Rethinking Individualism And Collectivism Evaluation Of -Books Download

Rethinking Individualism and Collectivism Evaluation of

2020 | 9 views | 70 Pages | 1.10 MB

these processes led to a Western cultural focus on individualism that is more salient in countries and ethnic groups with a Protestant heritage, applying the idea of Western individualism to both cross-regional and within-country comparisons of ethnic groups with different cultural heritages. Thus, within the United States it is



4 OYSERMAN COON AND KEMMELMEIER
role of individual choice personal freedom and self actualization self concept well being attribution style and relationality Last
e g Inglehart 1997 Sampson 2001 Researchers assume that we draw conclusions and articulate emerging questions about
these processes led to a Western cultural focus on individualism implications of a culture frame for psychology generally
that is more salient in countries and ethnic groups with a Protestant Methodologically we collected all English language literature
heritage applying the idea of Western individualism to both cross published since 1980 on individualism and collectivism that either
regional and within country comparisons of ethnic groups with assessed these constructs directly or related them to the basic
different cultural heritages Thus within the United States it is psychological domains of self concept well being attribution
commonly assumed that European Americans are higher in indi style or relationality We meta analyzed individualism and col
vidualism and lower in collectivism than are members of ethnic lectivism studies contrasting European Americans with other
minority groups e g Freeberg Stein 1996 Gaines et al 1997 groups and reviewed the empirical literature linking individualism
Rhee Uleman Lee 1996 Taken together current theorizing in and collectivism with basic psychological processes Together the
cultural psychology portrays European Americans as the most meta analyses and literature reviews clarify the extent to which
individualistic group European Americans are in fact uniquely high in individualism and
Considering European Americans the gold standard of individ make sense of themselves their lives and their relationships with
ualism seems at first glance unremarkable Certainly rugged others in terms of the values of individualism
individualism has been an American hallmark at least since de As we demonstrate in subsequent sections contemporary Amer
Tocqueville s 1835 1969 classic analysis of America that linked ican psychological research is particularly suited to an individual
individualism with individual rights and freedom equal opportu istic worldview and may not necessarily fit as a universal model of
nity and limited government Lukes 1973 Others have also human behavior to the extent that other peoples or regions of the
linked American individualism to the Puritans the founding fa world are sharply different from Americans in individualism and
thers the birth of a market economy and the vast American collectivism For example self concept research is dominated both
frontier Curry Valois 1991 From the beginning Americans by a focus on self esteem and by the belief that attainment of
have been enjoined to value life liberty and the pursuit of personal happiness is a basic motivational drive e g Baumeister
happiness and to think of themselves as separate and independent 1998 Likewise person perception and cognitive processes are
individuals isolated from others In de Tocqueville s words Such understood in terms of stable traits and equity is viewed as the
folk owe no man anything and hardly expect anything from any basis for successful relationships e g Triandis 1995 These
body They form the habit of thinking of themselves in isolation research frames fit individualistic not collectivistic conceptions
and imagine that their whole destiny is in their own hands of human nature To preview our findings our analysis of the
1835 1969 p 508 literature suggests first that differences in individualism do exist
Indeed modern American cultural icons continue to articulate and second that the influence of cultural frame is better docu
this belief in individualism Personal privacy individual rights mented for the domains of relationality and attribution than for
and personal freedoms are extolled personal pleasure and auton other domains
omy are valued and every American is exhorted to create a
personal private and unique self e g Sampson 1977 1988 CULTURAL ORIENTATIONS DEFINITIONS AND
Americans celebrate individualism as a uniquely American char
PSYCHOLOGICAL CONSEQUENCES
acteristic an integral part of their culture Yet in spite of the
seeming consensus that European Americans are the prototype In this section we provide an overview of individualism and
defining individualism we are aware of no systematic test of the collectivism as cultural orientations to provide a framework for
underlying assumption that European Americans value or behave reading the article as a whole
more individualistically than others
Furthermore there is a clear tension between the assumption Individualism
that European Americans are uniquely high in individualism and
low in collectivism and the assumption that the psychological The core element of individualism is the assumption that indi
models developed within this cultural frame of self concept con viduals are independent of one another From this core a number
tent and functioning well being attribution style and relational of plausible consequences or implications of individualism can be
ity are universal models not simply models derived from and discerned One question we explore further is whether research has
applicable to an individualistic worldview Therefore the basic empirically validated these plausible consequences or implications
aim of the present review is to address two questions a Are and whether these plausible consequences are in fact universally
European Americans higher in individualism and lower in collec part of individualism
tivism than people from other societies b Are theoretically Hofstede 1980 defined individualism as a focus on rights
derived implications of individualism and collectivism for psycho above duties a concern for oneself and immediate family an
logical functioning borne out in the empirical literature To answer emphasis on personal autonomy and self fulfillment and the bas
these questions first we present an overview of the theoretical ing of one s identity on one s personal accomplishments Water
implications of individualism and collectivism for basic psycho man 1984 defined normative individualism as a focus on per
logical domains Second we provide a guide to how psychologists sonal responsibility and freedom of choice living up to one s
have studied individualism and collectivism Third we meta potential and respecting the integrity of others Schwartz 1990
analyze empirical literature assessing individualism and collectiv defined individualistic societies as fundamentally contractual con
ism Fourth we review the empirical evidence of an influence of sisting of narrow primary groups and negotiated social relations
individualism and collectivism on basic psychological domains with specific obligations and expectations focusing on achieving
INDIVIDUALISM AND COLLECTIVISM 5
status These definitions all conceptualize individualism as a collectivism as a social way of being oriented toward in groups
worldview that centralizes the personal personal goals personal and away from out groups Oyserman 1993 Because in groups
uniqueness and personal control and peripheralizes the social can include family clan ethnic religious or other groups Hui
Bellah Madsen Sullivan Swidler Tipton 1985 Hsu 1983 1988 and Triandis 1995 among others have proposed that
Kagitcibasi 1994 U Kim 1994 Markus Kitayama 1991 collectivism is a diverse construct joining together culturally
Sampson 1977 Triandis 1995 disparate foci on different kinds and levels of referent groups In
Given these definitions plausible consequences of individual this way collectivism may refer to a broader range of values
ism for psychology self concept well being attribution style attitudes and behaviors than individualism
and relationality are easily discerned First with regard to self Plausible consequences of collectivism for psychology self
concept individualism implies that a creating and maintaining a concept well being attribution style and relationality are easily
positive sense of self is a basic human endeavor Baumeister discerned First with regard to the self collectivism implies that
1998 b feeling good about oneself personal success and hav a group membership is a central aspect of identity Hofstede
ing many unique or distinctive personal attitudes and opinions are 1980 Hsu 1983 U Kim 1994 Markus Kitayama 1991 and
valued Oyserman Markus 1993 Triandis 1995 and c b valued personal traits reflect the goals of collectivism such as
abstract traits as opposed to social situational descriptors are sacrifice for the common good and maintaining harmonious rela
central to self definition Fiske Kitayama Markus Nisbett tionships with close others Markus Kitayama 1991 Oyserman
1998 Second with regard to well being individualism implies 1993 Triandis 1995 Second with regard to well being and
that open emotional expression and attainment of one s personal emotional expression collectivism implies that a life satisfaction
goals are important sources of well being and life satisfaction derives from successfully carrying out social roles and obligations
Diener Diener 1995 Markus Kitayama 1991 Third indi and avoiding failures in these domains U Kim 1994 Kwan
vidualism implies that judgment reasoning and causal inference Singelis 1998 Markus Kitayama 1991 and b restraint in
are generally oriented toward the person rather than the situation or emotional expression rather than open and direct expression of
social context because the decontextualized self is assumed to be personal feelings is likely to be valued as a means of ensuring
a stable causal nexus Choi Nisbett Norenzayan 1999 Miller in group harmony
1984 Morris Peng 1994 Newman 1993 Consequently in Third with regard to judgment causal reasoning and attribu
dividualism promotes a decontextualized as opposed to a tions definitions of collectivism suggest that a social context
situation specific reasoning style one that assumes social infor situational constraints and social roles figure prominently in per
mation is not bound to social context son perception and causal reasoning Miller 1984 Morris Peng
Last with regard to relationality individualism implies a some 1994 and b meaning is contextualized and memory is likely to
what ambivalent stance Individuals need relationships and group contain richly embedded detail Last with regard to relationality
