments included seven judg-ments concerning how the subject would deal with conflict with variousingroups: These conflict resolution problems depicted situations inwhich the individual wants to do one thing and the ingroup wants theindividual to do another. The individual has the choice of three coursesof action: (a) Do what the ingroup wants, (b) do what the self wants, and(c) evade the issue (avoid confrontation). Table 2 shows the percentagesof each sample that gave the a and c responses. It is clear that the Illinoissample is more conforming than the Japanese, and although the Japa-nese avoid confrontation and take evasive action in the case of closefriends or persons they have jus t met, they are not especially differentfrom Americans in that tactic.Preliminary analyses. An examination of the data revealed that theJapanese sample tended to use the middle of the scale, whereas the Illi-nois and Puerto Rico samples used the whole scale. This is consistentwith previous findings (Zax & Takahashi, 1967). To eliminate this re-sponse set, we converted the data to standardized scores Z = M - MSD

ResultsReliability. To conserve space, we report here mostly c om-

parisons between the Illinois and the Japanese samples andmentio n the Puerto Rican data only when they differ from theJapanese. One way to judge the reliability of the data, a nd alsothe adequacy of the translations, is to correlate the 145 mea-sures across the two cultures. We computed the correlation ofthe mea n Japanese with the mean U.S. response to each of the145 items. That correlation was .87. We also computed thatcorrelation for each of the sections of the questionnaire. Thecorrelatio n for the simil arity judg ment s was .78, for the atten -tion-to-the-views-of-their ngrou p udg men ts it was .84, and forthe scenarios it was .93. Th e best tra nslati ons (or fewest cul turaldifferences) appear to have been done where there was morecontext (the scenarios), as predicted by Werner and Campbell(1970).

The reliabilities of the subscales obta ined from the varioussections of the que stionn aire were good (alphas of .78 to .96).A factor analysis that used the Similari ty o Ingrou ps and Atte n-tion to the Views of Ingroups scales and each o f the scenariosas separate variables showed the same factor patterns as thoseobtained by Triandis et al. (1985). That is, Subordination ofOwn Goals to Ingroup Goals, Concern for Ingroups, and Dis-tance of Self from In group (each acco unti ng from 25% to 35%of the variance in each sample) was found i n all three cultures.Overall results. To get an overview of the results, considerthat o ut of 138 ratings that were done by the Japanese a nd theIllinois samples, 19 showed the Japanese t o be more collectivist,at p < .01 or bet ter via analysis of variance (ANOVA) (7 of theseinvolved friends, 2 invo lved comparison s of greater perceivedsimilarity with the "average Japanese" as opposed to the "aver-age Ame rica n" of the Il linois sample, 5 involved feeling morehonor if an ingroup membe r is honored, and 3 involved helpingcoworkers); 105 showed no differences; and 14 showed the Illi-nois means to be more collectivist (6 of these were due to thefact that the Japanese indicated that they pay very little atten-tion to their neighbors, do not feel similar to their neighbors,and do not feel committed to help their neighbors, and 5 re-flected Japanese disinterest in religion, which resulted in lesssimilarity and less paying atten tion to religious views than wasthe case for Illinois sample).

Anot her overall result was that the Japanese older adults per-ceived themselves as more s imilar to their ing roups than did theJapanese younger adults via ANOVA, F( I, 252) = 7.3, p < .01,and the females saw themselves as more similar to their in-groups than did the males; there was also an int eract ion of thetwo, with older females particularly likely to see themselves assimilar to their ingroups (p < .04).

The pattern of results shows no differences between the Japa-nese and Illinois samples for reactions to parents. Most of thecultura l differences occurre d for acqu aintanc es, coworkers, andfriends. Specifically, the Ja panese indicate d that they pay moreatten tion to the views of coworkers than d id the Illinois sample.


11/16

INDIVIDUALISM AND COLLECTIVISM 33 3Specif ic results. On the whole, the s imilari ty judgm ents were

very similar across the three cultures. W hen there was a differ-ence (e.g., Japanese seeing themselves m ore sim ilar to the aver-age fellow cou ntry ma n than the Americans did, which was ex-pected from the charac teristics o f the societies, that is , thegreater homogenei ty of Japan) i t could not be considered adifference relevant to collectivism-individualism.

