ArticlePDF Available

Abstract and Figures

With data from 33 nations, we illustrate the differences between cultures that are tight (have many strong norms and a low tolerance of deviant behavior) versus loose (have weak social norms and a high tolerance of deviant behavior). Tightness-looseness is part of a complex, loosely integrated multilevel system that comprises distal ecological and historical threats (e.g., high population density, resource scarcity, a history of territorial conflict, and disease and environmental threats), broad versus narrow socialization in societal institutions (e.g., autocracy, media regulations), the strength of everyday recurring situations, and micro-level psychological affordances (e.g., prevention self-guides, high regulatory strength, need for structure). This research advances knowledge that can foster cross-cultural understanding in a world of increasing global interdependence and has implications for modeling cultural change.
Content may be subject to copyright.
DOI: 10.1126/science.1197754
, 1100 (2011);332 Science , et al.Michele J. Gelfand
Differences Between Tight and Loose Cultures: A 33-Nation Study
This copy is for your personal, non-commercial use only.
clicking here.colleagues, clients, or customers by , you can order high-quality copies for yourIf you wish to distribute this article to others
here.following the guidelines can be obtained byPermission to republish or repurpose articles or portions of articles
): May 26, 2011 (this infomation is current as of
The following resources related to this article are available online at
version of this article at: including high-resolution figures, can be found in the onlineUpdated information and services,
can be found at: Supporting Online Material
, 1 of which can be accessed free:cites 14 articlesThis article
1 articles hosted by HighWire Press; see:cited by This article has been
Psychology subject collections:This article appears in the following
registered trademark of AAAS. is aScience2011 by the American Association for the Advancement of Science; all rights reserved. The title CopyrightAmerican Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005.
(print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by theScience
on May 27, 2011www.sciencemag.orgDownloaded from
prot ected sugar beet seedlings from infection by
R. solani (fig. S7). Random transposon mutagen-
esis generated two mutants of strain SH-C52
with no in vitro activity against R. solani.The
single transposon insertions were mapped to a
nonribosomal peptide synthetase (NRPS) gene
with 69% sequence identity to syrE, the gene
of the syringomycin-syringopeptin (syr-syp)bio-
synthetic pathway in Pseudomonas syringae
pv. syringae (9). NRPS-mutant O33 colonized
the rhizosphere to the same extent as its parental
strain SH-C52, but did not protect sugar beet
seedlings from fungal infection (fig. S7). Subse-
quent genetic analyses revealed that the putative
biosynthetic pathway consisted of two gene clus-
ters, designated thaAB and thaC1C2D,whichwere
predicted to encode a nineamino acid chlorinated
lipopeptide (fig. S8).
The multifaceted approach adopted in this
study, linking culture-independent and culture-
dependent analyses, shows that plants, like mam-
mals and insects (1012), can rely on specific
constituents of the microbial community for pro-
tection against pathogen infections. We showed
that the g-Proteobacteria, and specifically mem-
bers of the Pseudomonadaceae, protect plants
from fungal infection through the production of
a putative chlorinated lipopeptide encoded by
NRPS genes. Functional analysis further revealed
a significant difference in plant disease suppres-
sion between haplotypes SH-A and SH-C (fig. S7),
suggesting that in situ antifungal activity is
governed by individual members of this bac-
terial taxon. Next to the Pseudomonadaceae,
several other bacterial taxa were found in this
study to be associated with disease suppressive-
ness (Fig. 3). Some of these taxa, including the
Burkholderiaceae, Xanthomonadales, and Actino-
bacteria, harbor genera and species with activ-
ity against plant pathogenic fungi, including
R. solani (13). These findings suggest that the
complex phenomenon of disease suppressive-
ness of soils cannot simply be ascribed to a single
bacterial taxon or group, but is most likely gov-
erned by microbial consortia. The observation
that bacterial strains, which lack activity against
pathogens when tested alone, can act synergis-
tically when part of microbial consortia (14 )fur-
ther exemplifies the complexity of adopting
Kochs postulates for identification of micro-
organisms involved in disease suppressiveness
of soils. The bacteria and biosynthetic pathway
identified here provide a set of microbial and
genetic markers to elucidate whether and how
plants recruit beneficial soil microorganisms for
protection against infections.
References and Notes
1. H. Marschner, Mineral Nutrition of Higher Plants
(Academic Press, London, ed. 2, 1995).
2. T. Bisseling, J. L. Dangl, P. Schulze-Lefert, Science 324,
691 (2009).
3. R. J. Cook et al., Proc. Natl. Acad. Sci. U.S.A. 92, 4197 (1995).
4. D. Haas, G. Défago, Nat. Rev. Microbiol. 3, 307 (2005).
5. D. M. Weller, J. M. Raaijmakers, B. B. M. Gardener,
L. S. Thomashow, Annu.Rev.Phytopathol.40, 309 (2002).
6. T. C. Hazen et al., Science 330, 204 (2010).
7. K. M. DeAngelis et al., ISME J. 3, 168 (2009).
8. P. D. Schloss, J. Handelsman, PLOS Comput. Biol. 2,
e92 (2006).
9. H. Feilet al., Proc. Natl. Acad. Sci. U.S.A. 102, 11064 (20 05).
10. R. E. Ley et al., Science 320, 1647 (2008).
11. J. Qin et al., MetaHIT Consortium, Nature 464, 59 (2010).
12. J. J. Scott et al., Science 322, 63 (2008).
13. J. Postma, R. W. A. Scheper, M. T. Schilder, Soil Biol.
Biochem. 42, 804 (2010).
14. P. Garbeva, M. W. Silby, J. M. Raaijmakers, S. B. Levy,
W. D. Boer, ISME J. (2011).
Acknowledgments: We thank T. Bisseling for critical
reading and valuable suggestions. We acknowledge
assistance by L. Sibbel-Wagemakers, N. Pangesti,
M. de Milliano, N. Sharma, R. de Vries, P.M.S. van Oorschot,
A. H. L. Schoone, and Y. Bakker. This work was financially
supported by grants from Netherlands Scie nce Organisation
(NWO)ERGO (#838.06.101) and Netherlands Genomics
InitiativeEcogenomics, Netherlands. Additional work was
performed at Lawrence Berkeley National Laboratory (LBNL)
(contract DE-AC02-05CH11231 with the U.S. Department of
Energy). The 16SrDNA sequences are available on
GenBank under accessions HQ848634 to HQ848643,
and the thaABCD sequences under accession HQ888764.
LBNL has a patent on the PhyloChip assay and Second
Genome has licensed this assay from LBNL. Although the
G3 PhyloChip is under patent (and under exclusive
license to Second Genome), the data generated from
the use of the chip are not patented or restricted.
T.Z.dS. owns stock in Second Genome valued at
under $10,000.
Supporting Online Material
Materials and Methods
Figs. S1 to S8
Tables S1 to S5
8 February 2011; accepted 20 April 2011
Published online 5 May 2011;
Differences Between Tight and Loose
Cultures: A 33-Nation Study
Michele J. Gelfand,
*Jana L. Raver,
Lisa Nishii,
Lisa M. Leslie,
Janetta Lun,
Beng Chong Lim,
Lili Duan,
Assaf Almaliach,
Soon Ang,
Jakobina Arnadottir,
Zeynep Aycan,
Klaus Boehnke,
Pawel Boski,
Rosa Cabecinhas,
Darius Chan,
Jagdeep Chhokar,
Alessia DAmato,
Montse Ferrer,
Iris C. Fischlmayr,
Ronald Fischer,
Mar t a Fülöp ,
James Georgas,
Emiko S. Kashima,
Yoshishima Kashima,
Kibum Kim,
Alain Lempereur,
Patricia Marquez,
Rozhan Othman,
Bert Overlaet,
Penny Panagiotopoulou,
Karl Peltzer,
Lorena R. Perez-Florizno,
Larisa Ponomarenko,
Anu Realo,
Vidar Schei,
Manfred Schmitt,
Peter B. Smith,
Nazar Soomro,
Erna Szabo,
Nalinee Taveesin,
Midori Toyama,
Evert Van de Vliert,
Naharika Vohra,
Colleen Ward,
Susumu Yamaguchi
With data from 33 nations, we illustrate the differences between cultures that are tight
(have many strong norms and a low tolerance of deviant behavior) versus loose (have weak
social norms and a high tolerance of deviant behavior). Tightness-looseness is part of a complex,
loosely integrated multilevel system that comprises distal ecological and historical threats
(e.g., high population density, resource scarcity, a history of territorial conflict, and disease and
environmental threats), broad versus narrow socialization in societal institutions (e.g., autocracy,
media regulations), the strength of everyday recurring situations, and micro-level psychological
affordances (e.g., prevention self-guides, high regulatory strength, need for structure). This
research advances knowledge that can foster cross-cultural understanding in a world of increasing
global interdependence and has implications for modeling cultural change.
How othercultures differ from ones
own has piqued the curiosity of scholars
and laypeople across the centuries. As
long ago as 400 B.C.E., Herodotus documented
a wide variety of cultural practices that he ob-
served in his travels in The Histories (1). Only
in the past few decades have scientists begun
to move beyond descriptive accounts of cultural
differences to empirically assess ways in which
national cultures vary. We examine a neglected
source of cultural variation that is dominating
the geo-political landscape and has the potential
to be a major source of cultural conflict: the differ-
ence between nations that are tight”—have strong
norms and a low tolerance of deviant behavior
and those that are loosehave weak norms and
a high tolerance of deviant behavior.
Early anthropological research showed the
promise of this distinction. In his study of 21 tra-
ditional societies, Pelto (2) documented wide var-
iation in the expression of and adherence to social
norms. The Hutterites, Hanno, and Lubara were
among the tightest societies, with very strong
norms and severe sanctions for norm violation,
whereas the Kung Bushman, Cubeo, and the Skolt
Lapps were among the loosest societies, with am-
biguous norms and greater permissiveness for norm
violation. Pelto speculated that these societies may
have different ecologies, with tight societies having
a higher population per square mile and a higher
dependence on crops as compared to loose socie-
ties. Later research indeed showed that agricultural
societies (e.g., the Temne of Sierra Leone), which
require strong norms to foster the coordination
necessary to grow crops for survival, had strict
child-rearing practices and children who were high
27 MAY 2011 VOL 332 SCIENCE
on May 27, 2011www.sciencemag.orgDownloaded from
on conformity. Hunting and fishing societies (e.g.,
the Inuit) had lenient child-rearing practices and
children who were low on conformity (3,4).
Despite evidence of the importance of this
contrast in traditional societies, there exists no
insight into how tightness-looseness operates in
modern nations. The goal of this research is to
fill this void. Drawing on theorizing in cultural
psychology (5,6), we propose that tightness-
looseness is part of a complex, loosely integrated
system that involves processes across multiple
levels of analysis (Fig. 1). We theorize that the
strength of social norms and tolerance of deviant
behaviorthe core distinction between tight and
loose culturesis afforded by numerous distal
ecological and human-made societal threats and
societal institutions and practices. The strength
of social norms and tolerance of deviant behav-
ior is further reflected and promoted in the pre-
dominance of strong versus weak situations that
are recurrent in everyday local worlds, and is re-
inforced through psychological processes that are
attuned to situational requirements. We provide an
empirical test that shows how ecological, histor-
ical, and institutional factors, along with everyday
situations and psychological processes, together
constitute cultural systems.