memberships to attain self relevant goals but relationships are definitions of collectivism imply that a important group mem
costly to maintain Kagitcibasi 1997 Oyserman 1993 Theorists berships are ascribed and fixed viewed as facts of life to which
assume that individualists apply equity norms to balance relation people must accommodate b boundaries between in groups and
ships costs and benefits leaving relationships and groups when out groups are stable relatively impermeable and important and
the costs of participation exceed the benefits and creating new c in group exchanges are based on equality or even generosity
relationships as personal goals shift Therefore theorists assume principles U Kim 1994 Morris Leung 2000 Sayle 1998
that for individualists relationships and group memberships are Triandis 1995
impermanent and nonintensive Bellah et al 1985 U Kim 1994
Shweder Bourne 1982 PSYCHOLOGICAL STUDY AND ASSESSMENT OF
INDIVIDUALISM AND COLLECTIVISM
Collectivism
The previous section highlighted plausible consequences of
The core element of collectivism is the assumption that groups individualism and collectivism without operationalizing individu
bind and mutually obligate individuals From this core theorists alism and collectivism In this section we provide a guide to the
discern a number of plausible consequences or implications of study and assessment of individualism and collectivism based on
collectivism One question we explore further is whether research the past 20 years of usage Our focus is on literature emerging
has empirically validated these plausible consequences or impli since 1980 because references to individualism and collectivism
cations cross culturally and whether these plausible consequences in the psychological literature increased dramatically at about that
are in fact universal consequences of collectivism time and because in 1980 Hofstede published his highly influential
Although sometimes seen as simple opposites it is probably analysis of cultural frame in which individualism was a central
more accurate to conceptualize individualism and collectivism as focus
worldviews that differ in the issues they make salient Kagitcibasi
1987 1997 Kwan Singelis 1998 According to Schwartz Overview and Background
1990 collectivist societies are communal societies characterized
by diffuse and mutual obligations and expectations based on As conceptual frameworks individualism IND and collectiv
ascribed statuses In these societies social units with common fate ism COL reflect clearly contrasting worldviews Our review
common goals and common values are centralized the personal is seeks to shed light on the question What is the empirical valida
simply a component of the social making the in group the key unit tion of these theoretical frames especially with regard to the
of analysis e g Triandis 1995 This description focuses on assumption that Americans are a gold standard of high individu
6 OYSERMAN COON AND KEMMELMEIER
alism and low collectivism We attempted to include all relevant expected if cross cultural differences in IND or COL exist without
studies with American participants published since 1980 For the directly showing the connection to IND or COL or assessing IND
purpose of the present review we include as American both or COL to show that differences in IND or COL predict differ
Canada and the United States for three reasons First the field has ences in the variables of interest Confusion arises in the current
clearly assumed an approximate equivalence between the cultures literature because authors commonly describe their research in
of Canada and the United States in terms of IND and COL cross national terms although their data are at the individual level
Second the one empirical study we found examining this assump This use of a single cross group comparison to generalize about
tion does support this conclusion Kemmelmeier et al 2001 cross societal differences is a central limitation of the existing
Third nearly all of the experimental cross cultural self concept database
research contrasts Canadians and Japanese So as not to introduce Three additional limitations of our database are the narrow focus
ambiguity we specify which country our American samples come on undergraduates as research participants 1 single group contrasts
from both in the appendixes and in the text so the reader can and enormous heterogeneity in how researchers conceptualize and
distinguish the source operationalize IND and COL Using undergraduates clearly limits
generalizability to other segments of society Narrow focus on
Source of Data differences between European American undergraduates and un
dergraduates from either a single East Asian country or a single
The meta analyses reported in this article included studies as American racial or ethnic minority group limits generalizability to
sessing IND and or COL whereas the review of psychological other countries racial groups or ethnic groups The bulk of cross
implications of IND COL included all studies in which IND and or national research comes from comparisons of American under
COL were evoked as an explanation for self concept well being graduates with undergraduates from Japan Hong Kong People s
attribution style or relationality We also included studies using Republic of PR China or Korea The final limitation of our
parallel terms such as discussions of Western versus Eastern database involves heterogeneity of usage and conceptualization of
worldviews e g Choi Nisbett 1998 Miller 1984 Morris IND and COL in the literature This heterogeneity challenged our
Nisbett Peng 1995 Morris Peng 1994 To obtain articles ability to generalize about IND and COL and required us to
we first searched electronic databases e g PsycINFO ERIC and establish a working structure to integrate diverse approaches for
Dissertation Abstracts International for publications from 1980 this review We outline these usage and conceptualization issues
through 1999 using the following keywords individualism col and our working solution in the next section
lectivism independence interdependence self construal allocen
trism and idiocentrism To obtain unpublished or in press re Measuring IND and COL
search we used professional e mail lists As listed in Appendixes
A B and C 83 different studies were included in the meta Hofstede s 1980 Approach
analyses of IND or COL and 170 studies were included in the
One approach to operationalizing and measuring IND is to focus
review of psychological implications of IND and COL studies are
on cultural values assessed at the aggregate level emphasizing
listed in each relevant appendix
difference between cultural units In his groundbreaking effort
Hofstede 1980 surveyed samples of employees of the same
Limitations of the Current Data multinational corporation in 39 nations On the basis of the dif
ferent samples responses to a work satisfaction questionnaire he
Like any review of extant literature the nature of the current
generated country level indicators of IND conceptualizing it as a
knowledge base limits the conclusions and generalizations one can
function of workplace values Hofstede assumed that IND and
draw Although Hofstede 1980 explicitly focused on differences
COL formed a single continuum with low IND isomorphic with
between countries and not individuals this element of his work has
high COL Hofstede argued that IND was in part a reflection of
not caught on and most research studies individual differences
social structural conditions He demonstrated consistent associa
Some authors do focus on analysis of cultural representations such
tions between his aggregate measure of IND and country level
as news reports e g F Lee Hallahan Herzog 1996 however
indicators of the society gross national product GNP country
most current models focus on assessing IND at the individual level
latitude population size and density
see Triandis 1995 These levels of analyses are not contradic
Hofstede 1980 was careful to point out the limitations of his
tory and it is possible in principle to contrast social artifacts
research First he emphasized that his country level analysis of
groups and individuals differing in IND
IND could not explain individual behavior which he regarded as
The preponderance of individual level analyses is not surpris
a theoretically distinct problem Second he saw country level IND
ing Country level comparisons require enormous resources be
as an indicator embedded in a dynamic process of cultural devel
cause these analyses require the researcher to sample a sufficient
opment shifting as social structural and historical changes oc
number of distinct groups to allow for quantitative analyses Not
cur Thus Hofstede warned that his own results were not stable but
only must sufficient groups be sampled but these groups must also
rather shaped by the economic and historical circumstances of the
be at least reasonably representative of the society as a whole if
one is to generalize comfortably to a society In contrast to these
desired characteristics we found that cross cultural analyses typ 1
Over 80 of studies in the meta analyses used undergraduates three
ically contrast two groups of students Researchers most com studies used adults three used management trainees and two used man
monly use a single cross group comparison Comparisons focus on agers as did a similar percentage of studies in the narrative review 7 of
either showing that variables of interest vary in ways that might be studies used children and 12 used adults See Appendixes A C
INDIVIDUALISM AND COLLECTIVISM 7
1970s when he collected his data He was able to demonstrate this vague quantifiers such as very much agree or very important
flux by comparing responses collected a few years apart in his first Comparing responses to IND or COL scales cross culturally re
and second data sets These data demonstrate cultural shift for quires that researchers assume that what respondents mean when
example Japanese workers shifted toward higher IND Whereas they say that they agree is sufficiently similar cross culturally to
Hofstede s work was highly influential his focus on social struc make comparisons meaningful We found two studies Ji
tures and use of a national survey to assign cultural values to a Schwarz Nisbett 2000 Peng Nisbett Wong 1997 that
country has not become a common practice but see Vandello address this issue both comparing China and the United States
Cohen 1999 Perhaps this is because implementation of this Both studies suggested that researchers cannot easily assume com
method is time and resource intensive and because attention has mon understanding of what responses such as very important
shifted to the ways that cultural frames influence individuals mean and that scale use can systematically differ between coun