Similarly, the hypo thesis th at collectivists will see themselvesalways as more under the influence o f ingroups m ust be re-jected. In ve ry specific cases it was suppo rted, such as the Japa-nese paying more at tent ion to the views of their coworkers thandid the I l l inois sample "whe n deciding where to shop and whatto buy," "when choosing an int imate fr iend ( including aspouse)," and "when deciding what kind o f work to do " So onthe whole there w ere no cultura l differences, bu t specific items,such as the importanc e o f the workgroups , did show the ex-pected results.

The hypothesis that in collectivist cultures people will reportthat they show more concern for their ingroup mem bers (corre-spon ding to Facto r 2 in the Triandis et al., 1985, study) wassupported in only 2 out o f 11 scenarios for the Japanese, a ndboth the I l l inois and the Puerto Rico samples showed the oppo-s i te pat tern, that is , more concern for ingroup than Japanshowed.

The hypoth esis that peop le in collectivist cultures will subor-dinate their needs to the needs of their ingroups more th anthose in individual is t cul tures wil l (corresponding to Factor 1in the Triandis et al., 1985, study) received mixed support. InJapan the hypothes is was supported in the case of 7 out of 18tests, and if the probabilities are com bine d, v ia Stouffer's (1949)test, that is significant at p < .001. But the Pue rto R icans did n otshow the same pat terns. Thu s i t does not seem to be a generalcollectivist pattern .The hypothesis that people in collectivist cultures will showless dis tance from their ingroups was supported for 6 out o f 13of the i tems for the Japanese, bu t the Pu erto R icans were likethe I l l inois sample. The com bined probabi l i ty supports the hy-pothes is for the Japanese only, a t p < .001.D i s c u s s i o n

The global characterizat ion of col lect ivism is inaccurate , a tleast when college students are studied. People in collectivistcul tures do not necessari ly conf orm more, feel more s imilar toothers , pay mor e at tent ion to the views of others , and/o r uni-formly subordinate their goals to the goals of others . Su ch re-sponses are much more selective. Thus, we see the Japanesepaying at tent ion to the views of coworkers and fr iends mo rethan the I l l inois sample, but the same phe nom enon was not ob-served in Puerto Rico. Such findings are consistent with thoseof Hui (1984), w ho identified different kinds of collectivists:family, coworkers, friend, neighbor, and so on. One may be acol lect ivis t in relat ion to o ne ingro up but not in relation to o thergroups .The Japanese feel honored w hen their ingroups are hono redand pay at tent ion to the views of some, but not a ll , ingroups;they subordinate their goals to the goals of some ingroups , butthey do not conform much.

The P uerto Rican s tudents , in L ucca 's subject ive judgm ent ,

are on the individual is t ic s ide of the neutral po int on the indi-vidual ism-col lect ivism dimension. That may explain why theyseem so close to the Illinois sample. However, on some itemsthey behaved m ore l ike individual ists than such an interpreta-t ion wou ld allow. A discuss ion of such deviat ions from expecta-tion will help us understa nd collectivism better.

The higher rates o f paying at tent ion to the view of othersfound in I l l inois relat ive to Puerto Rico requires a comment .One way to interpret this result is that in collectivist culturessuch as Puerto Rico, one interacts most ly with ingroup mem-bers who are e ither relatives, friends o f relatives, or friends o ffr iends . Since such relationships are d etermine d for the subject(i.e., one is born in a family and the family's friends becomeone 's fr iends) rather than b y the subject , one does n ot have todevelop high skills in cultivating relationships.

Paying at tent ion to the views of others and conf orm ing tothem are techniques for gaining acceptance by others . Suchtechniques are more important in an individual is t ic societywhere one 's ingroups are form ed by the individual . Th is inter-pretat ion is consis tent a lso with the f indings of Bond andCheung (1983) who s tudied the spontaneous self-concepts ofcollege s tudents in Japan, Hon g Kong, and the United States .They found, with the Twenty Statements Test, that the self-con-cepts of the U.S. sample had mo re "social content" than d id theself-concepts of the Far Ea stern sam ples.