We predict that tightness-looseness is afforded
by a broad array of ecological and human-made
societal threats (or lack thereof) that nations have
historically encountered (4,7). Ecological and
human-made threats increase the need for strong
norms and punishment of deviant behavior in
the service of social coordination for survival
whether it is to reduce chaos in nations that have
high population density, deal with resource scar-
city, coordinate in the face of natural disasters,
defend against territorial threats, or contain the
spread of disease. Nations facing these particular
challenges are predicted to develop strong norms
and have low tolerance of deviant behavior to
enhance order and social coordination to effec-
tively deal with such threats. Nations with few
ecological and human-made threats, by contrast,
have a much lower need for order and social
coordination, affording weaker social norms and
much more latitude (8).
The strength of social norms and tolerance of
deviant behavior is also afforded by and reflected
in prevailing institutions and practices. Institu-
tions in tight nations have narrow socialization
that restricts the range of permissible behavior,
whereas institutions in loose nations encourage
broad socialization that affords a wide range of
permissible behavior (9). Relative to loose na-
tions, tight nations are more likely to have auto-
cratic governing systems that suppress dissent, to
have media institutions (broadcast, paper, Inter-
net) with restricted content and more laws and
controls, and to have criminal justice systems
with higher monitoring, more severe punishment
(e.g., the death penalty), and greater deterrence
and control of crime. Tight nations will also be
more religious, thereby reinforcing adherence to
moral conventions and rules that can facilitate
social order and coordination (10). Challenges to
societal institutions (e.g., demonstrations, boy-
cotts, strikes) will be much less common in tight
nations than in loose ones. These institutions and
practices simultaneously reflect and support the
strength of norms and tolerance of deviance that
exists in nations.
Tightness-looseness is manifested not only in
distal ecological, historical, and institutional con-
texts but also in everyday situations in local
worlds (e.g., at home, in restaurants, classrooms,
public parks, libraries, the workplace) that indi-
viduals inhabit (5,6). We theorize that tightness-
looseness is reflected in the predominance of
strong versus weak everyday situations (11 ,12 ).
Strong situations have a more restricted range of
appropriate behavior, have high censuring poten-
tial, and leave little room for individual discre-
tion. Weak situations place few external constraints
on individuals, afford a wide range of behavioral
options, and leave much room for individual dis-
cretion. Situational strength has been long dis-
cussed among psychologists, sociologists, and
anthropologists (1114) but has yet to be linked
to cultural variation. Tight nations are expected
to have a much higher degree of situational con-
straint which restricts the range of behavior deemed
appropriate across everyday situations (e.g., class-
rooms, libraries, public parks, etc.). By contrast,
loose nations are expected to have a much weaker
situational structure, affording a much wider range
of permissible behavior across everyday situa-
tions. The strength (or weakness) of everyday re-
curring situations within nations simultaneously
reflects and supports the degree of order and so-
cial coordination in the larger cultural context.
We further theorize that there is a close con-
nection between the strength (versus weakness)
of everyday situations and the chronic psycho-
logical processes of individuals within nations.
In this view, individualspsychological processes
become naturally attuned to, and supportive of,
the situational demands in the cultural system
(15). Individuals who are chronically exposed to
stronger (versus weaker) situations in their every-
day local worlds have the continued subjective
experience that their behavioral options are lim-
ited, their actions are subject to evaluation, and
there are potential punishments based on these
evaluations. Accordingly, individuals in nations
with high situational constraint will have self-
guides that are more prevention-focused (16)and
thus will be more cautious (concerned with avoid-
ing mistakes) and dutiful (focused on behaving
properly), and will have higher self-regulatory
strength (higher impulse control) (17 ), a higher
need for structure (18), and higher self-monitoring
ability (19,20). Put simply, the higher (or lower)
degree of social regulation that exists at the
societal level is mirrored in the higher (or lower)
amount of self-regulation at the individual level
in tight and loose nations, respectively. Such
psychological processes simultaneously reflect
and support the strength of social norms and tol-
erance of deviance in the larger cultural context.
To provide a systematic analysis of tightness-
looseness in modern societies, we gathered data
Department of Psychology, University of Maryland, College
Queens School of Business, Kingston,
Ontario K7L 3N6, Canada.
Cornell University Industrial Labor
Relations School, Ithaca, NY 14853, USA.
Carlson School of
Management, University of Minnesota, Minneapolis, MN 55455,
Ministry of Defense, Singapore and Nanyang Business
School, Defense Technology Towers, 5 Depot Road, #16-01
Tower B, Singapore.
McKinsey & Company, Washington, DC
20036, USA.
BIP Institute of Psychology Ltd.,Vita Towers,11
Ben-Gurion Street, Bney-Brak 51260, Israel.
Nanyang Tech-
nological University, Nanyang Avenue 639798, Singapore.
Kaplaskjolsvegur 29, IS-107 Reykjavik, Iceland.
ment of Psychology, Koc University, Sariyer 34450, Turkey.
Bremen International Graduate School of Social Sciences,
Jacobs University Bremen, D-28759 Bremen, Germany.
saw School of Social Sciences and Humanities, 03-815 Warsaw,
University of Minho, Campus de Gualtar, 4710-057
Braga, Portugal.
Department of Psychology, Chinese Uni-
versity of Hong Kong, 3rd Floor, Sino Building, Shatin, N.T.,
Hong Kong.
Indian Institute of Management, Ahmedabad-
380015, India.
CENTRUM Catolica, Pontificia Universidad
Catolica del Peru, Lima 33, Peru.
Social Psychology Depart-
ment, University of Valencia, Avenida Blasco Ibáñez, 21, 46010
Valencia, Spain.
Johannes Kepler University, Institute for
International Management, Altenbergerstrasse 69, 4040
Linz, Austria.
School of Psychology, Centre for Applied
Cross-Cultural Psychology, Victoria University of Wellington,
PO Box 600, Wellington, New Zealand.
Institute for Psy-
chology, Hungarian Academy of Sciences, Victor Hugo Street
18-22, Budapest 1132, Hungary.
University of Athens, 11
Herodou Attikou, Athens 106 74, Greece.
School of Psy-
chological Science, La Trobe University, Bundoora, Victoria
3086, Australia.
Psychological Sciences University of Mel-
bourne, Victoria3010, Australia.
School of Social Sciences,
Sungkyunkwan University, Myungryun-dong 3 ga, Jongno-gu,
Seoul 110-745, Korea.
ESSEC Business School, Av. Bernard
Hirsch, B.P. 50105, 95021 Cergy Pontoise Cedex, France.
University of San Diego, 235 Olin Hall, 5998 Alcala Park, San
Diego, CA 92131, USA.
21 Jalan 5C/6, 43650 B B Bangi,
Selangor, Malaysia.
KU Leuven, Naamsestraat 67, B-3000
Leuven, Belgium.
University of Patras, 26500 Rio, Patras,
Human Sciences Research Council, Private Bag X41,
Pretoria 0001, South Africa.
Investigadora Colegio de la
Fontera Norte, Km 18.5 carretera escénica Tijuana - Ensenada,
San Antonio del Mar, Tijuana, Baja California C.P. 22560,
Psychology Department, Odessa National Uni-
versity, Dvorianskaya str. 2, Odessa 67027, Ukraine.
partment of Psychology, University of Tartu, Tiigi 78, Tartu
50410, Estonia.
Department of Strategy and Manage-
ment, Norwegian School of Economics and Business Admin-
istration, Breiviksveien 40, 5045 Bergen, Norway.
Koblenz-Landau, Fortstraße 7, D-76829 Landau, Germany.
Department of Psychology, University of Sussex, Falmer,
Brighton BN1 9QH, UK.
Department of Psychological Test-
ing, Guidance and Research, University of Sindh, Elsa Kazi
Campus, Hyderabad 71000, Pakistan.
33 Soonvijai 4,
Bangkok 10310, Thailand.
Department of Psychology
Gakushuin University, 1-5-1 Mejiro, Toshima-ku, Tokyo 171-
8588, Japan.
University of Groningen, Grote Kruisstraat 2/1,
9712 TS Groningen, Netherlands.
Indian Institute of Manage-
ment, Wing 12, IIM Campus, Vastrapur, Ahmedabad, Gujarat
380015, India.
School of Psychology, P.O. Box 600, Victoria
University of Wellington, Wellington 6140, New Zealand.
Department of Social Psychology, Graduate School of Hu-
manities and Sociology, University of Tokyo, Hongo 7-3-1,
Bunkyo-ku, Tokyo 113-0033, Japan.
*To whom correspondence should be addressed. E-mail: SCIENCE VOL 332 27 MAY 2011 1101
on May 27, 2011www.sciencemag.orgDownloaded from
from 6823 respondents across 33 nations (20).
Sample characteristics are shown in Table 1 (21).
In each nation, we surveyed individuals from a
wide range of occupations as well as university
students. Data on ecological and historical threats
and societal institutions were collected from nu-
merous established databases (20). When possi-
ble, historical data were included (e.g., population
density in 1500, history of conflict 19182001,
historical prevalence of pathogens).
Tightness-looseness (the overall strength of
social norms and tolerance of deviance) was mea-
sured on a six-item Likert scale that assessed the
degree to which social norms are pervasive, clear-
ly defined, and reliably imposed within nations.
Example scale items include There are many
social norms that people are supposed to abide by
in this country,”“In this country, if someone acts
in an inappropriate way, others will strongly dis-
approve,and People in this country almost
always comply with social norms.The results
show strong support for the reliability and valid-
ity of the measure (20). Ecological factor analyses
and Procrustes factor analysis in all 33 nations
illustrate that the scale exhibits factor validity
and measurement equivalence. Analyses show
that the strength of social norms and tolerance of
deviance is a shared collective construct: There
is high within-nation agreement in each nation
(M) = 0.85], high between-nation
variability [F(32, 6,774) = 31.23, P<0.0001;
intraclass correlation (ICC)(1) = 0.13], and high
reliability of the tightness-looseness scale means
[ICC(2) = 0.97]. The scale has high convergent
validity with expert ratings, unobtrusive measures,
and survey data from representative samples; is
able to adequately discriminate between cultural
regions; and is distinct from other cultural dimen-
sions (20) (tables S1 and S2).
The degree of constraint across a wide range
of everyday social situations was measured
through adaptations to Price and Bouffards
established measure (20). Participants rated the
appropriateness of 12 behaviors (i.e., argue, eat,
laugh, curse/swear, kiss, cry, sing, talk, flirt, listen
to music, read newspaper, bargain) across 15 sit-
uations (i.e., bank, doctors office, job interview,
library, funeral, classroom, restaurant, public park,
bus, bedroom, city sidewalk, party, elevator, work-
place, movies), resulting in a total of 180 behavior-
situation ratings (20). For a given situation, the
mean appropriateness ratings across behaviors
indicate the degree of situational constraint: Low
values indicate that there are few behaviors con-
sidered appropriate in that situation, whereas
high values indicate that a wide range of behav-
iors are considered appropriate in that situation.