tries 2 A final limitation of the direct assessment approach is the
Current Approaches assumption of cross cultural convergence in the questions that
must be answered to tap into the underlying dimensions of IND
Three general approaches are in current use by researchers and COL To date few studies have applied strict psychometric
studying IND COL We termed these approaches applying Hof criteria to carefully examine equivalence in cross cultural mea
stede direct assessment and priming cultural frame None of surement see Bontempo 1993 Rhee et al 1996 for exceptions
these approaches dominate the field each has limitations and any Results have focused on specific instruments and as will be seen
decision to limit our review to particular approaches would have a large number of instruments and operationalizations are in cur
dramatically narrowed its scope For example omitting research rent usage
that did not assess IND COL would have eliminated most U S
India comparisons and all self esteem research because researchers Priming Studies
in these areas do not assess IND COL We outline each approach
with its strengths and limitations below Given the limitations in the applying Hofstede and direct as
sessment approaches it is not surprising that researchers have
sought alternatives An emerging alternative based in social cog
Applying Hofstede nition research involves efforts to prime IND or COL values or
Despite Hofstede s 1980 admonitions a large proportion of independent interdependent self definitions before assessing their
research uses his ratings of country level IND as proxies for IND effect on a dependent measure Social cognition research consis
rather than assessing IND directly Some researchers use the tently shows first that accessible knowledge influences behavior
ratings he provided others simply note that Hofstede found a and second that temporarily accessible and chronically salient
difference between two countries and then use this as the basis of knowledge produce equivalent effects in laboratory settings e g
their assumption that the two countries still differ in IND and that Bargh Bond Lombardi Tota 1986 These findings form the
their findings relate to this difference Either variant of the apply theoretical underpinning of priming research in cultural psychol
ing Hofstede approach makes at least three assumptions about ogy These priming techniques attempt to create an experimental
mean levels of IND as assessed by Hofstede that they are a analogue of chronic differences between cultural groups by tem
accurate across life domains e g self concept or well being b porarily focusing participants attention on different cultural con
stable over time and c relevant to individual level assessment tent or values
Researchers who use this approach also typically assume European There are two types of priming manipulations The first type
Americans are higher in IND than their comparison group usually aims at making a participant s IND and COL values salient and
East Asians Lack of empirical support for these assumptions the second type aims at making a general IND or COL worldview
makes this approach vulnerable to criticism salient In its simplest version cultural values are primed as
follows Experimental participants complete an IND COL scale
immediately prior to responding to the dependent variable and
IND COL Rating Scales control participants respond to the dependent variable prior to
A second common approach is to measure IND and COL at the completing the IND COL scale e g Oyserman Sakamoto
individual level and to correlate this assessment with individual Lauffer 1998 This technique allows researchers to compare
outcomes behaviors attitudes and beliefs When using this ap responses of participants who have just brought to mind their
proach to assess IND COL researchers typically ask respondents cultural values and beliefs with the responses of participants not
to rate how much they agree with or how important they find a list focused on their cultural values Studies of this sort can show the
of behaviors attitudes and value statements This approach avoids effect of culture by focusing participants attention on cultural
the assumptions required to apply Hofstede however it has lim values This technique allows researchers to study the effect of
itations as well First the direct assessment approach assumes that bringing cultural frame to mind and whether the strength of this
cultural frame is a form of declarative knowledge e g attitudes
values and beliefs that respondents can report on rather than some 2
Perceived options anchor responses so even a little individualism
set of more subtle and implicit practices and social structures that within a Chinese context would stand out and be rated more extremely
respondents cannot report on because these practices are deeply Peng Nisbett Wong 1997 Response scales themselves can influence
woven into everyday life and are a normal part of living Second judgments of Chinese and American respondents differently scale struc
this approach assumes cross cultural convergence in the meaning ture influenced American but not Chinese respondents judgment of their
assigned to scale response choices Response choices are typically own public behaviors Ji Schwarz Nisbett 2000
8 OYSERMAN COON AND KEMMELMEIER
effect depends on individual endorsement of cultural values and social cues Bontempo 1993 Kagitcibasi 1987 Oyserman
Therefore this priming technique provides greater clarity in causal 1993 Rhee et al 1996 Singelis 1994 Sinha Tripathi 1994
reasoning than a simple correlational approach in which research Triandis Bontempo Villareal Asai Lucca 1988 Following
ers correlate an IND COL scale and a dependent measure because Schwartz 1994 it seems more reasonable to view societies as
systematic variation in the strength of the relationship can be dealing with collective and individual oriented value choices sep
attributed to salience of IND COL A limitation of this technique arately This means any given society is likely to have at least
is that the IND COL scales are typically the same kind of direct some representation of both individualistic and collectivistic
assessments just critiqued therefore the previously noted limita worldviews However this formulation leaves open the question of
tions of direct assessment can be said to apply to this type of whether in all societies IND focus on the independent individual
manipulation as well and COL focus on duty and obligation to in groups necessarily
A second type of priming manipulation focuses participants carry with them all the related constructs described in the section
attention either on IND or on COL and compares their subsequent on implications Thus for example a focus on personal achieve
responses on a dependent measure In other words this technique ment may be multiply determined and not always be related to
temporarily influences whether one s IND related or COL related individualistic values just as seeking the advice of parents may be
values beliefs and cognitions are more likely to come to mind but multiply determined and not always be related to collectivistic
does not attempt to measure these values beliefs and cognitions values
directly In the first published application of this method Trafi
mow Triandis and Goto 1991 asked participants to describe the
Terminology in Current Use
ways they were either different from or similar to their family and
friends The first prime elicited individualistic or independent It is most common for research to refer to an individual focus as
self knowledge whereas the second elicited collectivistic or in individualism and to a collective focus as collectivism This simple
terdependent self knowledge More recently Gardner and col usage allows much diverse research to be thought of as an inte
leagues Brewer Gardner 1996 Gardner Gabriel Lee 1999 grated whole and is the most commonly accepted way of describ
primed IND and COL self knowledge by having participants read ing the phenomena Although this simple usage introduces heter
a brief paragraph and circle either first person singular or plural ogeneity because it distinguishes neither differences in IND and
pronouns They found that focusing on pronouns like I me and COL scales nor differences in levels of analyses we feel that the
myself made independent self knowledge more accessible advantage of continuing to use these general terms outweighs the
whereas a focus on pronouns like we us and ourselves drew disadvantages because research primarily refers to the individual
attention to a person s embeddedness in a collective and thus made level of analyses and because scales contain many common
interdependent self knowledge more salient This technique has features
the advantage of avoiding measurement altogether by experimen That is not to say that a number of other terms are not applica
tally creating IND or COL differences in focus of attention This ble For example to clarify their focus on the individual level of
technique avoids problems of direct assessment and unlike the analyses Markus and Kitayama 1991 proposed the terms inde
applying Hofstede approach allows researchers to study culture as pendence and interdependence to describe the self related aspects
a dynamic process Like other approaches priming is limited by of IND and COL Triandis coined the terms idiocentrism and
lack of data comparing results using the different techniques allocentrism as the individual level equivalents of country or
leading to ambiguity as to the robustness of findings across mea society level IND and COL Triandis 1995 Whereas most
surement paradigms empirical research is at the individual level neither of these
alternative conceptions has fully captured the literature and
Summary different terminologies persist For example researchers de
scribing the effects of situationally primed IND COL refer to
In reviewing the implications of IND COL for basic psycholog priming individualism collectivism Kemmelmeier Wieczorkowska
ical domains we review studies that used any of the approaches Erb Burnstein in press Study 3 Oyserman et al 1998
outlined above We chose this broadly inclusive approach for three independence interdependence Gardner et al 1999 idio
reasons a Each approach has limitations b none dominates the centrism allocentrism and private self collective self Trafimow
field and c method and typical country of comparison differed et al 1991 To avoid unduly limiting the scope of this research
by content domain therefore picking one method would also result we include all relevant research however labeled We refer to the
in reducing the scope of our reviews Where possible we highlight effects of IND and COL using the more specific terms as needed
