Assess a Potential Foreign Market

Jn[. Siudict of Man. i Org.. Vol. Xlll. No. 1-2, pp. 46-74 M.E. Sharpc. Inc.. 1983

NATIONAL CULTURES IN FOUR DIMENSIONS A Research-based Theory of Cultural Differences among Nations

Geert Hofstede (the Netherlands)

This report summarizes a large research project,, in- volving 116,000 questionnaires, about the work-related value patterns of matched samples of industrial employees in 50 countries and 3 regions at 2 points in time. Half of the variance in the countries' mean scores can be ex- plained by four basic dimensions, here labeled power distance, uncertainty avoidance, individualism versus collectivism, and masculinity versus femininity. These dimensions are offered as a framework for developing hypotheses in cross-cultural organization studies. Index scores of the countries on the 4 dimensions correlate significantly with the outcomes of about 40 existing com- parative studies.

The four dimensions considered here relate to very fundamental problems which face any human society, but. to which different societies have found different answers. They are used to explain (1) different ways of structur- ing organizations, (2) differentjnotivations of people within organizations, and (3) different issues people and organizations face within society. On the basis of com- bined scores, the countries studied can be grouped by cul-

Dr. Hofstede is Director of the Institute for Research on Intercultural Cooperation, Arnhem, the Netherlands.

46

National Cultures in Four Dimensions • 47

tural clusters. The data also allow some conclusions about trends in value shifts over time.

The Cultural Relativity of Organization Theories

A certain U.S. business organization has a policy that salary- Increase proposals should be initiated by the employee's direct superior. In its French subsidiary, this policy is interpreted in such away that the superior's superior's superior — three levels above the employee — is the one who initiates salary- increase proposals. Why the difference? Because the French managers and their employees share some basic values about the exercise of authority that differ from the values that p r e - vail among their U.S. colleagues. These basic values are fos- tered in the majority of French and U.S. families, and are rein- forced in French and U.S. schools.

Organization theorists are also products of a national cul- ture: they were reared in families and trained in schools, and they absorbed the values prevailing in one particular society. They usually collect their life experience and their research data in the same society. Organization theories are therefore culturally bounded. The task of cross-cultural organization research is to broaden both the data bases available to r e - searchers/theorists and their awareness of value systems dif- ferent from their own.

If organization theory is to transcend national boundaries, it should go beyond statements such as "In the USA . . . , but in France " In this case we treat names of nations as r e s i - dues of undefined variance. For theories to be truly universal we should attempt to replace names of nations by explicitly defined variables. When we find that societies differ with r e - gard to a particular characteristic, we should try to specify what it is about these societies that causes this difference (Przeworski and Teune, 1970. P. 29).

A Multicountry Data Base (1) . .

Replacing country names by variables that explain differ-

48 Geert Hofstede (the Netherlands)

ences in the structure and functioning of organizations presup- poses a base of multicountry data to start from. It is very difficult to infer such variables from data about only two or three countries. The scai'city of studies covering more than a handful of countries and the almost inevitable loss of depth with a gain in breadth explain the slow development of truly universal organization theories.

For the present study (fully described in Hofstede, 1980), ac- cess was obtained to an existing body of organizational data from no fewer than 67 countries. All but one of the organiza- tions axe subsidiaries of the same large U.S.-based multina- tional Industrial corporation, which I shall call "Hermes," The remaining one is a Yugoslav worker-managed organiza- tion that (among other activities) imports and services Hermes products in Yugoslavia.

The data consist of answers of individual employees to stan- dardized paper-and-pencil attitude and value survey question- ' naires. The same questionnaires were used twice, around 1968 and around 1972. The entire data bank contains the answers on over 116,000 questionnaires, each with about 150 questions. Twenty different language versions were used.

The interest of the present study is in fundamental differ- ences in the way people in different countries perceive and interpret their world. It has therefore looked at only 32 ques- tions from the data bank, ones that (1) were conceptually r e - lated to the respondents' cosmologies (ways of looking at the world) and (2) showed differences among countries that r e - mained relatively stable between the 1968 and the 1972 survey rounds. Specifically, the questions retained for analysis deal with;

— perceptions of the organizational regime, (such as the presence or absence of subordinate consultation by superiors);

— perceptions of the organizational climate (such as feelings of job-induced stress);

— values in terms of the desirable (reactions to general, ideological statements such as "Competition among employees usually does more harm than good");

National Cultures in Four Dimensions 49

— values in terms of the desired (importance personally at- tached to various aspects of a job, such as earnings, coopera- tion, and security).

Treatment of the Data

The Hermes data bank represents a multilevel, multicriteria data base since responses can be analyzed across individuals, across occupations, across countries, between the sexes, amont age groups, and over time (1968 to 1972). This paper focuses on the analysis across countries. For this purpose, for each relevant question a score has been determined for each coun- try. This score was (1) based either on scale means or on agreement percentages, depending on the nature of the question; (2) composed for a constant mix of seven clerical, technical, professional, and managerial occupations, identical for all countries; and (3) averaged for the 1968 and 1972 survey rounds.