The very same poin t can be m ade by examining in m ore de -tail the results of Triandis et al. (1986). While on "F am ily Integ -ri ty" the Asians and Lat in A mericans were highly collectivistand the European s and Nor th Am ericans were highly individu-al is t ic , on "Interdependence and Sociabi l i ty" that pat tern wasnot observed. In fact, the highest score on interdependence wasobtained from the I l l inois females!

Thus , i t seems appropriate to change our view of col lect iv-ism. It is not that people are less sociable in the individualisticcountries-- in fact i t seems that they are more sociable in theindividualistic countries than in the collectivist because theyhave to work hard to get into and remain in their ingroups .

Lucca found the Puer to R ican sample ' s nona t t en t ion to theviews of others surpris ing, because she felt that in Puerto Ricothe "el qu6 dir~n" (what wil l others say about one) is verys trong. Perhaps our choice o f behaviors in this s tudy was notopt imal . A dis t inct ion is needed between behavior tha t is pro-scribed, in that what others will say about one is crucial, andbehavior that is prescribed in different degrees by others . T hei tems we used on the A ttent ion scenarios (paying at tent ion tothe views of others on how to vote , on religion, on shopping, onchoosing a fr iend, on what kind of work to do, on wh at musicto l isten to, and on what k ind o f educat ion to have) are behav-iors abo ut whic h cultures m ay prescribe very little. In furth erresearch, other behaviors (e.g. , who to mar ry) ab out w hich cul-tures both p roscribe and prescribe should be used.

Japan is supposed to be o n the col lect ivism s ide of the indi-vidual ism dimension in Hofs tede 's (1980) data , with a score of46 ( the U.S. score was 91). Thus , Japan may be more on thecol lectivist side of the cul tural dime nsion th an Puerto Rico, andthis is why we did obtain some support for the hypotheses inJapan, though the support was limited, specific, and suggestiveof different kinds of col lect ivism.

The data o f Table 2 are informative in several ways and con-


12/16

33 4 TRIANDIS, BONTEMPO, VILLAREAL, ASAI, LUCCAsistent with the interpretation that conformity is not high incollectivist cultures. First, note that the Japanese are less con-formin g than are the Illinois subjects. Thi s is consistent withresults obtained with the Asch paradigm in Japan and theUnited States. Frager (1970), using the Asch approach , foundconformity rates of 25% in Japan, which is lower than the usualAmerican rate of about one-third. Also, the Japanese showedmu ch anticon formity in Frager's study. Second, note th e differ-ence in the percentages of conformity to parent and personfrom another country. T he percentages in the Japanese datashow a 15:1 ratio; in the Am erican data, a 4:1 ratio. T his isconsistent with the argument, presented by Triandis (1972),tha t in collectivist cultu res the difference n the behav ior towardingroup and outgroup members is very large, whereas in indi-vidualist cultures it is greatly attenuated. A closer look at theJapan ese data in Table 2 suggests that the Japanese have an "in -ner ingroup" (parent, close friend), an "outer ingroup" (closerelative, coworker, neighbor), an d an outg roup (person hardlyknown, person from anoth er country) , whereas the I l l inois sam-ple has a w ider inner in group (parent, close friend, close rela-tive, coworker), a small "outer ingroup" (neighbors), and an"outg roup" that is not especially different from the "outer in-group." It seems desirable for social psychologists to stu dy m oreintensively, in the future, the social distances of individualsfrom a variety of natural group s in their social environment.