convergence and divergence in findings based on these to avoid confusion
approaches
Existing IND COL Scales
Operationalizing IND COL
Types of Scales
Although the assumption of IND being the conceptual opposite
of COL may be intuitively appealing an accumulation of recent Hofstede 1980 assessed only IND because he assumed that
research suggests this simple approach does not sufficiently rep COL is equivalent to low IND This bipolar single dimension
resent the impact of IND and COL on basic psychological pro approach continues to have some supporters see Appendixes
cesses Instead IND and COL are better understood as domain A C When researchers contrast countries on IND COL they
specific orthogonal constructs differentially elicited by contextual assess only IND or COL in a substantial minority of cases We
INDIVIDUALISM AND COLLECTIVISM 9
found this approach in almost a third of the studies in the meta in another scale or items from so many previously published scales
analyses 26 of 83 studies If anything this bipolar single dimen that no single source was primary Eleven of these distinct scales
sion approach is more popular among researchers studying psy measured IND COL as a single bipolar construct and 16 measured
chological implications of IND COL Of the 170 studies IND or independence or idiocentrism and COL or interdepen
examining psychological implications of IND COL only 87 as dence or allocentrism as orthogonal constructs In some cases
sessed IND or COL and only 40 of these n 36 assessed both authors refer to subscales of IND COL For example Triandis
IND and COL used more specific terms such as self reliance with hedonism
separation from in groups and family integrity Triandis et al
Measurement Technique 1986 and self reliance with competition distance from in groups
and concern for in groups Triandis et al 1988
Triandis and colleagues proposed multimethod assessment as
the most valid way to assess the cultural syndromes associated
Content of Current Scales
with IND and COL Triandis Chan Bhawuk Iwao Sinha
1995 Triandis McCusker Hui 1990 However perhaps be Given the diverse terminology used in IND COL research and
cause of the effort and difficulty involved in a multimethod as the different topics addressed in measurement instruments of IND
sessment Likert type ratings of values and attitudes are by far the COL we conducted a thorough review and content analysis of
most prevalent method used to assess IND COL Given their scales used in this area to clarify what IND and COL refer to We
dominance our review focuses on these types of scales In exam content coded each item on the 27 available IND COL scales
ining the scales cited in the past 20 years we did not find a single distinguishing COL focused and IND focused items with each
standard or most common measure though some items are com item assigned to only one content category for more information
mon across many scales In an effort to make sense of these see Coon Oyserman 2001 All three authors coded most scales
measures we sorted scales into messy categories finding 27 jointly using mutually agreed on guidelines with differences re
distinct scales Categories were messy and necessarily somewhat solved through discussion two authors coded remaining scales
arbitrary because researchers often modify scales with each use As shown in Table 1 we identified seven IND and eight COL
Distinct scales primarily contained items not previously published components accounting for 88 of items across each of the
Individualism and Collectivism Domains Assessed in Individualism Collectivism Scales
Domain name Description Sample item
Individualism
Independent Freedom self sufficiency and control I tend to do my own thing and others in
over one s life my family do the same
Goals Striving for one s own goals desires I take great pride in accomplishing what
and achievements no one else can accomplish
Compete Personal competition and winning It is important to me that I perform
better than others on a task
Unique Focus on one s unique idiosyncratic I am unique different from others in
qualities many respects
Private Thoughts and actions private from others I like my privacy
Self know Knowing oneself having a strong I know my weaknesses and strengths
Direct communicate Clearly articulating one s wants and I always state my opinions very clearly
Collectivism
Related Considering close others an integral part To understand who I am you must see
of the self me with members of my group
Belong Wanting to belong to and enjoy being To me pleasure is spending time with
part of groups others
Duty The duties and sacrifices being a group I would help within my means if a
member entails relative were in financial difficulty
Harmony Concern for group harmony and that I make an effort to avoid disagreements
groups get along with my group members
Advice Turning to close others for decision help Before making a decision I always
consult with others
Context Self changes according to context or How I behave depends on who I am
situation with where I am or both
Hierarchy Focus on hierarchy and status issues I have respect for the authority figures
with whom I interact
Group A preference for group work I would rather do a group paper or lab
than do one alone
10 OYSERMAN COON AND KEMMELMEIER
scales With regard to COL scales the overwhelming majority META ANALYSES OF THE LITERATURE
included at least one item focused on sense of duty to group COMPARING EUROPEAN AMERICANS WITH
85 of scales and relatedness to others 74 of scales OTHERS ON IND COL
Further more than half of the COL scales included at least one
item focused on seeking others advice 65 harmony In this section we present two sets of meta analyses focused on
57 and working in groups 57 The less commonly as difference in level of IND and COL including both international
sessed components of COL were sense of belonging to a group comparisons and within U S comparisons Note that the meta
39 contextual self 22 and valuing hierarchy 17 analyses do not address the implications of IND and COL for basic
With regard to IND scales 83 of scales included at least one item psychological processes we discuss implications later in the
focused on valuing personal independence One third or fewer of review
the scales included any of the remaining theoretically plausible six
components of IND personal achievement 33 self Overview
knowledge 33 uniqueness 30 privacy 22 clear Recall that a basic aim of the present review is to answer this
communication 19 and competition 15 Thus although question Are European Americans higher in IND lower in COL
scales are diverse there is enough overlap in scale content to than other cultural groups To begin to answer this question we
warrant use of meta analytic techniques conducted two groups of meta analyses The first group involved
Overall our content analysis of 27 IND COL scales allowed us a set of international comparisons contrasting Americans and
to identify consensual operationalization of IND and COL across Canadians with people from other countries The second group
researchers For IND the core or consensually agreed on element involved a set of within U S comparisons contrasting European
is valuation of personal independence For COL the core or Americans and Americans from other ethnic or racial groups We
consensually agreed on element is sense of obligation and duty to found too few studies using Canadians to warrant separate analy
the in group However scales vary widely in what other content ses For each set we conducted one analysis for IND scale scores
components they regard as relevant to the measurement of IND and a separate analysis for COL scale scores resulting in four sets
and COL For example some IND scales include personal unique of meta analyses These compared a Americans or Canadians
ness valuing privacy or self knowledge Family focus familial with people from other countries on measures of IND b Amer
ism is sometimes included in COL sometimes not and the same icans or Canadians with people from other countries on measures
is true of respect for hierarchy and competition which are some of COL c European Americans with other American ethnic and
times included as part of IND COL racial groups on measures of IND and d European Americans
Authors disagree as to whether familialism defined as relat with other American ethnic and racial groups on measures of COL
edness to family seeking harmony with family members or sup For international meta analyses we aggregated both by region
porting and seeking advice from family is separate from COL and by country For within U S meta analyses we aggregated
Gaines et al 1997 the essential core of COL Lay et al 1998 both across ethnic groups and by ethnic groups For both interna
or an important element of COL distinct from a non kin focused tional and within U S meta analyses we examined main and
type of COL Rhee et al 1996 As for hierarchy and competition moderator effects and analyzed the simultaneous effects of IND
Hofstede 1980 originally proposed that IND and power dis and COL by plotting IND COL effect sizes against each other
tance are separate cultural factors a view paralleled in Fiske s Main effects tell us about the size and direction of differences in
1992 taxonomy of basic social relationships and more recently IND and COL Moderator effects tell us to what extent scale
advocated by Triandis who proposed including hierarchical or reliability scale content and sample composition i e student vs
egalitarian aspects of social relationships in analyses of IND COL nonstudent influence size and direction of main effect differences
cf Singelis Triandis Bhawuk Gelfand 1995 Triandis Analyses of simultaneous effects of IND and COL can tell us
Gelfand 1998 By including a horizontal vertical dimension to whether European Americans are indeed both higher in IND and
discussion of cultural differences one can distinguish different lower in COL than are others
dimensions of IND and COL depending on whether they presume Eighty three studies contributed to the four types of compari
equal or different status between individuals namely horizontal sons described above Fifty of these studies provided data for the
IND horizontal COL vertical IND and vertical COL According international comparisons 35 provided data for the within U S
to this framework cultures high in horizontal IND tend to be comparisons Of these studies 2 provided data relevant to both sets
egalitarian with individuals being independent and of comparable of comparisons Matsumoto Weissman Preston Brown Kup
power and status whereas cultures high in vertical IND tend to perbusch 1997 Rhee et al 1996 resulting in the total 83
champion competition between individuals resulting in acceptable different studies Most studies reporting international comparisons