Of the 67 countries in the data bank, 27 were originally omitted from the analysis because more than half of the neces- sary occupational data was from fewer than eight respondents and was therefore considered "missing." This corresponded to a minimum number of about 50 respondents per country for each survey round. Thus, the bulk of the data summarized in this paper is based on 40 countries (Hofstede, 1980). At a later stage (Hofstede, 1982), countries were added for which sufficient data were available for at least two occupational groups; this increased the data base to 50 countries. Finally, data were also included from three multicountry regions (Arab- speaking countries. East Africa, and West Africa) for which the number of available responses from the individual countries was insufficient.

For the original set of 40 countries, the relationships among the country scores on the 32 questions were studied. This rep- resents an ecological, not an individual, analysis. The number of cases used is 40 (countries), not 116,000 (individuals). Eco- logical correlations among variables are mathematically differ- ent from individual correlations (Robinson, 1950), and should be interpreted differently.

50 Geert Hofstede (the Netherlands)

Questionnaire items were grouped according to ecological dimensions, based upon (1) theoretical relevance, and (2) statis- tical relationships. Four such dimensions were identified. There were labeled power distance, uncertainty avoidance, in- dividualism versus collectivism, and masculinity versus femi- ninity. The positions of each of the 40 countries on each of these four dimensions were expressed by an index. A factor analysis of the country scores for 32 questionnaire items and 40countriesshowed that 3 factors explained 49 percent of the total variance: one factor combined high power distance and low in- dividualism, one corresponded to uncertainty avoidance, and one, to masculinity. No other meaningful factors were found. The two dimensions of power distance and individualism, al- though negatively correlated (r = 0.67 across 40 countries), have been maintained as separate dimensions for conceptual reasons.

When, later, ten more countries and three regions were added, the index scores for these supplementary units were calculated according to the formulas derived from the first and larger 40 units. Adding these new units did not substan- tially change the structure of the dimensions (Hofstede, 1983).

Power Distance

A power distance index (PDI) has been composed of the coun- try scores on the following three questionnaire items, which are intercbrrelated with coefficients between 0.54 and 0.67 {see also Hofstede, 1977, 1979a).

(1) the percentage of subordinates who perceive that their boss makes his decisions in an autocratic or paternalistic (per- suasive) way;

(2) subordinates' perceptions that employees in general (their colleagues) are afraid to disagree with superiors (mean score on a five-point scale from 1 = very frequently to 5 = very seldom, multiplied by 25 to make it comparable with the per- centage scores for questions 1 and 3);

(3) the percentage of subordinates who do not prefer a boss

National Cultures in Four Dimensions 51

who makes his decisions in a consultative way, but who prefer a boss who decides either autocratically or paternalistically or, on the other hand, who does not decide himself, but goes along with the majority of his subordinates (for the rationale of combining the latter categories, see below).

The index is computed-according to the formula: PDI = 135 + a - b -I- c, which brings it into a range between 11 and 104: its theoretical range is from -90 (no power distance to + 210 (su- preme power distance). The index values for the 50 countries and 3 regions can be read in Table 1.

The term power distance has been inspired by the work of Mulder (1976, 1977). The country-level correlation of the pre- ferred type of decision making in the superior with the percep- tion of the behavior of both superior and colleagues reveals a fundamental fact about power distance in a hierarchy, namely, that a society's way of dealing with power relationships is e s - tablished through the values of superiors as well as of subordi- nates. It appears in the Hermes data that subordinates in a system in which superiors maintain considerable power dis- tance respond by preferring such superiors (dependent reac- tion) or by going to the other extreme and preferring superiors who do not decide at all, but govern by a majority vote of their subordinates. This latter type of decision making is rather unlikely to be practiced in complex work organizations: it would be feasible only if departments were autonomous and indepen- dent of other departments, whereas in fact modern work or- ganizations are complex interdependent systems. Therefore, the preference for a majoricy-vote type of decision making in a complex work organization is unrealistic; I interpret it as a collective counterdependent reaction to a situation of great power distance. So we see that in systems in which superiors maintain a great power distance, subordinates tend to polarize toward dependence or counterdependence. On the other hand, where superiors maintain less power distance, subordinates tend to prefer the consultative decision style, which can be in- terpreted as a compromise solution, an interdependence of superior and subordinate.

52 Geert Hofstede (the Netherlands)

Table i

Index Values and Rank of 50 Countries and 3 Regions on Four Cultural Dimensions

Country

Argentina Australia Austria Belgium Brazil Canada Chile Colanbia * Co6ta Rica Denmark • Equador Finland France Genrany (F.R.) Great Britain Greece * Guatetmla Hong Kcng • Indonesia India Iran Ireland I s r a e l I t a l y * Janaica Japan • Korea (S.1 " Malaysia Mexico Netherlands Norway N€V Zealand Pakistan * Panana Peru Philippines Portugal Scuth Africa * Salvador Singapore Spain Svedsn Switzerland Taiwan Thailand Turtcey • Uruguay U.S.A. Venezuela Yugcslavia Regions: ' East Africa * West Africa * Arab Ctrs.