Finally, the rejection of ma ny o f the hypothes es of this studyraises the question whether students in Puerto Rico and Japanare really allocentric. As menti oned earlier, Lucca judges thePuerto Ricans to be "on the individualist side of the mid-p oint."Asai reports that there is a large gap between students and theolder generation (which we found in the sim ilarity judgments).Japanese students go through a rather regimented and highlycompetit ive educational system until they are admitted intouniversity. O nce th ey are in university, for the first time in th eirlives, they can do " their ow n thing," and m an y do. This freedomlasts a few years, but once they join the wo rkforce they go backto collectivism. Thus, students may be a particularly inappro-priate s ample for studies of Japan ese collectivism, mu ch as theyare inappropriate for certain kinds of studies on other topics ofsocial psycholo gy (Sears, 1986).In any event, the data of this study tell us to restrict andsharpen our definition o f collectivism. We are not dealing witha broad co nstruct where similarity to all ingroups, attention tothe view of all ingroups, or concern with the needs of all in-group s is involved. The da ta suggest that we mu st consider eachingrou p and ea ch dom ain o f social behavio r separately, and col-lectivism defined as subordination to the ingroup's norms,needs, views, and em otion al closeness to ingro ups is very spe-cific to ingroup and to domain. Collectivism takes differentforms, such as the greater sense of doing one's dut y or n ot dis-rupting in terpersonal relationships within the ingroup, that arespecific to each culture.Stu dy 3." Replication of Relationships o fAllocentrism toSocial Support and ldiocentrism to Loneliness inPuerto Rico a nd Il l inois

The first study indicated which items might be most usefulin measuring U.S. idiocentrism. In addition, the study by Tri-

andis et al. (1985) suggested that allocentric persons in theUnited States tend to p erceive that they receive mo re and a bet-ter quality of social support, whereas idiocentric individualstend to repo rt that they are lonely. Are these findings relativelyrobust, so that they might be obtained in another culture?Puerto Rico is a good setting to answer this question for tworeasons. First, it is a Latin cu lture and m ost Latin cultures havebeen fo und to b e collectivist (H ofstede, 1980). Second, it is notso drastically different from the Unite d States that it differs inevery conceivable way, so that if he results show that on e cann otreplicate the U.S. findings this n onco nfirma tion o f expectationscould be explained by many other variables.

The item s used by Triandis et al. (1985) were translated intoSpanish and translated back into English, according to themeth od o f double translation (Brislin, 1980). Lucca exam inedthe items for appropriateness in Puerto Rico. She felt that thehighest loading items o f the Triandis et al. (1985) study weremeaningful in Puerto Rico.A total of 72 items that measured aspects of idiocentr ism-allocentrism were used. Th e design called for 100 subjects (50men an d 50 wom en) in each culture. We were able to get com-plete data from 97 subjects from the University of Puerto Rico(who also responded to m any o f the i tems used in Study 2 -that data is reported under Study 2) and 99 fro m the Universityof Illinois.

In addition to responding to the 72 items that m easured allo-centrism (Triandis et al., 1985), the subjects from Illinois andPuerto Rico respond ed to double-translated versions of the So-cial Supp ort Question naire developed by Sarason, Levine, Bas-ham, and Sarason (1983) and validated by Sarason, Sarason,and Lin dner (1983) an d the Loneliness scale of Schmid t andSermat (1983).Results

The first 21 items were factor analyzed separately, (a) fromboth the other i tems that were used in that questionnaire and(b) for each of the cultural groups. The Puer to R ico factors wereAffiliation (42.4% of comm on variance; high loadings on "I en-joy meeting and talkin g to m y neighb ors everyday, . . . Wh at Ilook forward to in a job is a friendly group of co-workers," etc.),Autonomy (31%; "One should live one's life independently ofothers," "I tend to do my own thing, and most people in myfamily do the same") , and Distance of Self f rom In group(26.7%; see Table 1, Factor 3). The Illinois factors were SelfoReliance With C ompetit ion (45.3% of com mon variance; seeTable 1, Factor 1), Distance of Self From Ingro up (29.1%; seeTable 1, Factor 3), and Kin-Interdependence (25.6%; see kin-related items in T able 1, Factor 2). In addition, the vario us sce-narios (which involved family memb ers in situations where thesubject had to act according to own or a family memb er 's goals)form ed their own factors.To test the hypothesis that allocentric subjects would be highin reported social support and idiocentrics high in loneliness,we correlated each subject's responses to the various factors(forming facto r scores consisting of the highest loading items)and the responses to the Social Support and the Lonelinessscales.We obtained four scores from the Social Support Question-naire: (a) num ber of relatives mentioned as providing support