inequality between individuals omitted description of the ethnic or racial makeup of American
Given that there is no consensus in the literature on the roles of undergraduate samples We assumed these were predominantly
family and hierarchy within the IND COL framework we paid European American samples given the ethnic composition of un
special attention to findings concerning these issues Wherever dergraduate students at the universities represented the same was
possible we traced the extent that familialism functioned indepen true for studies using Canadian samples In our within U S sam
dently of COL In addition we attempted to assess whether hier ples we were limited to comparisons among European Americans
archy and personal competition functioned independently of IND and African Americans Asian Americans and Latino Americans
and COL Given the heterogeneity of measurement more gener because these were the only groups studied in the literature This
ally we attempted to examine whether differences between coun means for example that we found no literature comparing Euro
tries or groups depended on how IND and COL were assessed pean Americans with Native Americans on IND and COL
INDIVIDUALISM AND COLLECTIVISM 11
Overview of Computation and Analysis the statistical information available usually with the help of a
pertinent computer program B T Johnson 1993 With this
Computing Effect Sizes method there were only two cases in which the available infor
mation did not allow us to obtain the standard deviations of the
Main Effects means necessary for the computation of g In these cases we
estimated this information on the basis of studies with similar
The effect size calculated was g the mean difference between measures and samples from the same country Gire Carment
samples on IND or COL divided by the pooled standard deviation 1993 Leung Iwawaki 1988 Examination of studies with
In international comparisons g represents the mean difference complete data revealed heterogeneity of variance was rarely a
between an American sample and a sample from another country problem therefore when studies used factor analysis and reported
For within U S comparisons g represents the mean difference standardized factor scores without corresponding standard devia
between a European American sample and a sample from a spe tions Kashima et al 1995 Triandis et al 1993 we assumed a
cific other American group e g African Americans Because standard deviation equal to one and homogeneity of variance
some samples were small we applied the sample size correction across samples within studies to generate standard deviation esti
suggested by Hedges and Olkin 1985 Thus the final statistic mates We lacked statistical information in only one case a
reported is the corrected difference score d rather than g nonsignificant subscale effect Leung Iwawaki 1988 for
Most commonly researchers reported a single mean IND score which we assigned an effect size of zero Using these methods we
when assessing IND and a single mean COL score when assessing were able to compute effect sizes on the basis of available mean
COL We computed a single mean score across multiple subscales scores and standard deviations in all but 2 of studies in these
in the few studies that used subscales to assess IND or COL when latter cases we computed effect sizes on the basis of F statistics
the scale s original author recommended this approach e g Hui
1988 This allowed a straightforward computation of an effect Adjustment for Correlated Effect Sizes
size for two samples However in a number of cases we could not
follow this approach either because the original author of a scale As can be seen in Appendixes A and B and as detailed in
did not recommend combining subscales into a single mean score subsequent sections not all research involved simple cross cultural
or because the author used multiple measures of IND or COL comparisons of a sample of European Americans with a sample
In these latter cases we first computed effect sizes for each from a single other country or group Some researchers compared
subscale separately and then used a weighting procedure by Gleser the same sample of European Americans with a number of other
and Olkin 1994 to integrate them into a single effect size If the country or group samples For example the researcher could
information concerning the correlation between subscales required compare the same sample of European Americans with both a
by this procedure was not available we simply averaged effect sample of African Americans and a sample of Hispanic Americans
sizes for all subscales or measures of the same construct All or the same sample of Americans with both a sample from Japan
relevant subscales or measures were included in this procedure and a sample from Hong Kong Because in these cases the re
with only one exception Triandis s most recent IND COL scale searcher used the same sample of Americans to compute effect
Singelis et al 1995 Triandis Gelfand 1998 assesses both sizes for more than one comparison e g the U S Japanese com
horizontal and vertical IND and COL The two COL subscales parison and the U S Hong Kong comparison the resulting effect
were correlated hence they were combined into one COL score sizes are stochastically nonindependent The presence of stochastic
However Triandis designed the vertical and horizontal IND scales dependence violates the assumed independence of observations in
to be orthogonal raising questions as to whether we could treat conventional meta analysis methods e g Hedges Olkin 1985
combined group differences on the two scales as reflections of the Hunter Schmidt Jackson 1982 and results in misspecification
same underlying concept Further the vertical IND subscale fo of the standard error of the combined effect size estimates and a
cused exclusively on competition content that is atypical for all of distortion of Type I error Becker 2000 Gleser Olkin 1994 cf
the other IND scales we found Therefore we selected only the Raudenbush Becker Kalaian 1988
horizontal not the vertical IND subscale for the six international Becker 2000 suggested a number of different strategies to deal
comparison studies and the nine within United States studies that with nonindependent effect sizes The most preferable strategy to
used both IND subscales statistically model this nonindependence in the data requires that
We relied on the recommendations of Cohen 1988 in inter the researcher have a database that allows for reliable estimation of
preting the meaning of the observed effect sizes effect sizes d the degree of association between pairs of effect sizes That is to
of less than 0 4 are described as small those from 0 4 to 0 7 as be able to model nonindependence the researcher needs a database
moderate and those above 0 7 as large For ease of interpre that includes a sufficient number of studies that make exactly the
tation we set comparisons such that positive effect sizes reflect same multiple comparisons For example multiple studies com
higher European American IND and COL and negative effect paring a U S sample to both a sample from Japan and a sample
sizes reflect lower European American IND and COL The from Hong Kong would allow the reliable estimation of the U S
expected pattern is then positive values for IND effect sizes and Japanese effect size and the U S Hong Kong effect size We
negative ones for COL effect sizes found multiple studies with the same multiple comparisons to
Overall data quality was excellent with little information miss statistically model nonindependence in the within U S analyses
ing When published studies did not report necessary information but not in the international analyses
we contacted authors if the publication date was not more than 6 Specifically for within U S comparisons we used a generalized
years earlier Otherwise we estimated effect sizes on the basis of least squares estimation method that allows for statistically correct
12 OYSERMAN COON AND KEMMELMEIER
estimation of combined effect sizes and their variance by taking with 70 and as low reliability scales those scales with 70
their covariance with other effect sizes into account see Gleser e g Stangor 1998 We assumed high reliability for scales cre
Olkin 1994 for computational details This state of the art gen ated through factor analyses unless the study provided information
eralized least squares method of dealing with stochastically depen that this was not the case
dent effect sizes Becker 2000 provides combined effect size
coefficient estimates I and confidence intervals i z 2 var Scale Content
i 1 2 that are similar in interpretation to di and its confidence
interval An additional benefit of this method is that it allows for As described previously in the section on content analyses and
comparison of effect sizes This means that the interested reader summarized in Table 1 we coded each scale for content For each
can easily compare size of effects for within U S comparisons scale we coded for presence absence of seven components of IND
Because our international comparisons did not provide suffi and eight components of COL This allowed us to contrast effect
cient data to apply this state of the art method we used the next sizes from studies differing in scale content
best method forming subsets of country comparisons for all coun
tries for which more than one data point was available cf Becker Sample Composition
2000 In this way we synthesized effect sizes pertaining to each
country comparison e g U S Japan separately using the uni As noted in the section on limitations of the database almost all
variate approach to meta analysis proposed by Hedges and Olkin studies used students as research participants However some
1985 Because we could not use the least squares estimation studies used nonstudent adults Therefore we coded for presence
method there was no straightforward way to compare effect sizes of students versus nonstudent adults in the samples This allowed
for the international comparisons us to contrast effect sizes from studies differing in their sample
composition Comparisons were possible only for Japan versus
America on IND PR China versus America on COL and within
Computation of Homogeneity Coefficients U S IND and COL comparisons
As suggested by Hedges and Olkin 1985 we generated the
homogeneity coefficient Q for both international and within U S International Comparisons
analyses This coefficient shows whether the estimated effect size
can be used as a population estimate In our case it provided a Overview
sense of whether we could generalize from the obtained effect size In all we found 50 studies reported in 48 research reports or
of difference between European Americans and each group to the articles with cross cultural comparisons involving Americans