Atiirev- i a t i c n

ARG AUL AOT BEL BRA CAN CHL O X COS DEN

FIN FRA GEK CZR OiE GOA HOK IDO IND IRA IRE ISR ITA JAM J A P KCR mL MEX NCT tCR NZL PAK PAN P I S PHI POR SAF SAL SIN SPA SWE SWI

m i THA TUR URU USA VEN VTJ3

EAF

ARA

Power Distance

lnde.v {PDD

49 36 11 65 69 39 63 67 35 18 78 33 68 35 35 60 95 68 78 77 58 28 13 50 45 54 60

104 81 38 31 22 55 95 64 94

63 49 66 74 57 31 34 58 64 66 61 40 81 7 6

64 77 80

Rank

19-19 13

1 33 39 15

29-30 36

10-12 3

43-44 8

37-38 10-12 10-12 26-27 48-49 37-38 43-44

42 24-25

5 2

20 17 21

26-27 50

45-46 14

6- 7 4

22 48-49 31-32

47 29-30 18-19 34-35

40 23

6- 7 9

24-25 31-32 34-35

28 16

45-46 41

(31-321 ( 421 (44-451

Uncertainty Avoidance

Index {UAl}

86 51 70 94 76 48 8 6 80 86 23 67 59 86 65 35

112 101

29 48 40 59 35 81 75 13 92 85 36 82 53 50 49 70 86 87 44

104 49 94

o 86 29 58 69 64 85

100 46 76 88

52 54 68

Rank

36-41 17

26-27 45-46 29-30 12-13 36-41

31 36-41

3 24

20-21 36-41

23 6- 7

50 48

4- 5 12-13

9 20-21 6- 7

32 28

2 44

34-35 8

33 18 16

14-15 26-27 36-41

42 10 49

14-15 45-46

1 36-41 4- 5

19 25 22

34-35 47 U

29-30 43

{17-181 (18-191 (24-251

Indiv a l i s n

Inde.v (IDV)

46 90 55 75 38 80 23 13 15 74

8 63 71 67 39 35

6 25 14 48 41 70 54 76 39 46 1 8 26 30 80 69 79 14 11 16 32 27 65 19 20 51 71 68 17 20 37 36 91 12 27

27 20 38

Rani;

28-29 49 33 43 25

46-47 15

5 8

42 2

34 40-41

36 48 22

1 16

6- 7 30 27 39 32 44 26

26-29 11 17 20

46-47 38 45

6- 7 3 9

21 18-19

35 12

13-14 31

40-41 37 10

13-14 24 23 50

4 18-19

(18-191 (13-141 ( 251

Masculinity

Index (MAS)

56 61 79 54 49 52 28 64 21 16 63 26 43 66 66 57 37 57 46 56 43 68 47 70 68 95 39 50 69 14

8 58 50 44 42 64 31 63 40 48 42

5 70 45 34 45 38 62 73 21

41 46 53

Rank

30-31 35 49 29 25 28

8 39-40 5- 6

4 37-38

7 17-18 41-42 41-42 32-33

11 32-33

22 30-31 17-18 43-44

23 46-47 43-44

50 13

26-27 45

3 2

34 26-27

19 15-16 39-40

9 37-38

14 24

15-16 1

46-47 20-21

10 20-21

12 36 48

5- 6

114-15) ( 221 128-291

EasecT c-.i data added l a t e r

National Cultures in Four Dimensions 53

Uncertainty Avoidance

An uncertainty avoidance index (UAI) has been composed of the Hermes country scores on the following three questionnaire items, which are intercorrelated with coefficients between 0.40 and 0.59:

1. How often do you feel nervous or tense at work? (mean score on a 5-point scale from 1 = always, to 5 = never).

2. Company rules should not be broken, even when the em- ployee thinks it is in the company's best interest (mean score on a 5-point scale from 1 = strongly agree, to 5 = strongly d i s - agree).

3. How long do you think you will continue to work for this company? (Percent answering [a] two years at the most, or [b] from two to five years. This is equal to 100 minus the per- cent planning to stay more than 5 years.)

The computation formula has been chosen to obtain equal contributions from all three questions to the variance in UAI, as follows: UAI = 3 0 0 - 4 0 d - 3 0 e - f . The values for UAI range from 8 to 112. The theoretical range is from -150 (no uncertainty avoidance) to + 230 (extreme uncertainty avoidance). These Index values can also be read in Table 1.

The term uncertainty avoidance has been inspired by Cyert and March;(1964). The ecological dimension revealed by the three questions has been associated with "uncertainty avoid- ance" because agreement with question 2 (rules should not be broken) and a low percentage for answer 1 + 2 on question 3 (few people want to leave the company) indicate two different ways of avoiding uncertainties in life. Feelings of uncertainty create anxiety; scores toward the "always" side on question 1 indicate greater anxiety.