Country-level scores of situational constraint were
derived by averaging scores across situations.
Analyses illustrate that the situational constraint
measure is a shared collective construct within
nations (20): There is high within-nation agree-
ment about the level of constraint in everyday
situations in each nation [r
high between-nation variability in situational con-
straint [F(32, 6790) = 92.9, P< 0.0001; ICC(1) =
0.31], and high reliability of the situational con-
straint means [ICC(2) = 0.99]. There is strong con-
struct validity of the measure (20). Respondents
in each nation also provided direct ratings regard-
ing whether the 15 situations had clear rules for
appropriate behavior, called for certain behaviors
and not others, required people to monitor their
behavior or watch what they do,and allowed in-
dividuals to choose their behavior (reverse-coded),
the average of which is highly correlated with the
behavior-situation ratings (r=0.74,P<0.001).
The correlation of the current situational constraint
data in the United States with those reported by
Price and Bouffard is 0.92 (P<0.001)(20),
which suggests that the degree of constraint across
situations is generally stable across time.
Psychological processes ( prevention focus, self-
regulation strength, need for order, self-monitoring)
were assessed with well-validated measures (20).
Procrustes factor analysis of all of the measures
across the 33 nations all evidenced high equiv-
alence and high degrees of cross-national varia-
tion (20).
To test our predictions, we first examine the
relationships between tightness-looseness and
ecological and historical institutions. Because
many of these variables are associated with na-
tional wealth, we controlled for nationsGNP
per capita to examine their unique relationships
with tightness-looseness. We next illustrate how
tightness-looseness is related to the strength of
everyday situations and examine the cross-level
relationship between the strength of situations
and numerous psychological processes with the
use of hierarchical linear modeling. We provide a
test of the overall model with multilevel struc-
tural equation analysis (20).
Table S3 illustrates that nations that have
encountered ecological and historical threats have
much stronger norms and lower tolerance of de-
viant behavior. Tight nations have higher popula-
tion density in the year 1500 (r=0.77,P=0.01),
in the year 2000 in the nation (r=0.31,P=0.10),
and in the year 2000 in rural areas (r= 0.59; P=
0.02), and also have a higher projected popula-
tion increase (r= 0.40, P= 0.03). Tight nations
have a dearth of natural resources, including a
lower percentage of farmland (r=0.37, P=
0.05), higher food deprivation (r=0.52,P<0.01),
lower food supply and production (r=0.36, P=
0.05, and 0.40, P= 0.03, respectively), lower
protein and fat supply (rs=0.41 and 0.46, Ps=
0.03 and 0.01), less access to safe water (r=0.50,
P= 0.01), and lower air quality (r=0.44, P=
0.02), relative to loose nations. Tight nations face
more disasters such as floods, tropical cyclones,
and droughts (r=0.47,P= 0.01) and have had
more territorial threats from their neighbors during
the period 19182001 (r=0.41,P= 0.04). His-
torical prevalence of pathogens was higher in tight
Fig. 1. A systems model of tightness-looseness.
27 MAY 2011 VOL 332 SCIENCE
on May 27, 2011www.sciencemag.orgDownloaded from
nations (r=0.36, P=0.05), as were the number
of years of life lost to communicable diseases (r=
0.59, P< 0.01), the prevalence of tuberculosis (r=
0.61, P< 0.01), and infant and child mortality
rates (rs = 0.42, P= 0.02, and 0.46, P= 0.01).
Tightness-looseness is reflected in societal
institutions and practices (table S3). Tight nations
suppresses dissent (r= 0.47, P= 0.01), less open
media overall (r=0.53, P< 0.01), more laws
and regulations and political pressures and
controls for media (rs = 0.37 to 0.62, Ps
0.05), and less access to and use of new com-
munication technologies (r=0.38, P= 0.04).
Tight nations also have fewer political rights and
civil liberties (rs=0.50 and 0.45, Ps0.01).
Criminal justice institutions in tight nations are
better able to maintain social control: There are
morepolicepercapita(r=0.31,P= 0.12), stricter
punishments (i.e., retention of the death penalty)
(r= 0.60, P< 0.01), and lower murder rates and
burglary rates (rs=0.45 and 0.47, Ps < 0.01)
and overall volume of crime (r=0.37, P= 0.04).
Tight nations are more religious, with more people
attending religious services per week (r=0.54,
P< 0.01) and believing in the importance of god
in life (r= 0.37, P<0.05) (20). The percentage of
people participating in collective actions (e.g., sign-
ing petitions, attending demonstrations) is much
lower in tight nations (r=0.40, P=0.03), and
more people report that they would never engage
in such actions (r= 0.36, P= 0.05) in comparison
to loose nations.
Tightness-looseness is also related to the
strength of everyday recurring situations within
nations. As predicted, there is much higher situa-
tional constraint in tight versus loose nations (r=
0.55, P<0.01) (22). In other words, there is much
higher constraint across everyday situations
including the bank, public park, library, restaurant,
bus, workplace, party, classroom, and the like
in loose nations, and much lower constraint across
such everyday situations in tight nations (20). Hi-
erarchical linear modeling intercept-as-outcomes
models showed that higher levels of situational
constraint are significantly related to greater pre-
vention self-guides [higher cautiousness: g
1.48, t(31) = 7.54, P<0.01; higher dutifulness:
=1.11,t(31) = 5.05, P<0.01], greater self-
regulation strength [higher impulse control: g
1.18, t(31) = 6.60, P<0.01], higher needs for
structure [g
=2.67,t(31) = 5.76, P<0.01], and
higher self-monitoring [g
= 0.94, t(31) = 3.69,
P<0.01] (23). This suggests that societal mem-
berspsychological characteristics are attuned to
and supportive of the degree of constraint versus
latitude in the larger cultural context. Multilevel
structural equation analyses that simultaneously
tested the proposed relations in Fig. 1 illustrated
very good fit to the data (20).
In all, the data illustrate that tightness-
looseness, a critical aspect of modern societies
that has been heretofore unexplored, is a part of a
Table 1. Sample characteristics of the 33 nations.
Nation Data collection site(s) Language
of survey
Number of
Mean age
Australia Melbourne English 230 25.4 T10.0 69.1 63.9 4.4
Austria Linz German 194 31.6 T11.8 51.5 41.8 6.8
Belgium Leuven (Flanders region) Dutch 138 33.3 T14.3 73.2 50.7 5.6
Brazil o Paulo Portuguese 196 27.5 T9.4 72.3 40.3 3.5
Estonia Tartu Estonian 188 32.0 T16.8 86.6 52.1 2.6
France Paris, Cergy English 111 25.2 T4.1 37.8 67.6 6.3
Germany (former East) Chemnitz German 201 31.6 T12.2 66.7 49.3 7.5
Germany (former West) Rhineland-Palatine/Frankfurt German 312 32.5 T14.5 63.8 51.6 6.5
Greece Athens Greek 275 30.9 T11.3 56.7 45.1 3.9
Hong Kong Hong Kong Chinese 197 27.3 T11.7 68.0 53.8 6.3
Hungary Budapest, Szeged Hungarian 256 30.8 T10.9 42.2 48.0 2.9
Iceland Reykjavík Icelandic 144 36.3 T13.3 67.4 41.7 6.4
India Ahmedabad, Bhubneswar,
Chandigarh, Coimbatore
Hindi 222 27.8 T9.6 54.1 52.3 11.0
Israel Tel-Aviv, Ramat-Gan,
Jerusalem, Petach-Tikva
Hebrew 194 30.2 T10.7 60.3 48.5 3.1
Italy Padova Italian 217 29.6 T10.3 40.1 53.0 6.8
Japan Tokyo, Osaka Japanese 246 33.2 T14.9 55.7 48.8 8.6
Malaysia Bandar Baru Bangi Malay 202 29.5 T9.1 49.5 45.0 11.8
Mexico Mexico City Spanish 221 27.7 T11.6 42.1 40.3 7.2
Netherlands Groningen Dutch 207 29.8 T11.9 55.6 53.1 3.3
New Zealand Wellington English 208 29.9 T13.0 64.4 61.1 3.9
Norway Bergen Norwegian 252 31.8 T11.0 56.7 46.0 9.5
Pakistan Hyderabad Urdu 190 30.0 T9.8 51.1 52.6 12.3
Peoples Republic of China Beijing Chinese 235 29.4 T11.5 45.9 53.2 7.9
Poland Warsaw Polish 210 28.5 T12.4 65.2 51.9 6.0
Portugal Braga Portuguese 207 28.5 T11.6 54.6 58.0 7.8
Singapore Singapore English 212 26.1 T6.7 59.0 49.1 10.4
South Korea Seoul Korean 196 26.2 T7.5 61.2 73.5 10.0
Spain Valencia Spanish 172 30.2 T9.6 66.9 40.1 5.4
Turkey Istanbul Turkish 195 32.0 T14.4 53.3 45.6 9.2
Ukraine Odessa Ukrainian 184 30.8 T12.7 56.5 44.6 1.6
United Kingdom Brighton English 185 29.9 T11.5 67.0 51.4 6.9
United States Washington, DC;
Maryland; Virginia
English 199 31.4 T13.7 60.3 48.2 5.1
Venezuela Caracas Spanish 227 35.8 T10.0 60.4 1.3 3.7
Totals/means 6823 30.1 T11.3 58.6 49.2 6.5 SCIENCE VOL 332 27 MAY 2011 1103
on May 27, 2011www.sciencemag.orgDownloaded from
system of interrelated distal and proximal factors
across multiple levels of analysis. In addition to
explicating how tight and loose cultures vary in
modern societies, this research has implications
for understanding and modeling how tight and
loose cultures are maintained and changed. Sub-
stantial top-down or bottom-up changes in any of
the levels in the model may trigger a rippling
effect to other levels, resulting in changes in tight
or loose cultures.
As culture is fundamentally a system, causal
inferences regarding the direction of the relation-
ships need further examination, particularly giv-
en that they are likely reciprocal. Future research
should also apply the basic principles of the
current work to explore variation in tightness-
looseness at other levels of analysis (e.g., regions).
We also note that the samples in this study are
not representative of each nation. However, the
diverse backgrounds of the participants, high agree-
ment among different subgroups, and correlations
with other measures drawn from representative
samples lend confidence to the generalizability of
the results (20).
This research illuminates the multitude of dif-
ferences that exist across tight and loose cultures.
From either systems vantage point, the other
systemcould appear to be dysfunctional, unjust,
and fundamentally immoral, and such divergent
beliefs could become the collective fuel for cul-
tural conflicts. Indeed, as Herodotus (1) remarked
centuries ago, if one were to order all mankind
to choose the best set of rules in the world, each
group would, after due consideration, choose its
own customs; each group regards its own as
being the best by far(p. 185). Such beliefs fail
to recognize that tight and loose cultures may be,
at least in part, functional in their own ecolog-
ical and historical contexts. Understanding tight
and loose cultures is critical for fostering cross-
cultural coordination in a world of increasing global
References and Notes
1. Herodotus, The Histories (Oxford, New York, 1998;
R. Waterfield, Transl.).
2. P. J. Pelto, Trans Action 5, 37 (1968).
3. H. Barry III, I. L. Child, M. K. Bacon, Am. Anthropol. 62,
51 (1959).