expected effect size in the population between European Ameri since 1980 3 Appendix A summarizes each study used in the
cans and each group When Q coefficients are nonsignificant international meta analyses Studies are organized alphabetically
samples from the same source show consistent differences with the by author and information is provided about year of publication
comparison group meaning that it is possible to generalize from which countries were compared with the United States the number
the samples to the population When Q coefficients are significant of participants in each sample the percentage of women in the
samples drawn from the same source e g region country or sample sample composition e g college students managers the
ethnic group differ systematically meaning it is not possible to IND COL measurement used effect size and reliability codes for
generalize to the population Moderator analyses are particularly moderator analyses
promising when Q is significant because they may reveal system Twenty seven studies assessed IND and COL 2 studies as
atic variability in the effect sizes explaining the within class sessed IND only and 21 studies assessed COL only Thus 29
heterogeneity For example when researchers studying the same studies provided IND data and 48 studies provided COL data
two countries compare them on scales that differ in reliability Participants were mostly undergraduates notable exceptions being
scale reliability may moderate or systematically influence the studies by Brett and Okumura 1998 C C Chen Meindl and Hui
extent that the countries appear different from each other the 1998 Earley 1989 1993 1994 Robert 1998 and Yang
effect size of the comparison 1996 who mostly used employees or managers as participants
Most studies were cross national comparing IND and or COL in
Moderators the United States and one other country Multinational studies
contributed more than one data point to the meta analyses These
We analyzed three possible moderators scale reliability scale studies compared Americans with 2 Earley 1993 1994 3
content and sample composition Chew 1996 Gudykunst Matsumoto Ting Toomey Nishida
Kim Heyman 1996 M Kim Hunter Miyahara Horvath
Scale Reliability 1996 Matsumoto et al 1997 Robert 1998 4 Kashima et al
1995 6 Kemmelmeier et al 2001 9 Triandis et al 1993
Whenever possible we obtained Cronbach s alpha for each or 38 other countries Oishi 2000 In this way the 29 IND studies
measure of IND and COL from each study used If a single study provided 90 data points and the 48 COL studies provided 109 data
presented more than one reliability coefficient we averaged them points
We did not attempt to impute reliability for studies for which we
could not obtain information because in our sample of studies
reliability was not consistent even across studies that used the 3
We could not include studies that lacked specific sample information
same measure We classified as high reliability scales those scales e g Schwartz 1994 Triandis et al 1986 1988
INDIVIDUALISM AND COLLECTIVISM 13
Figure 1 International meta analysis on individualism Comparing samples from America and eight regions
A large number of countries were included in these cross Country Level Analyses
national comparisons 46 countries for IND comparisons 50 coun
tries for COL comparisons However few countries provided more Regional analyses ignore existing differences between countries
than a single comparison and we obtained two or more IND of the same region and increase the risk of regional overgeneral
comparisons for 12 countries and two or more COL comparisons ization cf Dien 1999 Therefore we also performed a series of
for 14 countries Most of these more frequently studied countries country level analyses for a more appropriate examination of
were South Asian or East Asian 8 of 12 for IND and 7 of 14 for cross national differences This method has the advantage of in
COL creased specificity and allows us to explore the nature of within
region heterogeneity We computed effect sizes for all individual
country comparisons and present them graphically These displays
Analyses Plan provide the reader with a visual overview of the stability of IND
We first present analyses describing differences between Amer and COL differences across all countries
icans and others in IND and then analyses describing differences
between Americans and others in COL For each set of analyses Results of International Meta Analyses on IND
we use the following order a regional comparisons b country
level comparisons and c moderator analyses scale reliability Regional Analyses
scale content and sample composition
Figure 1 presents effect sizes comparing samples from America
with those from eight world regions In line with the assumption of
Regional Level Analyses an overarching Western culture we found no difference between
We grouped countries into eight regional blocks to contrast America and other English speaking countries and only relatively
regions of the world with Americans and Canadians Regional small differences between Europe Western and Central and
analysis allowed us to integrate many comparisons into a single America with Americans being higher in IND The majority of
effect size The regions we used were a English speaking coun available studies focused on comparisons between East Asians and
tries with an early British settlement White New Zealanders Americans making them the most common comparison in the
Australians South Africans b Western Europe France Italy
Spain c Central Europe Bulgaria Greece Poland d East 4
We note two exceptions a Israel was excluded from regional anal
Asia Japan PR China Vietnam e other Asian countries India yses it did not make sense to include it in the Middle Eastern region nor
Nepal Pakistan f Central Africa Nigeria Zimbabwe g Mid could we determine whether it best fit in another region e g Western or
dle East Egypt Turkey and h Latin South America Brazil Eastern Europe and b we included Puerto Rico as part of the Latin
Mexico Puerto Rico 4 American not the U S region
14 OYSERMAN COON AND KEMMELMEIER
INDIVIDUALISM AND COLLECTIVISM 15
cross national IND literature Combined effect sizes for compari Table 2
sons with East Asia were moderate in size as were combined International Meta Analyses Overall Individualism Results
effect sizes for Africa and the Middle East It is surprising that Compared With the United States and Canada
whereas these findings corroborate conventional expectations of
cultural theorists the combined effect size for other Asia was Mean weighted Homogeneity
effect size within
small and we found no difference in IND between the United Country n di 95 CI country Qwi
States and South America Recall however that student samples
dominated our data set Because students tend to be higher in Australia 3 0 02 0 09 0 13 6 32
socioeconomic status than nonstudents and higher socioeconomic Germany 3 0 01 0 12 0 14 51 29
Hong Kong 8 0 66 0 58 0 74 35 10
status is associated with higher IND and lower COL Freeman
India 3 0 29 0 17 0 42 40 64
1997 Hofstede 1980 Triandis et al 1990 it is possible that Indonesia 2 0 08 0 06 0 23 0 32
these results underestimate regional IND COL differences among Japan 15 0 25 0 18 0 31 549 01
nonstudents Korea 5 0 39 0 31 0 48 75 01
Nonetheless regional analyses generally support the assumption People s Republic
of China 4 0 46 0 36 0 55 57 07
that Americans are more individualistic than others are We turn to Poland 3 0 16 0 05 0 28 3 20
more fine grained analyses of country level effects as each re Puerto Rico 2 0 31 0 47 0 16 0 13
gional comparison yielded a highly significant Q coefficient all Singapore 2 0 24 0 11 0 38 1 62
values of p 004 meaning that regional effect estimates do not Taiwan 2 0 77 0 65 0 89 6 72
generalize to the population at large
Note Positive values indicate higher American individualism negative
values indicate lower American individualism Combined effect sizes that
Country Level Analyses differ reliably from 0 are bold CI confidence interval
p 05 p 01 p 001
Figure 2 shows how each country in each region compared with
America highlighting sources of within region heterogeneity As
seen earlier for the most part Americans were more individual Indonesians and that the U S Singapore and U S Poland differ
istic than were others Exceptions to this general pattern come ence in IND is small Moreover the small difference in IND
from single studies comparing America with a country located in between Puerto Rico and the United States suggests that Puerto
either an English speaking or a European region South Africa Ricans are more individualistic than European Americans Signif
New Zealand Finland Italy or Bulgaria Perhaps more surprising icant homogeneity coefficients in other country comparisons imply
is the observation that Americans were less individualistic than significant variation in effect size among studies with the same
were Latin Americans in more than half of the country samples country comparison limiting generalizability We therefore turn to
Unfortunately we could not assess the stability of these differ moderator analyses to explore the impact of differences in scales
ences because except for comparisons with Puerto Rico these and other factors influencing effect sizes
exceptions occurred for countries represented by only one sample Scale reliability moderator analyses Our first set of modera
in our database tor analyses focused on the possibility that effect sizes vary sys
The meta analysis allowed examination of the stability of dif tematically because of differences in scale reliability Psychomet
ferences between European Americans and others in a subset of ric theory leads to the assumption that if validity is held constant
comparisons based on more than one sample this subset of mul more reliable scales will show uniformly higher effect sizes than
tisample countries included Puerto Rico and some European and less reliable scales e g Nunnally 1994 To examine this possi
Asian countries As seen in Table 2 European Americans were bility we focused on country comparisons in which we had at least
significantly more individualistic than Hong Kong Indian Japa two studies with high reliability and at least two studies with low
nese Korean Polish Singaporean and Taiwanese respondents reliability the minimum number required for this type of analysis
However they also reported significantly lower IND than Puerto Therefore country comparisons that provided fewer than four data
Rican respondents and were not reliably different in IND from points overall and country comparisons with fewer than two data
Australian German or Indonesian respondents Furthermore al points in the high or low reliability categories specifically could
though Americans were more individualistic than were East not be included in the moderator analyses