High scores on the UAI thus mean a higher mean anxiety level among respondents in a country and avoidance of attitudes and behaviors that could increase this anxiety (doing away with rules, considering leaving one's employer).

Individualism-Collectivism and Masculinity-Femininity

The other 2 ecological indices are derived from mean

54 • G e e r t Hofstede {the Netherlands)

country s c o r e s on 14 questions dealing with "values in t e r m s of the desired": the importance attached by respondents to the aspects of jobs indicated a s challenge, desirable (living) a r e a , e a r n i n g s , cooperation, training, benefits, recognition, physical conditions, freedom, employment security, advancement, ( r e l a - tion with) manager, use of skills, and personal time (time for personal life).

Answers to these questions were scaled according to five points ranging from "of utmost importance" to "of very little or no importance." The scores for groups of respondents were standardized a c r o s s the 14 goals (that i s , given the same overall mean and standard deviation) so that tendencies in a group to score everything as more or as l e s s important were eliminated (these tendencies were investigated separately; see Hofstede, 1980. Pp. 77 ff. and 224).

The standardized country s c o r e s for the 40 countries on the 14 goals were factor analyzed and showed two clear factors explaining, respectively, 24 and 22 percent of the variance (Hofstede, 1980. P . 241). The factor s c o r e s for the countries on these two factors were transformed into a country " i n d i - vidualism" and a country "masculinity'" s c o r e . For "individu- alism" the formula used was IDV = 50 + 25 x (factor score); for "masculinity," MAS = 50 - 20 x (factor s c o r e ) , yielding an IDV of 6 - 9 1 and a MAS of 5-95 (see Table 1).

Individualism (IDV) Indicates the relative importance in the country of the job aspects personal t i m e , freedom, and c h a l - lenge and the relative unimportance of training, of use of s k i l l s , of physical conditions, and of benefits. It thus s t r e s s e s goals in which the individual is an active ageni v e r s u s those in which he or she is dependent on the organization (being trained, skills being used, working conditions, and benefits being provided).

W h e r e a s the power distance index indicated dependence on the s u p e r i o r , the individualism index indicates (non-) dependence on the organization. As mentioned e a r l i e r , the two a r e nega- tively correlated (r = -0.67). The combination of a low PDI and a low IDV does aot occur (except marginally for Costa Rica); but the combination of a high PDI and a high IDV occurs for the

National Cultures in Four Dimensions S5

Latin European countries France, Belgium, and Italy and mar- ginally for Spain and South Africa. In order to sho_w cle.arly the unique values pattern of Hermes employees in these Latin Euro- pean countries, I have maintained the PDI and IDV as separate indices and not coUapsed them into a single index. The Latin European values pattern means that people in Hermes subsidi- aries in these countries have a need for dependence (or counter- dependence) on hierarchical superiors but, at the same time, s t r e s s their personai independence from the organization to which they belong: they are dependent individualists.

Masculinity (MAS) indicates the relative importance in the country of the job aspects earnings, recognition, advancement, and challenge and the relative unimportance of (relation with) manager, cooperation, desirable (living) area, and employment security. These tend to be also the job aspects on which, v/ith- in countries. Hermes men, in the same jobs, score significantly differently from Hermes women. This is in line with the domi- nant pattern of sex roles found to exist in nearly all societies, even nonliterate ones (Barry, Bacon, and Child, 1957): boys are socialized toward assertiveness and seU-reliance, and girls, toward nurturance and responsibility.

The Hermes data show that not only do men and women in the same jobs emphasize different j.ob aspects but that coun- tries also differ along these same lines: in some countries all respondents (both men and women) emphasize job aspects usually associated with the male role; in others, all emphasize job aspects usually associated with the female role. Moreover, it appears that in more "masculine" countries, the gap between. the values for the men and for the women in Hermes is wider, whereas in the most "feminine" countries this gap is reduced to zero (Hofstede, 1980. P. 282).

Relevance of the Hermes Dimensions for National Cultures in General

Employees in Hermes subsidiaries are, of course, an ex- tremely narrow and specific sample of their countries' popu-

56 Geert Hofstede (the Netherlands)

lations. They belong to the middle class of their society rather than to the upper, working, or peasant class. They have doubt- less all been socialized, to some extent, to the international perspective a corporation like Hermes maintains.

Valid comparisons among countries in the study of cultural differences can use two strategies: a broad sample strategy, or a narrow sample strategy. Broad samples should be r e p r e - sentative of entire populations; they are typically used in pub- lic-opinion research and consumer market research. Narrow sarr^les use respondents who are very well matched: they should be similar in many aspects, except nationality, anU be- long to functionally equivalent categories in each country.