4. J. W. Berry, Adv. Exp. Soc. Psychol. 12, 177 (1979).
5. S. Kitayama, Psychol. Bull. 128, 89 (2002).
6. A. Fiske, S. Kitayama, H. R. Markus, R. Nisbett,
in The Handbook of Social Psychology, D. Gilbert,
S. T. Fiske, G. Lindzey, Eds. (Oxford, New York, 1998),
vol. 2, pp. 915981.
7. H. C. Triandis, The Analysis of Subjective Culture
(Wiley, New York, 1972).
8. We acknowledge that these relationships are only
probabilistic, as cultures can find equifinal solutions to
ecological and historical threats (24). Moreover, the
degree of tightness-looseness in societies can further
reinforce the ecological context (6), making these
relationships potentially reciprocal.
9. J. J. Arnett, J. Marriage Fam. 57, 617 (1995).
10. A. Norenzayan, A. F. Shariff, Science 322,58
11. W. Mischel, in Personality at the Crossroads,
E. Magnusson, N. S. Endler, Eds. (Erlbaum, Hillsdale,
NJ, 1977).
12. R. H. Price, D. L. Bouffard, J. Pers. Soc. Psychol. 30,
579 (1974).
13. E. D. Boldt, Can. J. Sociol. 3, 349 (1978).
14. E. Goffman, Behavior in Public Places: Notes on the
Social Organization of Gatherings (Greenwood, Westport,
CT, 1963).
15. S. Kitayama, H. R. Markus, H. Matsumoto,
V. Norasakkunkit, J.Pers.Soc.Psychol.72, 1245
16. E. T. Higgins, Psychol. Rev. 94, 319 (1987).
17. R. F. Baumeister, T. F. Heatherton, Psychol. Inq. 7,
1 (1996).
18. S. L. Neuberg, J. T. Newsom, Personal. Processes Indiv.
Diff. 65, 113 (1993).
19. M. Snyder, J. Pers. Soc. Psychol. 30, 526
20. See supporting material on Science Online.
21. Most samples corresponded to nations; however, where
subnational boundaries could be identified on the basis
of historical circumstances, they were treated as separate
samples (e.g., East and West Germany; Hong Kong and
Peoples Republic of China).
22. For ease of interpretation, the situational constraint score
was reversed such that high values are indicative of
higher constraint.
23. We also ran these analyses with a split-sample approach
(25) to eliminate single-source bias as an alternative
explanation for our findings. Within each country we
randomly assigned participants to one of two groups:
One group provided the situational constraint scores and
the other group provided the individual-difference scales.
These hierarchical linear modeling results were the same
as with the full sample.
24. D. Cohen, Psychol. Bull. 127, 451 (2001).
25. C. Ostroff, A. J. Kinicki, M. A. Clark, J. Appl. Psychol. 87,
355 (2002).
Acknowledgments: Supported by NSF grant 9910760
and U.S. Army Research Lab and Research Office
grant W911NF-08-1-0144 (M.J.G.), Turkish
Academy of Sciences (Z.A.), Polish Academy of
Sciences (P.B.), Australian Research Council (Y.K.),
and Estonian Ministry of Science (A.R.). We thank
C. B. Bruss and R. Mohr for their help in preparing
this manuscript.
Supporting Online Material
Materials and Methods
Tables S1 to S6
14 September 2010; accepted 6 April 2011
27 MAY 2011 VOL 332 SCIENCE www.sciencemag.org1104
on May 27, 2011www.sciencemag.orgDownloaded from
Supporting Online Material for
Differences Between Tight and Loose Cultures: A 33-Nation Study
Michele J. Gelfand,* Jana L. Raver, Lisa Nishii, Lisa M. Leslie, Janetta Lun, Beng Chong Lim,
Lili Duan, Assaf Almaliach, Soon Ang, Jakobina Arnadottir, Zeynep Aycan, Klaus Boehnke,
Pawel Boski, Rosa Cabecinhas, Darius Chan, Jagdeep Chhokar, Alessia D’Amato, Montse Ferrer,
Iris C. Fishchlmayr, Ronald Fischer, Marta Fülöp, James Georgas, Emiko S. Kashima,
Yoshishima Kashima, Kibum Kim, Alain Lempereur, Patricia Marquez, Rozhan Othman,
Bert Overlaet, Penny Panagiotopoulou, Karl Peltzer, Lorena R. Perez-Florizno,
Larisa Ponomarenko, Anu Realo, Vidar Schei, Manfred Schmitt, Peter B. Smith, Nazar Soomro,
Erna Szabo, Nalinee Taveesin, Midori Toyama, Evert Van de Vliert, Naharika Vohra,
Colleen Ward, Susumu Yamaguchi
*To whom correspondence should be addressed. E-mail:
Published 27 May 2011, Science 332, 1100 (2011)
DOI: 10.1126/science.1197754
This PDF file includes:
Materials and Methods
Table S1 to S6
*To whom correspondence should be addressed. E-mail:
Materials and Methods
Participants. Data were gathered
from a total of 6960 respondents in 33
nations across five continents. After
removing incomplete surveys with
unusable data, the final sample for
analyses consisted of 6823
participants. The final sample sizes
included: Australia (n = 230), Austria
(n = 194), Belgium (n = 138), Brazil
(n = 196), Estonia (n = 188), France
(n = 111), Former East Germany (n =
201), Former West Germany (n =
312), Greece (n = 275), Hong Kong (n
= 197), Hungary (n = 256), Iceland (n
= 144), India (n = 222), Israel (n =
194), Italy (n = 217), Japan (n = 246),
Malaysia (n = 202), Mexico (n = 221),
the Netherlands (n = 207), New
Zealand (n = 208), Norway (n = 252),
Pakistan (n = 190), People’s Republic
of China (n = 235), Poland (n = 210),
Portugal (n = 207), Singapore (n =
212), South Korea (n = 196), Spain (n
= 172), Turkey (n = 195), Ukraine (n
= 184), the United Kingdom (UK, n =
185), the United States (US, n = 199),
and Venezuela (n = 227). The gender
distribution was 58.6% female and
41.4% male. The average percentage
of university students in the samples
was 49.2%, with adults comprising
the remainder (50.8%). The mean age
of participants was 30.1 years, and the
average amount of work experience
was 8 years. With regard to the socio-
economic status of the participants,
73.2% reported that they were middle
We employed a theoretically based
sampling strategy that aimed at
maximizing the variability of nations
with regard to the expected correlates
of tightness-looseness (e.g.,
population density, scarcity of
resources) (S1). All data were
collected during 2000-2003. Within
nations, our participant sampling
strategy was aimed at maximizing the
variability of participants (S2). In each
nation, a diverse sample of adults was
recruited through a combination of
strategies, which included directly
recruiting adults who were either
waiting in public areas or were
enrolled in non-credit continuing
education classes. They represented a
variety of occupations, including
business and financial operations,
management and sales (18.4%),
education, training and library
services (17.2%), office and
administrative support (9.8%),
architecture (5.8%), food preparation
and personal services (5%), computer
and mathematical (4.5%), community
and social services (3.8%), among
others. Collaborators also recruited
approximately 100 students who filled
out the survey in exchange for course
credit (excluding Venezuela, where
we were unable to obtain a separate
student subsample). These procedures
resulted in a sample comprised of
approximately 200 respondents in
each nation who were diverse with
regard to their personal and
professional characteristics.
Participants responded to our
tightness-looseness scale, our measure
of situational constraint, individual
differences measures, and
demographics. The survey instrument
was administered in 21 languages.
The five English-speaking nations
administered the English version of
the survey (with spelling and grammar
adapted to local norms; i.e., Australia,
New Zealand, UK, US). Surveys were
also administered in English in
Singapore and France where the
respondents attended or were
affiliated with English-speaking
institutions, and English was deemed
the most appropriate language by
collaborators. Three nations
administered a Spanish-language
version of the survey (i.e., Mexico,
Spain, and Venezuela). Two nations
each administered the surveys in
Portuguese (i.e., Brazil, Portugal),
German (East and West Germany
samples, Austria), Dutch (i.e.,
Belgium, Netherlands), and Chinese
(i.e., Hong Kong, PRC). Collaborators
Differences between Tight and Loose Cultures:
A 33-Nation Study
Michele J. Gelfand*, Jana L. Raver, Lisa Nishii, Lisa M. Leslie, Janetta Lun,
Beng Chong Lim, Lili Duan, Assaf Almaliach, Soon Ang, Jakobina Arnadottir,
Zeynep Aycan, Klaus Boehnke, Pawel Boski, Rosa Cabecinhas, Darius Chan,
Jagdeep Chhokar, Alessia D’Amato, Montse Ferrer, Iris C. Fischlmayr,
Ronald Fischer, Marta Fülöp, James Georgas, Emiko S. Kashima, Yoshishima
Kashima, Kibum Kim, Alain Lempereur, Patricia Marquez, Rozhan Othman,
Bert Overlaet, Penny Panagiotopoulou, Karl Peltzer, Lorena R. Perez-
Florizno, Larisa Ponomarenko, Anu Realo, Vidar Schei, Manfred Schmitt,
Peter B. Smith, Nazar Soomro, Erna Szabo, Nalinee Taveesin, Midori
Toyama, Evert Van de Vliert, Naharika Vohra, Colleen Ward, Susumu
This PDF File includes Materials and Methods
Tables S1 to S6
Gelfand et al.
in the remaining nations administered
versions in their local languages:
Estonian (Estonia), Greek (Greece),
Hungarian (Hungary), Icelandic
(Iceland), Hindi (India), Hebrew
(Israel), Italian (Italy), Japanese
(Japan), Korean (Korea), Malay
(Malaysia), Norwegian (Norway),
Urdu (Pakistan), Polish (Poland),
Turkish (Turkey), and Ukrainian
We used the translation-
backtranslation procedure, which is
the most widely accepted method for
conducting survey translations (S3).
This procedure entails having the
survey instrument translated from the
original language (i.e., English) to the
second language (i.e., local languages)
by one translator, and then having a
second independent translator re-
translate the survey back to the
original language. In cases where
discrepancies between the two
versions arose, the translators
discussed the discrepancies and
resolved them by selecting the most
appropriate and understandable
translation. In each nation where
translation was necessary,
collaborators selected the two
translators and oversaw this process to
ensure that the final version of the
survey was translated accurately.
Scales in all languages are available
from the first author.