Asians effects for comparisons with India Japan and Singapore These statistical requirements limited our moderator analyses to
were small Given nonsignificant Q scores in Indonesia Poland three countries Hong Kong Japan and Korea For these countries
Puerto Rico and Singapore current evidence is sufficient to con results did not follow the simple rule of greater effect size being
clude that there are no differences in IND between Americans and associated with higher reliability As can be seen in the first
Figure 2 opposite International meta analysis on individualism Comparing samples from America and 47
countries Country comparisons are grouped by region with countries appearing alphabetically within region
Light bars show effect sizes based on a single data point a single sample comparison and dark bars show effect
sizes based on multiple data points These latter bars represent the countries for which a statistical integration
of the available effect sizes is provided and where possible moderator analyses were performed PR China
People s Republic of China
16 OYSERMAN COON AND KEMMELMEIER
International Meta Analyses Individualism Scale Reliability Moderator Analyses
Between class Mean weighted Homogeneity within
Country effect QB n effect size di 95 CI class Qwi
Hong Kong 0 27
Low reliability 2 0 65 0 51 0 79 12 60
High reliability 4 0 69 0 59 0 80 16 88
Japan 39 01
Low reliability 9 0 08 0 001 0 16 371 34
High reliability 6 0 48 0 38 0 57 138 66
Korea 14 65
Low reliability 2 0 55 0 43 0 66 3 18
High reliability 3 0 22 0 10 0 34 57 18
Note Positive values indicate higher American individualism negative values indicate lower American
individualism Studies with Cronbach s 70 were classified as low reliability studies with Cronbach s
70 were classified as high reliability Combined effect sizes that differ reliably from 0 are bold CI confidence
p 10 marginally significant p 05 p 001
column of Table 3 the between class effect was not significant for valuing independence Additional components assessed in either
Hong Kong was significant in the expected direction for Japan Japan or Hong Kong were valuing personal uniqueness privacy
and was significant in the opposite direction for Korea For Hong direct communication and self knowledge see Table 1 for a
Kong effect sizes were not dependent on scale reliability For description of each domain
Japan reliable scales were associated with moderate effect sizes For America Hong Kong comparisons scale content did not
and less reliable scales were associated with nonsignificant differ affect the effect size Americans were higher in IND no matter
ences in IND between Japan and America Finally for Korea whether assessment of IND included uniqueness privacy or direct
effect sizes were smaller for scales that are more reliable and larger communication suggesting that the difference between these two
for less reliable scales Given these diverse findings it is unlikely countries in IND does not reside in any of these but rather in
that reliability alone is responsible for variation in effect sizes At valuing personal independence For America Japan comparisons
least in the Korean case our findings point to a problem of scale content was critical for both the size and the direction of
validity perhaps related to scale content as explored below effects Specifically Americans were higher in IND than were
Scale content moderator analyses The next set of moderator Japanese when in addition to personal independence IND assess
analyses examined whether observed effect sizes differed as a ment included personal uniqueness valuing privacy and direct
function of scale content We examined the moderating role of communication In fact when IND assessment did not include
scale content for all country comparisons for which there were at personal uniqueness Americans were lower in IND than were
least two studies that included specific content in their assessment Japanese Further the effect size jumped from small to large when
of IND and at least two studies that did not include the same valuing personal privacy was included with a similar but less
content in their assessment of IND As with the scale reliability dramatic increase when direct communication was included
analyses this criterion excluded country comparisons with fewer These data are suggestive of the notion that the between U S
than two data points in a given content category Thus as detailed Japan difference in IND resides in a combination of different
below although there were five studies with Korea comparisons valuation of independence personal uniqueness personal privacy
we were unable to complete a moderator analysis of scale content and direct communication
First all five Korea comparisons assessed IND with two of the In contrast when competition was included in the scale the
same IND components valuing personal independence and difference between Americans and Japanese in IND disappeared
uniqueness hence we could not compare studies that assessed suggesting that competitiveness is a construct unrelated to IND
IND with and without these components Second each of the other This conclusion finds support in Triandis s 1995 Singelis et al
components of IND appeared only in a single study Because 1995 advocacy of assessing competition as a cultural factor sep
comparisons require at least two studies with a given content arate from IND Further the self knowledge component did not
component and two studies without the same content component have any influence on effect size suggesting it is not relevant to
we could not make these other comparisons Similarly for one of the assessment of IND with respect to differences between Japan
the four studies involving PR China we were unable to obtain any and America These findings are interesting and for self
information about the content of the measure used leaving only knowledge provocative because they seem to contradict basic
three studies too few to compute moderator analyses assumptions about the influence of IND on American and Japanese
As reflected in Table 4 sufficient observations were available self concepts However caution in generalizing is necessary be
only for Hong Kong and Japan In these studies all IND scales cause as can be seen in the last column of Table 6 the combined
included a common component valuing personal independence effect sizes are quite heterogeneous Moreover some of the effects
making it a constant in this analysis Therefore analyses of the may be due to the specific scale used rather than the content
effect of other components necessarily included the effect of domains more generally privacy and self knowledge each ap
INDIVIDUALISM AND COLLECTIVISM 17
International Meta Analyses Individualism Scale Content Moderator Analyses
Between class Mean weighted Homogeneity within
Country effect QB n effect size di 95 CI class Qwi
Hong Kong independent
Unique 0 07
Included 6 0 66 0 57 0 75 28 56
Not included 2 0 68 0 51 0 86 6 46
Private 2 04
Included 2 0 76 0 60 0 92 8 97
Not included 6 0 63 0 53 0 72 24 09
Direct communicate 0 48
Included 5 0 65 0 55 0 74 27 23
Not included 3 0 71 0 55 0 87 7 39
Japan independent
Competition 5 37
Included 2 0 01 0 24 0 22 0 39
Not included 13 0 27 0 20 0 33 543 25
Unique 216 80
Included 11 0 49 0 42 0 55 140 80
Not included 4 0 66 0 79 0 52 191 42
Private 70 28
Included 2 0 82 0 67 0 96 1 44
Not included 13 0 12 0 06 0 19 477 30
Direct communicate 16 45
Included 6 0 39 0 30 0 48 69 07
Not included 9 0 13 0 05 0 22 463 49
Self know 0 32
Included 2 0 29 0 12 0 45 22 85
Not included 13 0 24 0 17 0 31 525 84
Note Positive values indicate higher American individualism negative values indicate lower American
individualism In parentheses after the country name is the content contained in all or all but one research
instrument used in studies contributing to the country comparison Studies for which no reliability information
was available were excluded Combined effect sizes that differ reliably from 0 are bold CI confidence
p 05 p 01 p 001
peared in only one scale Gudykunst et al 1996 for privacy and Results of International Meta Analyses on COL
Singelis et al 1995 for self knowledge which means that idio
syncrasies of these scales cannot be disentangled from the effect of Regional Analyses
scale content Overall Americans were higher in IND than East As displayed in Figure 3 Americans were lower in COL than
Asians were However effects varied from sample to sample and were others from all regions of the world with the exception of
scale reliability or content did not fully account for this variability English speaking countries We found it surprising that Americans
Sample composition moderator analyses Following the crite were lower in COL than Europeans were befitting the idea of a
ria of needing at least four studies in a country comparison for a uniquely American way of being high individualism and low
moderator analysis in this case two using students and two using collectivism but challenging the notion of a single Western
nonstudent adults in their samples we were able to analyze the culture In addition effect sizes for Asian regions were similar to
possibility of moderation due to sample composition for the those for European regions and we found large effects only for
Japan America comparison only Two of these 15 studies sampled comparisons with Africa Almost half of all studies focused on
nonstudent adults We found that with nonstudent adults the effect comparisons between East Asian regions and America However
size was somewhat larger d 0 44 confidence interval CI 0 20 there were considerable within region differences as revealed by
to 0 68 than with college students d 0 23 CI 0 17 to 0 30 the significant homogeneity statistic Q for each regional compar
suggesting that analyses of cultural differences between samples of ison all ps 00001 We turned to more fine grained cross
college students may provide a conservative estimate of the nature national analyses to unpack this regional heterogeneity
of the difference between Japanese and Americans on IND How
ever this shift was significant only at trend level between class
Country Level Analyses
effect QB 2 80 p 10 and effect size estimates for both
nonstudent adults and college students were quite heterogeneous As shown in Figure 4 findings across the 50 country compar
Qw 5 18 p 03 and Qw 514 03 p 001 respectively isons confirmed the general conclusion that relative to others
Thus even within samples student and nonstudent effect sizes Americans were lower in COL There were notable exceptions
were heterogeneous and it is not possible to generalize findings though showing higher American COL in comparison to New
either to student or to nonstudent populations Zealand France Singapore Tanzania Egypt Costa Rica and