Most cross-cultural studies use narrow samples, such as students or industrial managers. The Hermes subsidiary r e - spondents are also such narrow, but well-matched, samples: they share the same company superstructure and policies; they are selected to belong to the same occupational categories, so they do very much the same kind of work; they are of the same education level and vary only marginally in age and sex composition; they differ systematically only in nationality. Because the data analyzed are differences between Hermes employees in one country and another, they paradoxically tell us nothing about the Hermes corporate culture, because this - is shared by all employees. Systematic and stable differences among the Hermes respondents irom different countries caji be explained only by country culture; in fact, the differences within Hermes because of the shared corporate structure should be a conservative estimate of differences to be found in organizations outside Hermes.

The validity of the Hermes indices (PDI, UAI, IDV, MAS) for characterizing countrywide culture patterns can be checked with other data. If the dimensions they measure are meaning- ful for national cultures at large, not just within Hermes, the indices should correlate with measures found in other narrow samples, in broad samples, and with country-level indicators. The last represent characteristics of countries as total sys- tems not measured by aggregating data collected from indi-

National Cultures in Four Dimensions 57

viduals, but measured for the country as a whole, such as its per capita gross national product (GNP) or its traffic accident death rate.

In an extensive search of theoretically relevant other data, I found 13 comparative survey studies, covering between 5 and 19 countries at a time, whose results are significantly c o r r e - lated across the particular countries with one or more of the Hermes indices (Hofstede, 1980. Pp. 326-27). I also found 31 country-level indicators, available for between 5 and 40 coun- t r i e s , that were significantly correlated with at least one of the dimensions (Ibid. Pp. 328-31). Few of these studies and indicators had ever been related to each other before. Often the sets of countries for which data are available overlap only very partially from one study to another. The four Hermes indices provide a framework for fitting these disparate studies together; they drop into place like pieces in a large jigsaw puzzle.

Here are a few examples of the correlations found:

With the Power Distance Index (PDI)

— The country's political system since 1950: the presence or absence of periods with autocratic or'oligarchic govern- ments, cross-tabulated against the PDI above or below the mean, yields a chi-square of 16.9*** with one degree of free- dom. (2)

— Results obtained with translated versions of Gordon's (1976. P. 55) survey of interpersonal values among students from 17 countries: r = 0.80*** for conformity', r = —0.79*** for independence, and r = -0.70*** for "support" (expecting to be treated with understanding).

With the Uncertainty Avoidance Index (UAI):'

— A "neuroticism" factor found in the data of 12 medical and related indices for 18 developed countries by Lynn and Hampson (1975. P. 237): Spearman rank correlation coefficient rho = 0.73***.

58 Geert Hofstede (the Netherlands)

— Several questions in a public-opinion study on Images of the World in the Year 2000, by Ornauer et al. (1976. Pp. 674- 95), are comparable with Hermes data across nine countries — for example, the statement "To compromise with our opponents is dangerous because it usually leads to the betrayal of our side": r = 0.90***.

With the Individualism Index (IDV):

— The coimtry's income (1970 per capita GNP). Across the 40 countries, the correlation between the individualism index and income is r = 0.82***.

— Data from the International Research Group on Manage- ment (IRGOM) data bank obtained with Bass's exercise "life goals" from managers in 12 countries (Bass and Burger, 1979). After having performed an ecological factor analysis, I found a factor opposing "pleasure," "security," and "affection" to "expertness," "prestige," and "duty" that correlates with an IDV with rho = 0.76***.

With the Masculinity Index (MAS):

— The percentage of GNP spent on government-sponsored development aid to third-world countries for 15 wealtny coun- tries correlates (negatively) with r = -0.81***.

— Data from the IRGOM data bank (see above) on "life goals" from managers in 12 countries show a second factor opposing "leadership" and "independence" to "service"; this correlates with a MAS with rho = 0.84***.-

Combining the connotations of the four dimensions found within Hermes with the connotations of their correlates in other studies, I have made up the integrated lists of connota- tions shown in Tables 2 through 5. For each dimension a table shows the connotations associated with the extreme positive and negative poles. Most countries, of course, are somewhere between these poles.

To avoid misunderstanding, I want to stress again that the

National Cultures in Four Dimensions 59

four dimensions are ecologically derived: they apply to coun- tries as social systems, not to individuals within those coun- tries. For example. Table 3 shows that societies with a high uncertainty avoidance index are characterized by "more show- ing of emotions" and a "need for written rules and regulations." This means that the two features tend to be associated at the country level, but not that individuals who show their emotions more also tend to need more written rules: the reverse may even be the case. Psychologists, especially, often have diffi- culties recognizing that characteristics associated with each other at the society level need not be associated at the indi- vidual level, and vice versa. Societies are not "king-size in- dividuals": eco-logic is not the same as psycho-logic.

Fundamental Problems of Societies

The previous section argues that the four dimensions power distance, uncertainty avoidance, individualism—collectivism, and masculinity—femininity represent universal categories for characterizing national societies. They transcend the narrow borders of the Hermes corporation. If this is true, they should relate to underlying, fundamental problems of societies, to which every society should find its own answers.

From the cultural anthropological literature (for example, Mead, 1962. Pp. 102-107), we know that the nonliterate soci- eties that have been the preferred object of study of anthro- pologists differ remarkably in their answers to such funda- mental problems. Some societies have kings and hierarchies; others are ruled by consensus among eq,uals. Some societies have clearly defined and different sex roles; others have side- by-side collaboration of women and men on the same tasks.