Response sets vary across
nations, such that individuals in
some nations are systematically more
likely to provide extreme responses
and acquiesce to survey items than in
others (S4-5). To reduce the
influence of cross-cultural response
sets on our data, we used procedures
outlined by Van de Vijer and Leung
(S5). We used the within-subject
standardization procedure that
adjusts the scores for each individual
using the mean for that individual
across all variables (S5-6). To do so,
the mean for each person’s responses
to all of the items in the survey was
first calculated. We then standardized
all items in the survey by subtracting
each item from that person’s mean
response to all items. Standardized
data were used in all analyses. The
results did not change substantially
whether standardized or
unstandardized scores were used. All
data are available from the first
Tightness-Looseness Scale:
Strength of Social Norms and
Tolerance of Deviance
We developed a generalized
measure of tightness-looseness that
assessed the degree to which social
norms are pervasive, clearly defined,
and reliably imposed within nations.
As per recommendations for scale
development, items were generated
deductively based on our construct
definition in order to maximize
content validity (S7-9). Nine items
were first generated by a set of 5 team
members, and thereafter collaborators
involved in the study evaluated the
items in terms of the degree to which
the items mapped onto the construct
definition, how clear, concise,
readable, distinct, and redundant they
viewed each of the items to be,
whether the items would be easily
understood by respondents in their
country as intended (once translated,
where appropriate), and the extent to
which the items as a set demonstrated
content validity and adequate
construct coverage. Minor wording
changes were made to the items and
three items were dropped due to
redundancy and/or problems in
wording. !
The final version of the scale
included six statements regarding the
clarity and number of social norms,
the degree of tolerance for norm
violations, and overall compliance
with social norms in each nation. The
survey respondents received the
following instructions:
The following statements refer to
[COUNTRY NAME] as a whole.
Please indicate whether you agree or
disagree with the following statements
using the following scale. Note that
the statements sometimes refer to
"social norms,” which are standards
for behavior that are generally
We limited the number of reverse
coded items in the scale because
psychometric research suggests that
1. There are many social norms that people are supposed to abide by
in this country.
2. In this country, there are very clear expectations for how people
should act in most situations.
3. People agree upon what behaviors are appropriate versus
inappropriate in most situations this country.
4. People in this country have a great deal of freedom in deciding how
they want to behave in most situations. (Reverse coded)
5. In this country, if someone acts in an inappropriate way, others will
strongly disapprove.
6. People in this country almost always comply with social norms.
Gelfand et al.
reverse scoring can introduce method
factors that supersede substantive
factors (S10-14), resulting in a
separate dimension that has little
theoretical meaning.
In keeping with prevailing
standards for scale development (S7-
9) we established the reliability and
validity of the scale by assessing the
degree to which the scale has factor
validity at the national level, scale
equivalence across the 33 nations,
adequate reliability, high within-
nation agreement and between-nation
variability, high convergent validity,
and is distinct from other known
cultural values and beliefs. Each is
discussed in turn below.
Reliability and Validity of the
Tightness-Looseness Measure and
Scale Equivalence
The validation of a new instrument
should involve an empirical
evaluation of the underlying factor
structure using exploratory factor
analysis (S7-9). Evidence of construct
validity is indicated by the extent to
which the factor structure that
emerges from exploratory factor
analyses aligns with theoretical
expectations. We expected that
tightness-looseness scores would be
explained by one underlying factor at
the national level with all items
loading in the expected direction. As
predicted, an exploratory factor
analysis with principal axis estimation
indicated a clear one-factor solution,
accounting for 62% of the variance
1 = 3.70,
2-6 < 1.01). Item loadings
were.68 or greater, with the exception
of the reverse-coded item that had a
loading of .26 which was in the
theorized direction. The scale also
demonstrated very good reliability
(S7) at the national level (
= .85).
We also demonstrate validity for
the scale by showing its structural
equivalence (i.e., similarity in factor
structures) across nations. Following
established standards, we used
Procrustes Factor Analysis (PFA) to
examine the measurement equivalence
of the scale across cultures (S5, S15).
PFA is a special form of exploratory
factor analysis that involves (a) using
individual-level data to calculate an
overall, or normative, factor solution
across nations and (b) calculating an
individual-level factor solution
separately in each nation and rotating
those solutions so that they match the
normative solution as closely as
possible. Through this targeted
rotation process, one can examine the
extent to which the factor solution in
each nation deviates from the
normative structure. PFA has been
used to establish structural
equivalence across nations for a wide
variety of constructs, and evidence
suggests that it is an effective means
of establishing structural equivalence
across nations (S2, S15-19). We used
the steps outlined below to establish
structural equivalence of the
tightness-looseness measure.
To conduct the PFA, we first
calculated item intercorrelations in all
33 nations. We then transformed the
item intercorrelations into z scores,
averaged the z-scores across samples,
and transformed the z-scores back into
correlations to form the normative
item intercorrelation matrix. We
conducted an exploratory principal
axis factor analysis on the normative
item intercorrelation matrix to
determine the normative factor
structure of the scale. Next, we
conducted an exploratory principal
axis factor analysis separately in
nation and then subjected the nation-
specific solutions to Procrustes
rotation. The Procrustes rotation
procedure rotates the country-specific
solution so that it matches the
normative solution as closely as
possible by maximizing the fit of
country specific solutions with the
normative solution structure (S5). To
demonstrate scale validity, we
calculated the identity coefficient, the
most stringent index of the fit between
the normative loadings and the
Procrustes-rotated nation specific
loadings (S5). The mean identity
coefficient across nations was .97
(Mdn = .98, SD = .03), and the
identity coefficient exceeded
recommended .90 cutoff for 32 of the
33 countries. The one exception was
Brazil, which had an identity
coefficient of .87. In sum, factor
analyses at both the national and
individual level illustrate strong
validity for the measure of tightness-
Tightness-looseness is a shared
cultural construct with high within-
nation agreement and high between-
nation variance. Tightness-looseness
is conceptualized as a shared construct
regarding the degree to which social
norms are pervasive, clearly defined,
and reliably imposed. We used the
nation as the level of analysis to test
our predictions and provide empirical
evidence that justifies this level of
analysis below. Although there can be
variability in nations, many have
argued that there are forces toward
integration (e.g., common political
and educational systems, media,
markets, dominant language, national
symbols) that produce substantial
sharing of culture within nations (S6,
S20-22). There is also substantial
shared knowledge among people
within a nation because they are a
coordinating unit in dealing with
distinct ecological and territorial
threats and in forming and supporting
cultural institutions that regulate
social behavior. Examining the
relationship between national levels of
tightness-looseness and ecological and
historical factors, socio-political
institutions, and citizens’ attitudes is
thus theoretically justified.
We specifically theorized that
tightness-looseness emerges as a
referent-shift collective construct
Gelfand et al.
(S23) where perceptions of what is
normative in a given nation are,
generally speaking, shared among its
members. Consistent with this level of
theory, our level of measurement
reflects individuals’ ratings of the
country as a whole (S23). Following
recommendations from levels of
analysis experts (S24-25), we tested
our assumption regarding the nation
as the appropriate level of analysis by
examining whether (a) individuals
have low variability in their
perceptions of the strength of social
norms and degree of tolerance for
deviance in their nation (which is
indicated by high rwg(j) values; (S26)
(b) there is significant between-nation
variance in the construct (which is
indicated by high ICC(1) values), and
(c) national means are reliable at the
culture level (which is indicated by
high ICC(2) values).
We first calculated the rwg(j) value
for each nation (S26-27), which is an
index of the extent to which
individuals within a given nation
agree on the level of tightness-
looseness within that nation, and
therefore provide similar responses to
the tightness-looseness items. For
each nation, this index is calculated by
comparing the observed variance in
tightness-looseness perceptions to the
variance that would be expected by
chance. If there was no agreement in
perceptions of tightness-looseness
within a given nation, all response
options for the tightness-looseness
measure would be selected with equal
frequency. Thus, the distribution that
would be expected by chance is a
uniform distribution. Specifically:
where is the mean of the
observed variances of the J items and
2 is the expected variance of the
uniform distribution that would be
expected if there was no agreement in
tightness-looseness. The expected
variance of the uniform distribution
equals (A2 1)/12, where A equals the
number of response options for each
of the J items. The tightness-looseness
items had six response options, which
means that
2 = 2.92. We calculated
rwg(j) in each nation (M = .85, Mdn =
.86, SD = .08), and found that the
mean rwg(j) value across all nations
exceeded the recommended cutoff
point of .70 (S25). Across nations,
tightness-looseness scores from the
student and working adult subsamples
are highly correlated (r = .83), as are
the scores of females and males (r =
.84), suggesting substantial cultural
unity (S20). This illustrates that there
is high within-nation agreement on
We calculated two additional
aggregation statistics, ICC(1) and
ICC(2). Two types of inferences can
be drawn from the ICC(1) statistic
(S24). First, ICC(1) is an index of the
degree of variance in tightness-
looseness that is explained by cultural
membership. Second, ICC(1) is an
inter-rater reliability index that
reflects the extent to any individual
response is a reliable indicator of the
mean tightness-looseness score for a
given nation. ICC(2) is similarly a
reliability index, but it reflects the
extent to which the national-level
mean scores for tightness-looseness
are reliable. ICC(1) and ICC(2) are
both a function of the degree of
variance in tightness-looseness that
resides within versus between nations.
Thus, the first step in calculating
ICC(1) and ICC(2) is to conduct a
one-way random-effects analysis of
variance (ANOVA) in which nation
membership is used to explain
individuals’ responses on the
tightness-looseness measure. The
following formulas (S28) can then be
used to calculate ICC(1) and ICC(2).
where MSB is the between-nation
mean square, MSW is the within-
nation mean square, and k is the
average number of individuals per
nation in the sample. The one-way
ANOVA for the tightness-looseness
measure produced a highly significant
F-value [F(32, 6,774) = 31.23, P <
.0001], indicating that there is high
between-nation variability in
tightness-looseness. Moreover, the
ICC(1) value exceeded the
recommended cutoff of .06 [ICC(1) =
.13], indicating that 13% of the
variance in tightness-looseness is
explained by nations and that the
tightness-looseness scale has high
inter-rater reliability. In addition, the
ICC(2) value far exceeded the
recommended cutoff of .70 [ICC(2) =
.97], thus indicating that the national-
level mean scores of tightness-
looseness scores are highly reliable.
Collectively, these results provide
strong justification for aggregation,
and show that tightness-looseness is a
shared, reliable construct with
significant between-nation variance
(S24-25). Tightness and looseness
index scores across the 33 nations can
be found in Table 1 in the main text.
Higher values indicate greater
tightness. The tightness and looseness
index scores are the original
standardized scores multiplied by 10
for easier reference.
Tightness-looseness scores have
convergent validity. Table S1 shows
that the tightness-looseness scale
demonstrated strong convergent
Gelfand et al.
validity. Scores on the measure are
highly correlated with expert ratings
on tightness-looseness of nations
(given by Harry Triandis, a leading
cultural psychologist) (r = .61, P <
.01). Convergent validity data also
suggest that there are greater pressures
toward uniformity in tight as
compared to loose nations. There are
fewer people who report they write
with their left-hand in tight than loose
cultures (r = -.61, P = .05) (S29). In
addition, there is much greater
accuracy of public clocks in tight as
compared to loose nations (r = -.60; P
< .01; lower values are indicative of
greater accuracy) (S30), illustrating a
greater collective concern with order
and uniformity in the former as
compared to the latter. The tightness-
looseness scale also correlates with
higher monitoring (more police per
capita), more severe punishments
(e.g., the death penalty), and fewer
challenges to societal institutions (see
the main text and table S3).