Related Books

Rethinking Individualism and Collectivism: Evaluation of ...

Rethinking Individualism and Collectivism: Evaluation of ...

Rethinking Individualism and Collectivism: Evaluation of Theoretical Assumptions and Meta-Analyses Daphna Oyserman, Heather M. Coon, and Markus Kemmelmeier University of Michigan Are Americans more individualistic and less collectivistic than members of other groups? The authors

Continue Reading...
European Journal of Social Theory Rethinking ...

European Journal of Social Theory Rethinking ...

Rethinking individualization: ... individualism Liza Cortois University of Leuven, Belgium Rudi Laermans University of Leuven, Belgium Abstract This article proposes a more culturalist and variegated conception of the individual than that presented by individualization theorists. Inspired by the approach of the individual advocated by E´mile Durkheim, Talcott Parsons and John Meyers, it first ...

Continue Reading...
Explanatory strategies beyond the individualism/holism debate.

Explanatory strategies beyond the individualism/holism debate.

1 Explanatory strategies beyond the individualism/holism debate. Jeroen Van Bouwel Ghent University, Belgium [email protected] (This is a draft of a paper to be published in: Collin Finn ...

Continue Reading...
Rethinking Palestine: settler-colonialism, neo-liberalism ...

Rethinking Palestine: settler-colonialism, neo-liberalism ...

Rethinking Palestine: settler-colonialism, neo-liberalism and individualism in the West Bank and Gaza Strip Jamil Hilal* Independent sociologist, occupied West Bank The 1967 occupied Palestinian territories have undergone three major types of development since the Oslo agreement between the Palestinian Liberation

Continue Reading...
Simmons, B. & Watson, D. (2015) From Individualism to Co ...

Simmons, B. & Watson, D. (2015) From Individualism to Co ...

Simmons, B. & Watson, D. (2015) From Individualism to Co-construction and Back Again: Rethinking Research Methodology for Children with Profound and Multiple Learning Disabilities, Child Care in ...

Continue Reading...
Culture Posts: Individualism, Collectivism – and Relationalism

Culture Posts: Individualism, Collectivism – and Relationalism

Culture Posts: Individualism, Collectivism – and Relationalism [1] In honor of the New Year (both West and East), I would like to share a relatively new lens for viewing relations in public diplomacy. Many may have heard of the terms individualism, which privileges the individual, and collectivism, which favors the collective or group. What ...

Continue Reading...
Rethinking Micro-Macro Relations - ResearchGate

Rethinking Micro-Macro Relations - ResearchGate

Rethinking Micro-Macro Relations Petri Ylikoski Abstract: This paper proposes a new approach to the micro-macro problem in the social sciences.

Continue Reading...
UKRAINIANCULTURE: INDIVIDUALISMORCOLLECTIVISM? BorysenkoL.H.

UKRAINIANCULTURE: INDIVIDUALISMORCOLLECTIVISM? BorysenkoL.H.

individualism scores obtained therein were close to the average values, implying that Ukrainian individualism has, at best, a slight prevalence over collectivism [15-17].

Continue Reading...
Rethinking accountability in a knowledge society

Rethinking accountability in a knowledge society

Rethinking accountability in a knowledge society Pasi Sahlberg Published online: 20 November 2008 Springer Science+Business Media B.V. 2008 Abstract Competition between schools combined with test-based accountability to hold

Continue Reading...
Social Capital and Equality: Tocqueville’s Legacy ...

Social Capital and Equality: Tocqueville’s Legacy ...

Social Capital and Equality: Tocqueville’s Legacy 77 The first appearance of the concept, with the exception of Hanifan6 (1916), is at the beginning of 1960’s in a famous book by Jane Jacobs (1961) called The Death and Life of Great American Cities. Social capital is used as a device to criticize the artificial development

Continue Reading...
RETHINKING THE ROLE OF KIERKEGAARD’S ‘AUTHENTIC INDIVIDUAL ...

RETHINKING THE ROLE OF KIERKEGAARD’S ‘AUTHENTIC INDIVIDUAL ...

Rethinking the role of Kierkegaard’s ‘authentic individual’ 131 In my article, I am exploring this potential on the basis of Kierkegaard‘s later work The Sickness unto Death (1849) [12],arguing that his relational understanding of a human (authentic) self can serve as an antidote to the acute sense of loneliness and alienation of postmodern humans, over-burdened with

Continue Reading...