There is no. reason to assume that such differences among societies have entirely disappeared from present-day nation- states; in fact, it would be very surprising if they had. How- ever, because of the complexity of nation-states, the differences have become less clearly visible to any single observer. So- ciologically oriented anthropologists such as Inkeles and

60 Geert Hofstede (the Netherlands!

Table 2

Connotations ol the Power Distance Dimension

Low PDI Inequality in society should be mini-

mized.

AU should be interdependent.

Hierarchy means an equality of roles, established for convenience.

Subordinates are people like me.

Superiors are people like me.

The use of power should be legiti- mate, and is subject to the judg- ment between good and evil.

All should have equal rights.

Powerful people should try to look less powerful than they are.

Stress on reward and legitimate and expert power. (1)

The system is to blame for things that go wrong.

The way to change a social system is by redistributing power.

People at both high and low power levels feel less threatened and more prepared to trust people.

There is latent harmony between the powerful and the powerless.

Cooperation among the powerless can be based on solidarity.

High PDI There should be an order of in-

equality in this world in which everybody has his rightful place: high and low are protected by this order.

A few should be independent; most should be dependent.

Hierarchy means existential inequality.

Superiors consider subordinates as different from themselves.

Subordinates consider superiors as different from themselves.

Power is a basic fact of society and antedates good or evil, its legitimacy is irrelevant.

Power-holders are entitled to privileges.

Powerful people should try to look as powerful as possible.

Stress on coercive and referent power. (1)

The underdog is to blame ior things that go wrong.

The way to change a social system is by dethroning those in power.

Other people are a potential threat to one's power and can rarely be trusted.

There is latent conflict between the powerful and the powerless.

Cooperation among the powerless is difficult to achieve because little faith in people is the norm.

1) French and Raven (1959).

National Cultures In Four Dimensions

Table 3

Connotations of the Uncertainty Avoidance Dimension

Low TJAI High UAI The uncertainty inherent in life is

more easily accepted and each day is taken as it comes.

E a s e , lower s t r e s s Time is free. Hard work is not a virtue per se. Weaker superegos Aggressive behavior is frowned

upon. Less showing of emotions Conflict and competition can be con-

tained on the level of fair play and used constructively.

More acceptance of dissent Deviance not felt as threatening;

greater tolerance Less nationalism More positive attitude toward

younger people Less conservatism More willingness to take risks

in life Achievement determined in terms

of recognition Relativism, empiricism

There should be as few rules as possible.

If rules cannot be kept, we should change them.

Belief in generalists and common sense

The authorities are there to serve the citizens.

The uncertainty inherent in life is felt as a continuous threat that must be fought.

Higher anxiety and s t r e s s Time is money. Inner urge to work hard Strong superegos Aggressive behavior of self and

others is accepted. More showing of emotions Conflict and competition can un-

leash aggression and should therefore be avoided.

Strong need for consensus Deviant persons and ideas are

dangerous; intolerance Nationalism Younger people are suspect

Conservatism; law and order Concern with security in life

Achievement defined in terms of security

Search for ultimate, absolute truths and values

Need for written rules and regu- lations

If rules cannot be kept, we are sinners and should repent.

Belief in experts and their knowledge

Ordinary citizens are incompe- tent compared with the author- ities.

62 Geert Hofstede (the Netherlands)

Table 4

Connotations of the Individualism-Collectivism Dimension

Low rov In societ}', people are born into ex-

tended families or clans that pro- tect them in exchange for loyalty'.

"We" consciousness Collectivity-orientation (1) Identity is based in the social system. Emotional dependence of individual

on organizations and institutions

Emphasis on belonging to organiza- tion; membership ideal

Private life is invaded by organiza- tions and clans to which one belongs; opinions are predetermined.

Expertise, order, duty, security p r b - vided by organization or clan

Friendships predetermined by stable social relationships, but need for prestige within these relationships

Belief in group decisions Value standards differ for in-groups

and out-groups; particularism (1) "Jen" philosophy of man (2)

Gemeinschaft (community-based) social order (3)

Involvement of individuals with organizations primarily moral (4)

High

1) Parsons and Shils (1951) 2) Hsu (1971) 3) Tdnnies {1887) 4) Etzioni (1975)

In society, everybody is supposed to take care of him/herself and his/her immediate family.

"I" consciousness Self-orientation {I) Identity is based in the individual. Emotional independence of indi-

vidual from organizations or institutions

Emphasis on individual initiative and achievement; leadership ideal

Everybody has a right to a p r i - vate life and opinion.

Autonomy, variety, pleasure, in- dividual financial security

Need for specific friendships

Belief in individual decisions Value standards should apply to

all: universalism (_1) "Personality" philosophy of

man (2) Gesellschaft (society-based)

social order [3] Involvement of individuals with

organizations primary calcu- lative (4)

National Cultures in Four Dimensions

Table 5

Connotations of the Masculinity-Femininity Dimension

Low MAS

People orientation

Quality of life and environment are important.