Table S1 shows that the tightness-
looseness scores are correlated in
expected ways with other data that
reflect higher compliance with norms
and intolerance for deviance. Data
from representative samples from the
World Values Survey (S31) show that
people in nations that score higher on
our tightness-looseness measure find
socially deviant behavior to be much
less justifiable. Respondents were
specifically asked to rate how
justifiable each of the following
behaviors is: claiming government
benefits to which you are not entitled,
avoiding a fare on public transport,
cheating on taxes if you have a
chance, buying stolen goods, someone
accepting a bribe in the course of
one’s duties, homosexuality,
prostitution, abortion, divorce,
euthanasia (or ending the life of the
incurably sick), and suicide. People in
tight nations find these behaviors
much less justifiable than people in
loose nations (r = -.48, P < .01), and
the variability of responses to these
questions is lower in tight nations (r =
-.56, P <. 01), suggesting a higher
degree of consensus in negative
attitudes toward deviance. Greater
restriction on behavior in tight nations
is also evident in more restricted
sexual behavior (e.g., more negative
attitudes toward casual sex, fewer
sexual partners) (r = -.44, P = .04)
(S32), and lower alcohol consumption
rates as compared to loose nations (r =
-.46, P = .01) (S33).
Other convergent data show that
there is a much greater concern with
social order in nations that are higher
on the measure of tightness-looseness.
People in tight nations more strongly
endorse the belief that the most
important responsibility of
government is to maintain order in
society (r = .61, P < .01), and are
more likely to believe that political
systems with “a strong leader who
does not have to bother with
parliament and elections” as well as
“army rule systems” are more
favorable (r = .38, P = .04) (S31). The
measure also shows that tight nations
are much less open to outside
influences given their potential threat
to social order. Individuals in tight
nations believe their way of life needs
to be protected against foreign
influence (r = .57, P = .02) (S34), and
prefer to not have immigrants as
neighbors (r = .43, P = .02) (S31).
Tight nations indeed have a smaller
population of international migrants (r
= -.32, P = .08) (S35), and they are
more likely to believe that their
culture is superior to others (i.e., more
likely to agree with the statement that
Our people are not perfect, but our
culture is superior to others”) (r = .60,
P = .01) (S34).
The tightness-looseness scale also
discriminates between cultural regions
in expected ways. Nations were
categorized into cultural regions
according to the Global Leadership
and Organizational Behavior
Effectiveness (GLOBE) clusters (S36)
as well as the Inglehart-Wenzel’s
cultural zones (S37). A one-way
ANOVA was performed on regions
that have more than one nation. There
were significant differences in the
measure of tightness and looseness
across the GLOBE clusters [F(6, 25)
= 7.94, P < .01]. The data showed that
Southern Asian and Confucian Asian
nations are the tightest (M = 11.69, SD
= 0.62 and M = 8.64, SD = 1.84
respectively) and Eastern European
nations are loosest (M = 3.60, SD =
1.65). Nordic/Germanic European (M
= 6.04, SD = 2.21), Latin European
(M = 5.87, SD = 1.80), Anglo (M =
5.09, SD = 1.30), and Latin American
(M = 4.78, SD = 2.11) nations are in
the mid-range. Turkey was not
included in this analysis because it
was the only nation in the Middle East
cluster, but as expected, is tight (M =
9.2), similar to the South Asia cluster.
The Inglehart-Wenzel cultural zone
classification showed the same pattern
of results [F(6, 26) = 9.62, P < .01].
South Asian nations (M = 10.92, SD =
1.22) are tightest, followed by
Confucian nations (M = 8.21, SD =
1.52), Catholic Europe (M = 5.95,
SD= 1.46), Protestant Europe (M =
5.75, SD = 2.40), English speaking
nations (M = 5.09, SD = 1.30), and
Latin American nations (M = 4.78,
SD= 2.11). Nations in the Ex-
communist cultural zone (i.e., Estonia,
Ukraine) are the least tight (M = 2.13,
SD = 0.69).
Tightness-looseness is related to
but distinct from other cultural
dimensions. Analyses show that
tightness-looseness is distinct from
other available culture level data,
including Hofstede’s (S38) five
dimensions, Schwartz’s value
dimensions (S39), Leung and Bond’s
(S2) five social axiom dimensions, the
nine value dimensions examined by
the GLOBE project (S36), Smith,
Dugan, and Trompenaar’s (S40)
Gelfand et al.
dimension of loyal involvement
versus utilitarian involvement,
Inglehart and Baker’s (S41) two value
dimensions, and Smith, Peterson, and
Schwartz’s (S42) five sources of
guidance. Tightness-looseness was
expected to have only small to
moderate correlations with some of
these dimensions, illustrating its
validity as a novel construct.
Distinction from Hofstede’s
cultural values. We expected
tightness-looseness to be related to but
distinct from Hofstede’s (2001) five
value dimensions (S38). Collectivism
refers to the degree to which
individuals feel strong ties to their
ingroups (S38), and does not pertain
to how pervasive social norms are or
how much tolerance there is for
deviant behavior. Tightness-looseness
and collectivism have been also been
empirically differentiated in
traditional societies in research using
the Human Area Relations Files.
Carpenter (S43) showed that the
correlation between the constructs is
.44. In modern nations, tightness and
collectivism are also expected to be
related but distinct constructs. Indeed,
Table S2 illustrates that individualism
(the opposite pole of collectivism)
was moderately and negatively
correlated with the tightness-looseness
measure (r = - .47, P = .01). The
distinction between collectivism and
tightness can also be discerned in
comparative correlations with other
national variables. Collectivism, for
example, is highly correlated with
national wealth whereas tightness has
no relationship with national wealth
(see data in economic indicators
section below). Tightness, but not
collectivism, is correlated with
variables such as history of conflict on
one’s territory, greater monitoring
(population per police officer), and
greater desire for societal order
(controlling for wealth) (all data are
available from the first author) (S44).
Tightness-looseness was also
expected to be related to but distinct
from power distance, or the extent to
which power is distributed equally in
societies (S38). Although tight
societies may be more hierarchical
given that hierarchy helps to reinforce
order and coordination, this need not
always be the case (cf. Pelto’s
example of Israeli Kibbutzim, which
traditionally were highly egalitarian)
(S45). As expected, Table S2 shows
that tightness-looseness and power
distance are distinct and moderately
and positively correlated (r = .42, P =
We did not expect any strong
relationship between tightness-
looseness and masculinity-femininity,
which is the degree to which societies
emphasize competition and
materialism versus cooperation and
fairness (S38). Theoretically,
tightness-looseness can emphasize
either of these two poles of Hofstede’s
dimension. Finally, tightness-
looseness was expected to be related
to but distinct from uncertainty
avoidance (S38). Although tight
societies may be higher on uncertainty
avoidance (i.e., the level of stress that
is experienced in a society in the face
of an unknown future), it is also
possible that the converse is true. That
is, because tight societies have many
clear norms, stress deriving from
uncertainty may be dramatically
reduced amongst its citizens.
Singapore, for example, is expected to
be tight, yet it ranked lowest on
Hofstede's index of uncertainty
avoidance. We also did not anticipate
any strong relationship between
tightness-looseness and Hofstede’s
dimension of long-term orientation.
The results presented in Table S2
show that uncertainty avoidance,
masculinity and long/short-term
orientation each was not significantly
related to tightness-looseness (P’s >
Distinction from cultural values
dimensions of Schwartz, GLOBE,
Smith, Dugan, & Trompenaars, and
Inglehart and Baker. We expected
that tightness-looseness is related to
but distinct from Schwartz’s (S39)
dimensions of harmony, conservatism,
hierarchy, mastery, affective and
intellectual autonomy, and egalitarian
commitment, with the strongest
correlations expected between
tightness-looseness and conservatism
(i.e., emphasis on maintaining status
quo, group solidarity and traditions)
and hierarchy (i.e., more accepting of
unequal distribution of power and
resources), similar to the predictions
for Hofstede’s collectivism and power
distance dimensions above. As shown
in Table S2, tightness-looseness is
related to but distinct from all of these
value dimensions. Tightness-
looseness has moderate correlations
with Schwartz’s (S39) scores on
conservatism, hierarchy and
egalitarian commitment (r = .43, P =
.04, r = .47, P = .03, and r = -.41, P =
.06, respectively). Affective and
intellectual autonomy, mastery, and
harmony are not related to tightness-
looseness (r’s = -.28 to .18, P’s > .16).
We also expected low to moderate
correlations with GLOBE’s “as is”
value dimensions (e.g., family
collectivism, institutional
collectivism, performance orientation,
power distance, gender egalitarianism,
assertiveness, uncertainty avoidance,
future orientation, and humane
orientation; S36), with the strongest
correlations anticipated for family and
institutional collectivism and power
distance for reasons cited above.
Correlations with the GLOBE
cultural dimensions of how things are,
or the “as is” value dimensions, (S36)
were consistent with expectations.
Tightness-looseness is moderately
correlated with family collectivism, or
the degree to which individuals
express pride, loyalty and
Gelfand et al.
cohesiveness to in-groups (r = .49, P
=.01) as well as institutional
collectivism, or the degree to which
institutional practices encourage and
reward the collective distribution of
resources and collective action (r =
.43, P = .03). Tightness-looseness is
also moderately related to GLOBE’s
future orientation (r = .47, P = .02).
There are also trends that tighter
nations are more focused on
performance and excellence
(performance orientation; r = .35, P =
.08) but have less investment in
minimizing gender inequality
(r = -.35, P = .08). Assertiveness,
power distance, uncertainty
avoidance, and humane orientation are
not significantly related to tightness-
looseness (r’s = -.29 to .32, P’s > .11).
Tightness-looseness was expected
to be related to but distinct from
Smith et al.’s (S40) dimension of loyal
involvement versus utilitarian
involvement, with tighter nations
having more loyal involvement and
looser nations having more utilitarian
involvement. As expected, Table S2
also shows that tightness-looseness is
moderately correlated with Smith et.
al.’s scores on loyal versus utilitarian
involvement, with tightness associated
with being involved in organizations
based on loyalty over utilitarian goals
(r = .45, P = .02).
Finally, we expected that
tightness-looseness would have low to
moderate correlations with Inglehart
and Welzel’s traditional versus
secular rational values and survival
versus self-expression values (S37,
S41), which reflect the contrast
between economic and physical
security with an emphasis on
subjective well-being, self-expression,
and quality of life. Table S2 indeed
shows that Tightness-looseness is
uncorrelated with Inglehart and
Welzel’s traditional versus secular-
rational and survival versus self-
expression values (S37) (r’s = -.11
and -.13, P’s > .50). In sum, tightness-
looseness is related to but distinct
from extant value dimensions.