13/16

I N D I V I D U A L I S M A N D C O L L E C T I V I S M 335(SSR), (b) number of nonrelatives (usually friends) mentioned(SSnR), (c) satisfaction with social support received (SSS), and(d) total social support received (SST). We found that the corre-lations between Social Support and Loneliness scores were sim-ilar in Illinois and Puerto Rico, but the size of the correlationswas attenuated in Puerto Rico. For example, SSR correlated-.48 (p < .001) with Loneliness in Illinois but only -.23 (p


14/16

336 TRIANDIS, BONTEMPO, VILLAREAL, ASAI, LUCCAIn short , the em pirical s tudies suggest that we need to con-

sider individualism and collectivism as multidimensional con-structs,

Wh en we analyze data w ithin cul ture we f ind factors emerg-ing that are s imilar to those that emerge across cultures . In thepresent s tudy we conducted within-cul ture factor analyses .These analyses reflect mo re faithfully the respon ses of the sub-jects in the particular sam ples. However, they increase the com -plexity of the findings. Th e U.S. subjects o f Study 1 gave threefactors. The first is Self Reliance W ith Co mp etition and is aculture-specific factor that is similar to Self-Reliance With H e-donism from the ecological analysis, but is clearly different inthat the competition items loaded together with self-reliance.The second factor is Conce rn for Ingroup, with i tems fr om bo ththe ecological Interdependence With Sociabi li ty and negativeloadings on i tems o f the ecological Separat ion Fr om Ingroupsfactor. The third factor is Dis tance From Ingroups , a factor thatdoes no t emerge across cul tures .

In Study 2 the factor analyses carried out with data from Ja-pan and I l l inois provided very s imilar factors. Th ey were nam edSubordina tion of Own Goa l s to Ingroup Goa l s, Concern forIngroups , and Dis tance o f Self From Ingroups . The f irs t is mor eabstract than the ecological factors , but the Concern for In-groups factor again is s imilar to the Interdependence W ith So-ciabi l i ty and is negat ively related to the Separat ion From In-grou ps ecological factors. Finally, the Distance of Self Fro m In-groups fac tor has much in common wi th the th i rd fac tor ofStud y 1, suggesting tha t Jap anese students are sufl~ciently indi-vidualistic to provide a factor that looks much like a U.S. indi-vidualistic factor.In Study 3, Puerto Rico provided a classic collectivist factor(Affiliation), a classic individualist fa ctor orthog onal t o the co l-lect ivist (Autonomy), and a Dis tance of Self From Ingro up fac-tor that was also found in Japa n and the U nited States. The U.S.sample o f Study 3 repl icated the factors of the U.S. sample ofStudy 1.

In summa ry, the main ecological factors emerge also in thefactor analyses of the within-cul ture data . But there are in addi-t ion cul ture-specific factors that do n ot ap pear in the ecologicalanalyses.One point can be mad e qui te generally: In both the ecologicaland the within-cul ture analyses and those done in al l the cul-tures so far, the emphases on individualistic and collectivistthemes are orthogonal . Furthe rmore, col lect ivism depends verymu ch o n which ing roup is present, in wha t contex t (e.g., lottery,trip), and what behavior (e.g., paying attention to the views ofothers, feelings similar to others, competing with others) wasstudied.Although the analyses presented here suggest considerablecomplexi ty in the subjects ' concept ions o f behavior towardmemb ers of various natural groups , i t i s impo rtant to em pha-s ize that w hether individual ism-col lect ivism is one dimensionor a mult idimensional co nstruct depends on the contex t of thes tudy. If one s tudies a bro ad range of values (as did Hofs tede,1980) then in th at c ontext individual ism-col lectivism is one di-mension. I f one focuses only on self- ingroup relat ionships (aswe did in the present studies), then a multidimensional struc-ture emerges. In short, i t depends on the distance between theobserver and the data . Jus t as when observing a town from a

satel li te on e m ay gain the impress ion of a l ine spread along ahighway whereas when looking at the tow n fr om a close dis tanceone can see other s t reets and neighborhoods that are far fromthe highway, so in this case one observes mor e comp lexity fr oma closer vantage poin t.