Work to live.

Service ideal

Interdependence ideal

Intuition

Sympathy for the imfortunate

Leveling: Don't try to be better than others.

Small and slow are beautiful.

Men need not be assertive, and can also assume nurturing roles.

Sex roles in society* should be fluid.

Differences in sex roles should not mean differences in power.

Unisex and androg>'ny ideal

H igh MAS

Money and things orientation

Performance and growth are im- portant.

Live to work.

Achievement ideal

Independence ideal

Decisiveness

Sympathy for the successful achiever

Excelling: Try to be the best.

Big and fast are beautiful.

Men should behave assertively, and women should be nurturing.

Sex roles in society should be clearly differentiated.

Men should dominate in all set- tings.

Machismo (ostentative manliness) ideal

64 Geert Hofstede (the Netherlands)

Levinson (1969. Pp. 447 ff.) have predicted categories for classifying the "national character" of nation-states. The four dimensions found in the Hermes data represent such categories, and they fit Inkeles and Levinson's predictions remarkably well (Hofstede, 1980. P. 313).

The underlying, fundamental problems of societies to which the four dimensions apply (compare Tables 2 through 5) in- clude:

Power distance: the problem of human inequality and the translation of biological differences in strength and talents into social differences in power and wealth.

Uncertainty avoidance: the problem of life and death asso- ciated with the one-way arrow of time, the inescapable uncer- tainty about tomorrow, and the ways in which societies never- theless try to enable their members to sleep in peace.

Individualism: the problem of the relationship of the indi- vidual to his or her fellows, from tightly to loosely Integrated primary groups. This dimension, as we see in Table 4, is r e - lated to some classical dichotomies in sociology: Tonnies's (1887) Gemeinschaft (low IDV) versus Gesellschaft (high IDV) and Etzioni's "moral involvement" versus "calculatlve involve- ment" in organizations. There are indications (Hofstede, 1980. Pp. 224-39) that it also relates to Merton's (1968. P. 447) "local" versus "cosmopolitan" mentality. All, obviously, are associated at the society level, not necessarily at the indi- vidual level.

Masculinity: the problem of the division of mankind into two sexes, and what represents the appropriate role for men (who tend to make their concept of their own role a model for soci- ety as a whole).

Implications for Organizations

Organizations serve two main functions: distribution of power, and control of uncertainty. Organizing is a symbolic activity: it consists of the manipulation of symbols — e.g., uniforms, orders, rules, forms, rituals, and policies — that

National Cultures in Four Dimensions 65

have meaning only to the initiated. It is not surprising, there- fore, that the functioning of organizations in a country and the" way of thinking about organizations in that country are related to the country's position on the power distance and uncertainty avoidance scales.

Across the 40 countries surveyed, the PDI and UAI are weakly correlated (r = 0.28*). We nevertheless find countries . in all four quadrants of a PDI x UAI plot. In interpreting the meaning of the countries' positions on the indices, I use con- cepts from two sources. One is the "Aston" approach to the study of organizations (Pugh, 1976; Pugh and Hickson, 1976). The Aston researchers found empirically that different organi- zations within one country (Great Britain) varied mainly ac- cording to two dimensions: "concentration of authority" and "structuring of activities" (not counting a third and weaker di- mension). It is obvious that the first is conceptually related to the PDI, and the second, to the UAI. This suggests that the Aston typology of different organizations in one country applies, mutatis mutandis, also to similar organizations in different countries. The second source of concepts is an unpublished study by a former colleague, O. J. Stevens (3), about the im- plicit models for well-functioning organizations that he found among British, German, and French management students at the INSEAD school in Fontainebleau, France.

Hermes subsidiaries with a low PDI and a low UAI (mainly Anglo and Nordic countries) are characterized by smaller power distances and weak uncertainty-avoidance tendencies. The Aston typology refers to organizations in the low concen- tration of authority, low structuring of activities quadrant as "implicitly structured." Stevens finds that British INSEAD students see the effective organization as a "village market": equality of partners and few rules. In these countries we can expect relative sympathy for decentralized and flexible struc- tures, such as participative management and matrix organiza- tion, and theories that defend the effectiveness of such struc- tures.

Hermes subsidiaries with a low PDI and a high UAI (mainly

66 Geert Hofstede (the Netherlands)

German-speaking) combine smaller power distances with strong uncertainty avoidance. The corresponding quadrant in the As- ton typology shows "work-flow bureaucracies," and Stevens finds for the German INSEAD students a model of the effective organization as a "well-oiled machine." For these countries we can e ^ e c t relative sympathy for decentralized structures, which will, however, be somewhat rigid; there is a greater need for creating rules and living by them. Job content, pro- motion criteria, and decision competencies will tend to be strictly codified. Rules will rarely be broken, because they have been internalized in people's superegos. A typical theory from one of these countries is Weber's theory of bureaucracy as an impersonal system (Weber, 1970 [1921], P. 196).