Distinction from social axioms and
sources of guidance. Tightness-
looseness is expected to be related to
but distinct from social axioms (S2).
We expected that people in tight
nations will have higher fate control
beliefs, given that fatalism has been
associated with the perception that
others have total control over one’s
actions (S46-47). By contrast, given
there is more latitude in loose
cultures, we expected that people in
loose nations will have higher
flexibility beliefs. Finally, we
expected that tightness-looseness will
be positively related to spirituality
given that this construct is highly
related to religious practices and
observance. We did not have any
predictions regarding the relationship
between tightness-looseness and
beliefs in rewards for good effort and
application of relevant knowledge
(reward for application) or
expectations about negative outcomes
in life (cynicism). As predicted,
tightness-looseness is related to but
distinct from social axiom
dimensions. Table S2 shows that
nations higher on tightness are more
likely to endorse beliefs in fate (r =
.44, P = .03), and spirituality and
supernatural forces (r = .52, P <.01).
Tightness-looseness, however, is not
correlated with flexibility (r = -.20, P
= .33) or cynicism (r = .14, P = .49).
Tightness-looseness did correlate with
beliefs in reward for application (r =
.60, P <.01).
Lastly, we examined the
relationship between tightness-
looseness and sources of guidance in
nations. Smith et al. demonstrated that
cultures vary in the sources of
guidance upon which they rely when
managing everyday events, including
vertical sources (e.g., reliance on
formal rules and superiors), unwritten
rules, specialists, other coworkers, and
beliefs that are widespread in one’s
nation (S42). We expected that in tight
nations people will be particularly
likely to rely on beliefs that are
widespread in their nation as well as
vertical sources (i.e., the extent to
which managers rely on formal rules
and procedures), for similar reasons
for power distance and hierarchy
described above. As expected,
tightness-looseness is positively
related to the reliance on beliefs that
are thought to be widespread in one’s
nation (r = .54, P < .01) and to the use
of vertical sources (r = .40, P= .03).
Tightness-looseness was unrelated to
the tendency to consult unwritten
rules, specialists and coworkers (r’s =
-.18 to .18, P’s > .35).
We also note that the tightness-
looseness scale is not significantly
correlated with the acquiescence
index constructed by Hofstede (S6) (r
= .16, n = 26, P = .43) or with the
acquiescence index constructed by
Smith (S48) based on Schwartzs
value survey (r = .14, n = 31, P = .45)
Economic Indicators. We did not
expect tightness-looseness to be
highly related to how well a nation
performs economically. Table S2
shows that tightness-looseness was
not related to Gross National Product
(GNP) per capita (r = .05, P = .79) or
global growth competitiveness (r = -
.08, P = .68).
In sum, the results illustrate that
the tightness-looseness measure is
both reliable and valid. The factor
analysis results provided evidence of
factor validity and scale equivalence;
aggregation statistics showed high
within-nation agreement, high
between nation variability, and high
reliability of the national means.
Analyses of the scale illustrate that it
has high convergent validity and is
also distinct from other extant culture
constructs, including cultural values,
axioms, and sources of guidance.
Gelfand et al.
Together, these steps illustrate that the
measure of tightness-looseness
demonstrates strong validity and
reliability (S7-9).
Situational Constraint Measure
A central feature of situations that
has received attention in psychology
and numerous disciplines is the extent
to which situations differ in the range
of behavioral responses that are
considered appropriate, or the extent
to which the situation constrains or
affords opportunities for behavioral
options (S49-S54). When situations
are strong, there is a restricted range
of behavior that is deemed
appropriate, leaving little room for
individual discretion in determining
behavior. Because such situations
have strong behavioral demands,
deviations from expected patterns are
associated with an increased
propensity for social censure (S49).
By contrast, weak situations are
ambiguously structured, place few
external constraints on individuals,
and afford a wide range of behaviors
that are appropriate (S49). This
dimension of situations has been
referred to as the strength of situations
(S49), or situational constraint (S50).
In their seminal paper, Price &
Bouffard (S50) showed that
situational constraint can be reliably
assessed. Based on daily diary studies,
they selected situations that were
commonly recurring including being
in a class, in the park, on a bus, at a
family dinner, in the park, on a date,
in church, at a job interview, on a
sidewalk, at the movies, in a bar, in an
elevator, in a restroom, in one’s room,
in a dormitory lounge, and at a
football game. Participants in their
study rated the appropriateness of
numerous behaviors (run, talk, kiss,
write, eat, sleep, mumble, read, fight,
belch, argue, jump, cry, laugh, shout)
across each of the situations for all
possible behavior-in-situation
judgments on a scale of 0 (the
behavior is extremely in appropriate
in this situation to 9 (the behavior is
extremely appropriate in this
situation). For a given situation, one
can collapse the mean appropriateness
ratings across behaviors; a low value
is indicative of the fact that there are
few behaviors that are considered
appropriate in that situation. By
contrast, high values on a given
situation indicate that a wide range of
behaviors are considered as
appropriate in that situation. They
showed that situations such as church,
job interview, and elevator are high on
situational constraint whereas own
room, park, and dorm lounge are low
on situational constraint. Price &
Bouffard also showed construct
validity for the measure. After
computing situational constraint for
each situation from the behavior-
situation matrices, a different set of
participants provided direct ratings on
the same situations. They were asked
to rate all 15 situations on such items
as “To what extent does the situation
require that people monitor their own
behavior or ‘watch what they do?”;
“To what extent would the approval
of other people make a difference in
what most people would do in the
situation?”; “To what extent does the
situation call for or demand certain
behaviors and not others?”; and “To
what extent is the situation loaded in
terms of its potential for
embarrassment?” As predicted,
situations higher on constraint (from
the behavior-situation matrices) were
judged to be much more loaded for
personal embarrassment, to elicit
higher self-monitoring, to be
associated with approval-disapproval
by others, and to demand certain
behaviors and not others. In all, they
illustrated that situational constraint is
a valid construct that can be reliably
We built on this work and suggest
that although all cultures invariably
have strong and weak situations, tight
cultures maintain a much higher
overall degree of situational constraint
across everyday situations, whereas
loose cultures maintain a much lower
overall degree of situational constraint
across everyday situations.
Importantly, we theorize (and test the
assumption) that perceptions of
situational constraint are generally
shared among individuals in
culturesthat is, situational constraint
is a collective construct.
We based our survey assessment
of situational constraint upon the
methodology developed by Price and
Bouffard (S50), and we employed
additional steps (e.g., diary pilot
study, focus groups, iterative process
for selecting final behaviors and
situations) to ensure that the stimuli
and methodology would be cross-
culturally valid (S4).
Selection of Behavior and
Situation Stimuli: Pilot Study and
Multi-Nation Focus Groups. We
began with the set of 15 behaviors and
15 situations used by Price and
Bouffard (S50). In order to generate a
more extensive list of behaviors and
situations to evaluate for inclusion in
this cross-cultural study, we had six
individuals each complete a behavior
and situation diary for a 24-hour
period. They kept lists of all behaviors
that they performed and the situations
in which they performed these
behaviors. All participants’ lists were
compiled and synonyms were merged.
Next, four multi-cultural individuals
reviewed these lists and added
behaviors and situations that had
potential for cross-cultural
generalizability. Based upon these
methods, a total of 34 behaviors and
35 situations were chosen for further
investigation in multi-nation focus
groups across the 33 nations. We
limited the list of behaviors to those
that can physically be performed in
any setting (e.g., we excluded running
Gelfand et al.
and exercising because they cannot be
performed on a bus).
In line with recommendations for
selecting stimuli in cross-cultural
research (S4), collaborators ran focus
groups in their nations to ensure that
the behaviors and situations chosen
for the investigation were meaningful
and appropriate in each nation, and to
add any additional relevant behaviors
and situations. Each focus group was
conducted in the respondents’ native
language. The stimuli were translated
and bi-lingual collaborators
coordinated the sessions. Each
behavior and situation was evaluated
for: (1) whether it could be translated
into the native language, (2) whether
it was relevant in that culture, and (3)
whether there was more than one
interpretation for the situation being
evaluated. Collaborators provided a
report that summarized their focus
group discussions and provided any
additional behaviors and situations to
consider. The data from the focus
groups were compiled and the list of
behaviors and situations was updated
and re-evaluated. Additional rounds of
feedback from collaborators were
used to evaluate additional behaviors
and to refine the lists.
The final list of situations and
behaviors included those stimuli that
were translatable, relevant,
unambiguous in all cultures, and
representative of a wide variety of
behaviors and situations. We also
ensured that the final lists were
representative of theoretical
dimensions of behaviors and
situations identified in the literature.
For example, Wish and colleagues
(S55-56) identified formal vs.
informal and intense vs. superficial
(i.e., situations involving close
personal connections vs. situations
involving the general public or
acquaintances) as key dimensions
upon which interpersonal situations
varied, and thus the list includes a
range of formal, informal, personal,
and public situations. Similarly, the
degree to which behaviors are
cooperative and helpful vs.
competitive and/or neglecting has
been identified as a key dimension of
behaviors (S55-56; labeled as
associative vs. disassociative by
Triandis, S57; see also S58), and the
final list of behaviors reflects variance
on this dimension. For purposes of
this study, we included 15 situations
and 12 behaviors that were shown to
be clearly understood and relevant in
each nation, and they also represented
of a range of theoretically-based
dimensions of behaviors and
situations. The 15 situations included
bank, doctor’s office, job interview,
library, funeral, classroom, restaurant,
public park, bus, bedroom, city
sidewalk, party, elevator, workplace,
and movies. The 12 behaviors
included argue, laugh, curse/swear,
kiss, cry, sing, talk, flirt, listen to
music, read newspaper, bargain, and
Behavior x Situation Matrices. The
measure was administered to
participants through the behavior x
situation matrix procedure validated
by Price and Bouffard (S50).
Judgments about the appropriateness
of each of the 12 behaviors in each of
the 15 situations, which comprised the
measure, were made for all possible
combinations (N=180 judgments).
Participants were specifically asked:
From various sources in our
everyday lives we have all
developed a subjective
“impression” or “feeling” for the
appropriateness of any given
behavior in a particular situation.
In this study, we are interested in
your judgment of the
appropriateness of some particular
behaviors in some particular
settings. Your task in each case is
simply to rate, on a scale from 1
through 6, the appropriateness of
the particular behavior in the
situation that is given. The rating
scale is as follows:
1 = extremely inappropriate, 2
= very inappropriate, 3 =
somewhat inappropriate, 4 =
somewhat appropriate, 5 = very
appropriate, and 6 = extremely
They responded to the question
“How appropriate is this behavior in
this setting?” for each of the 180
behavior x situation pairings. For
example, participants were asked to
judge how appropriate it was to curse
in a library, eat in a bank, talk in the
movies, cry in a public park, etc. for
all possible behavior and situation
pairs (see Table S4 for example
items). Consistent with Price and
Bouffard, we calculated the situational
constraint of each situation by
averaging the appropriateness ratings
of all behaviors for a given situation.
We calculated average situational
constraint scores for each country by
averaging across situations. The
scores were reversed for presentation
in the main text such that high values
are indicative of high constraint.