Finally, the replication o f the findings ofTria ndi s et al. (1985)that a l locentr ic persons report m ore social support an d perceivea bet ter qual i ty of such suppo rt while idiocentric persons reportbeing mor e lonely suggests that the constructs may have rele-vance for our unders tanding o f the relat ionship of cul ture andhealth. It seems plausible (see Cohen and Syme, 1985) that,other things being equal, allocentric individuals m ay have betterheal th than idiocentric individuals . I f GN P per capi ta is keptconstant (and tha t is essential because i t correlates abou t .8 withindividualism [see Hofstede, 1980; Triandis, 1987]) people incol lectivis t cul tures should show fewer symp tom s o f s t ress andsmaller rates o f stress-related diseases (e.g., cardio vascula rdeath rates) than people in individualist cultures. That expecta-t ion seems supported by the data reported by Frederichs et a l.(1984).

In conclus ion, this art ic le has provided an analysis of the con-s tructs col lectivism and individualism and the hypotheses con-cerning the way these constructs are l inked to social phenom -ena, social behavior, and health. It has presented, also, proce-dures for the mea surem ent o f the constructs . I t has shown thatsome o f the s impler ideas abou t wha t col lect ivism m eans mustbe discarded and that a more complex unders tanding of theconstruct is possible. Finally, it has shown that within culture,al locentr ic individuals experience more social support and abet ter qual ity of support tha n do idiocentric individuals , w hotend to repo rt that they are lonely mo re than al locentr ic personsdo. Thus, cultural and personality differences exist in the wayself and ingr oup a re related. T his is a first step towar d a betterundersta nding of how culture affects social behavior.

R e f e r e n c e sAdelman, I., & Morris, C. T. (1967). Society, politics and economic de-velopment: A quantitative approach. Baltimore: John s Hopkins Uni-versity Press.Ader, R. (Ed.) (1981). Psychoneuroimmunology New Y ork: AcademicPress.Antill, J. K. (1983). Sex role complemen tarity versus similarity in mar-ried couples. Journal o f Personality and Social Psychology, 45, 145-155.Bakan, D. (1966). The duality of human existence. Chicago: Rand Mc-Nally.Banfield, E. C. (1958 ). The mor al basis o f a backward society Glencoe,IL: Free Press.Barry, H., Child, I., & Bacon, M. (1959). Relation o f child training tosubsistence economy .American Anthropologist, 61, 51-63.Bellah, R. N., Madsen, R., Sullivan, W. M., Swindler, A., & Tipton,S. M. (1985). Habits o f the heart: Individualism and com mitm ent inAmerican life, Berkeley,CA: University of California Press.Berger, P., Berger, B., & K elner, H. (1973). The homeless mind. NewYork: Random House.Berry, J. W. (1979). A cultural ecology of social behavior. In L. Berko-witz (Ed.), Advances in experimental socialpsychology (Vol. 12, pp.177-207). N ew York: Academic Press.Bond, M., & Cheung, T. (1983). College students' spontaneous self-con-cepts: The effect of culture am ong respondents in Hong Kong, Japan,

and the U.S. Journal of Cross-Cultural Psychology, 14, 153-171.


15/16

INDIVIDUALISM AND COLLECTIVISM 337Bon tem po , R., Lobel, S., & Triandis, . C. (in press). omp lia nce an d

value internalization mo ng Brazilian and U.S. students. ournal ofPersonality an d S ocial PsychologyBreckler, S., & Gree nwald , A. G. (1986, April). Social motivational con-ceptions of the self. Paper presented at a meeting of Polish and U.S.social psy

Documents

Individualism Collectivism 1988