We theorized that situational
constraint is a shared collective
construct (S23) where perceptions of
the range of behaviors seen as
appropriate across situations in a
given nation is, generally speaking,
shared among its members. Analyses
strongly supported this assumption.
We calculated aggregation statistics
separately for each of the 15 situations
in each of the 33 countries, treating
each of the 12 behaviors as items. The
mean and median of the rwg(j) values
across countries for each situation
ranged from .95-.96. Furthermore, all
rwg(j) values were greater than the
recommended cutoff point of .70. We
also assessed aggregation statistics for
the overall situational constraint
measure. To assess within country
agreement for the overall scores we
Gelfand et al.
calculated an rwg(j) value for each
country, treating the 15 situations as
items. The rwg(j) values provided strong
evidence of agreement for all 33
countries (M = .99, Mdn = .99, SD =
The one-way ANOVA for the
situational constraint measure
produced a highly significant F-value
[F(32, 6790) = 92.9, P < .0001],
indicating that there is high between-
nation variability in situational
constraint. Moreover, the ICC(1)
value exceeded the recommended
cutoff of .06 [ICC(1)=.31], indicating
that 31% of the variance in situational
constraint is explained by nations and
that the situational constraint measure
has high inter-rater reliability. The
ICC(1) values calculated across
countries separately for each situation,
ranged from .09 to .36. The ICC(2)
value for the overall situational
measure was .99, far exceeding the
recommended cutoff of .70, and the
ICC(2) values for the specific
situations ranged from .95 to .99.
Collectively, these results provide
strong justification for aggregation,
and namely that situational constraint
is a shared, reliable construct with
significant between-nation variance.
As per Price and Bouffard (S50),
we established the construct validity
of the situational constraint measure
by asking participants to provide
direct ratings of each of the situations.
The validation questions included: (1)
To what extent does the setting allow
people to behave as they choose? (2)
To what extent does the setting have
clear rules regarding appropriate
behavior? (3) To what extent does the
setting call for or demand certain
behaviors and not others? (4) To what
extent does the setting require that
people monitor their own behavior or
“watch what they do”? Participants
responded to these questions on a
scale ranging from 1 (not at all) to 5
(very much). See Table S5 for
example items. To avoid participant
fatigue, four different versions of the
survey were created, such that each
participant provided ratings for only
one of the situational constraint
validity questions for all 15 situations.
The measure of situational
constraint showed strong construct
validity. Mean ratings for each
situation in each country were
correlated with the index of situational
constraint for the situation from the
behavior-situation matrices. The four
ratings, namely whether the 15
situations allowed individuals to
choose their behavior (rating 1), have
clear rules regarding appropriate
behavior (rating 2), call for certain
behaviors and not others (rating 3),
and require people to monitor their
behavior or “watch what they do”
(rating4) were all highly
intercorrelated and loaded onto a
single factor (
1 = 2.71,
2-4 < .70;
explaining 68% of the variance) and
were averaged for an overall direct
rating of situational constraint. The
direct ratings on situations and
behavior-situation ratings were
correlated at .74 (P <.001), illustrating
construct validity for the measure
As another indication of construct
validity, we compared the situational
constraint scores found in the present
data in the United States, with those
reported by Price and Bouffard (S50)
collected in the United States more
than 30 years ago. We assessed eight
of the 15 situations originally studied
by Price and Bouffard. As shown in
Table S6, the rank order of the
situations assessed in 2003 and 1974
were highly similar. The correlation
between the degree of constraint for
the eight situations assessed in 1974
and 2003 was .92 (P < .001). This
suggests that the degree of constraint
across situations is stable across time.
Situational Constraint Affordances
of Psychological Processes
We assessed a number of
psychological constructs to test our
theory, including aspects of
prevention focus, self-regulation
strength, epistemic needs, and self-
monitoring ability.
Situational Constraint Affordances
of Prevention Focus: Cautiousness
and Dutifulness. Higgins (S59) argued
that people with strong normative
ought self-guides are concerned with
conforming to normative rules,
injunctions, and prescribed duties and
obligations. Normative guides thus
represent the “generalized other” and
evoke a prevention focus (S59). We
assessed aspects of prevention
regulatory focus through measures
from the cautiousness and dutifulness
subscales from Goldberg’s validated
International Personality Item Pool
(S60-61). We theorized that
individuals in nations that have high
situational constraint will be more
cautious and dutiful as compared to
individuals in nations that have low
situational constraint. Example
cautiousness items include, “I am very
careful to avoid making mistakes” and
“I choose my words with care,” “I
reflect on things before acting,” “I act
without thinking (reverse coded)”.
Example dutifulness items include, “I
behave properly” and “I stick to the
rules,” “I pay no attention to what is
asked” (reverse coded). All items
were rated on a 6-point scale where 1
= strongly disagree and 6 = strongly
Situational Constraint Affordances
of Self-Regulation Strength. We
theorized that individual in nations
where there is high situational
constraint will have higher self-
regulatory strength (self-control) than
individuals in nations where there is
low situational constraint. Having
high self-control is adaptive to the
preponderance of strong situations
Gelfand et al.
and is functional to the extent that it
helps individuals avoid the possibility
of being censored for inappropriate
behavior. By contrast, in cultures
characterized by primarily weak
situations, individuals will have less
of a need to show restraint, and thus,
will tend to be lower on self-control.
We assessed participants’ self-
regulation strength through the
impulse self-control subscale from
Goldberg’s International Personality
Item Pool (S60-61). Example impulse
self-control items include, “I keep my
emotions under control” and “I easily
resist temptations” All items were
rated on a 6-point scale where 1 =
strongly disagree and 6 = strongly
Situational Constraint Affordances
of Epistemic Needs. We expected that
individuals’ epistemic needs, or the
desire for clear knowledge and
information, will be related to the
degree of situational constraint within
nations. We assessed an aspect of
epistemic needs, the need for
structure, to examine this prediction.
Individuals who have a high need for
structure prefer an ordered
environment and rely on formalized
social scripts in their interactions with
others (S62). Such tendencies are
adaptive to strong situations with high
censoring of behavior. Accordingly,
individuals in nations with high
situational constraint were expected to
have a greater preference for structure.
By contrast, individuals in nations
with low situational constraint were
theorized to have a lower need for
structure as this is adaptive to a
weaker normative environment
wherein there is a wide range of
behavior that are permissible. We
assessed participants’ desire for order
and discomfort with ambiguity
through the Neuberg and Newsom’s
(S62) personal need for structure
(PNS) scale. Participants responded to
8 items on a 6-point scale where 1 =
strongly disagree and 6 = strongly
agree. Example items include: “I
enjoy having a clear and structured
mode of life”, “I like to have a place
for everything and everything in its
place”, “I find that establishing a
consistent routine enables me to enjoy
life more”, and “I don’t like going
into a situation without knowing what
I can expect from it.”
Situational Constraint Affordances
of Self-Monitoring Ability. Cultural
differences in situational constraint
were expected to be related to
individuals’ self-monitoring abilities,
or the ability to monitor and adjust
one’s behavior to the context.
Individuals in nations with high
situational constraint will engage in
more frequent self-monitoring, as the
need to comply with social norms so
as to avoid punishment is much
greater. In comparison, individuals in
nations with a low situational
constraint experience less of a need to
constantly monitor their behavior to
ensure compliance with social norms,
and thus should exhibit overall lower
self-monitoring ability. We assessed
participants’ self-monitoring ability
with Lennox and Wolfe’s (S63)
revised version of Snyder’s (S64) self-
monitoring scale, which assesses the
ability to modify one’s self-
presentation. Sample items are “Once
I know what a situation calls for, it’s
easy for me to regulate my actions
accordingly,” “I have found that I can
adjust my behavior to meet the
requirements of the situations I find
myself in,” andI have trouble
changing my behavior to suit different
people and different situations”
(reverse coded). Participants
responded to these items on a 6-point
scale where 1 = certainly not, always
false to 6 = certainly, always true.
We conducted a Procrustes Factors
Analysis for all scales, including
prevention focus (cautiousness,
dutifulness), self-regulation strength
(impulse control), need for structure,
and self-monitoring ability. For the
personal need for structure scale, the
normative EFA revealed that, as
expected, the 8 items loaded cleanly
onto two factors: desire for structure
and response to lack of structure.
When the cultural samples were
subjected to procrustes rotation, the
identity coefficient for the desire for
structure factor exceeded the .90
cutoff for all 33 countries, but the
identity coefficient for the response to
lack of structure factor only met the
.90 cutoff for 17 countries. Thus, the
desire for structure factor
demonstrated structural equivalence,
but the response to lack of structure
factor did not. Accordingly, we only
report analyses for the desire for
structure scale. We note, however,
that the response to lack of structure
scale revealed parallel effects in all of
our analyses (all data are available
from the first author). Each of the
remaining scales loaded onto a single
factor in the normative solution.
Procrustes rotations were conducted
separately for each of these scales,
and for each scale the identity
coefficient met the .90 criteria in all
countries. Thus, each of these scales
demonstrated structural equivalence.
Based on the analyses, some items
were dropped due to low loadings in
the normative solution. In the final
solutions the cautiousness scale
included nine items (
= .85), the
dutifulness scale included 10 items (
= .86), the impulse control scale
included five items (
= .73), and the
self-monitoring scale had five items
= .77).
Hierarchical linear modeling
(HLM) analyses indicated that there
were large cultural differences in all
of the psychological adaptations. The
ICC(1) values all exceeded the
recommended cutoffs for ICC (1) of
.06 (cautiousness = .12; dutifulness =
.13; impulse control=.08; desire for
Gelfand et al.
structure=.19; and self-monitoring-
=.10, P’s all <.01).
Sources and Descriptions of
Archival Data
Ecological archival data.
Population density was gathered from
a published source for the year 1500
(S65), from the United Nations for the
year 2000 (S66), and from the World
Bank’s World Development
Indicators for rural areas in the year
2000 (S67). All population density
measures were transformed by a
natural log function to reduce
skewness in distribution. Data on
arable land, food production, food
supply, protein and fat supply, and
food deprivation in year 2002 were
taken from the Food and Agriculture
Organization of the United Nations
(S68). Population pressure, air quality
and natural disaster vulnerability were
obtained from the 2005
Environmental Sustainability Index
Report (S69). Data on percentage of
farmland and access to safe water
were taken from Kurian’s world
ranking (S70). The index of historical
prevalence of pathogens was taken
from Murray and Schaller’s research,
in which they constructed the disease
prevalence index based on early
epidemiological atlases (S71). The
World Health Organization (WHO)
provided data for years of life lost to
communicable disease in 2002 and
prevalence of tuberculosis per
100,000 in 2000 (S72). Mortality rate
for infants from 2000-2005 and
children under 5 in the year of 2000
were also gathered from the United
Nations (S66). The number of threats
from neighboring nations for
integration and annexation, or
territorial threats, from 1918-2001,
was gathered from the International
Crisis Behavior Archives (S73).
Socio-Political Institutions and
Practices. Autocratic polity 2002 data
were ga