Hermes subsidiaries with a high PDI and a high UAl (all Latin countries, European as well as Latin American, and some Asian countries) show greater power distances plus strong uncertainty avoidance tendencies. The Aston studies place in this quadrant the "full bureaucracies," and Stevens finds that French INSEAD students model the ideal organiza- tion as a "pyramid" of people. The tendency toward centrali- zation is strong. The combination of strong uncertainty avoid- ance with great power distances means that in these coimtries one will look to powerful people for resolving uncertainties. There is a need for formal rules, too, but powerful people can break them — and so will less powerful people if they can get away with it when there are no powerful people around. Rules are personal, not impersonal. Typical theories from these countries are Pareto's (1976 [1896]) thoughts about elites and Crozier's (1964) analysis of the French bureaucratic phe- nomenon.

Hermes subsidiaries with a high PDI and a low UAl (most Asian and all African countries) combine great power distance; with weak uncertainty avoidance. The Aston studies place in this quadrant "personnel bureaucracies" that have rules for status relationships among people, but not to the same extent for daily work activities. Stevens's study at INSEAD did not cover Indian students, but discussions with Indian colleagues

National Cultures in Four Dimensions ' 67

have led me to fill in the "family" as an implicit model of the organization among them, with a father-type manager of un- questioned authority, but no deep-seated need for working ac- cording to formal rules.

Implications for Motivation

McClelland (1961. P. 461) has published scores by country for need for achievement (nAch)> need for affiliation (n^ff), and need for power (npow)> These scores were based on a content analysis of children's stories from two periods in time: from about 1925 (for 25 countries) and from about 1950 (for 41 coun- tries). McClelland's scores for nAch around 1925 appear to correlate strongly with the combination of low uncertainty avoidance and high masculinity in the Hermes indices (across 22 countries represented in both sets; multiple correlation coefficient R = 0.74***). It is remarkable that McClelland's 1925, but not his 1950, data correlate with the (1970) Hermes scores; I explain this by the assumption that the 1925 stories, collected in the 1950s, were largely traditional and therefore more representative of fundamental cultural themes than the 1950 stories, which were more affected by conscious selection by contemporary educators, who at that time were more sensi- tive to nontraditional fashions.

The correlation of McClelland's country scores with Hermes country scores shows that need for achievment as a traditional theme as defined by McClelland corresponds to a willingness to take risks (low uncertainty'' avoidance) plus a masculine de- sire for visible success (high masculinity). The countries showing this combination are all Anglo-Saxon countries (Aus- tralia, Canada, Great Britain. Ireland, New Zealand. South Africa, the United States), plus a number of their former colo- nies (Hong Kong, India, Jamaica, Malaysia, the Philippines, Singapore). A remarkable fact is that nearly all these coun- tries speak English: the word achievement: is hardly translat- able into any other language.

The discovery that the "achievement motivation" pattern

68 Geert Hofstede (the Netherlands)

corresponds to the dominant value pat t erns of one p a r t i c u l a r group oTcouritrie's", including the United States, should lead to skepticism about the universal vaiidit)' of U.S.-made motiva- tion theories in general. 1 am thinking, in p a r t i c u l a r , of M a s - low's (1970) theory of human needs. In Maslow's need h i e r - archy, self-actualization and esteem needs are placed above social (belongingness) needs, and the latter above security needs. We can identify different combinations of the UAI and MAS with a predominance of different needs. We have already seen that a low UAI and a high MAS stand for achievement needs: weak uncertainty avoidance indicates a predominance of risk-taking over security, and masculinity implies a p r e - dominance of a s s e r t i v e n e s s (self-actualization, esteem) over nurturance (belongingness).

Maslow's hierarchy is thus a reflection of the value pattern of low-UAI, high-MAS countries. With a low UAI and a low MAS (Scandinavia and the Netherlands), we should find a r e l a - tive predominance of belongingness over self-actualization. We might think, for example, of the Swedish approach to hu- manization of work (group centered) v e r s u s the U.S. approach (individual job-enrichment centered). With a high UAI we should find a relative predominance of security over r i s k - taking; if the MAS is low, this relative predominance is c o m - bined with belongingness (e.g., Yugoslav w o r k e r s ' s e l f - m a n - agement); if the MAS i s high, it is combined with a s s e r t i v e - n e s s (e.g., Japanese performance motivation combined with, life-time employment).

Country C l u s t e r s

On the b a s i s oi each country's s c o r e s on the four d i m e n - sions, the units have been clustered, using W a r d ' s grouping method (according to a program by F o r s t and Vogel, 1977). The resulting "dendrogram" is shown in the figure. It should be read from left to right: it indicates what percent of e r r o r should be tolerated to bring the s c o r e s of two countries t o - gether. A split into two large c l u s t e r s cuts off Guatemala

National Cultures in Four Dimensions 69

90 100

ERROR SUM OF SQUARES IN PERCENT OF TOTAL

Results ot a cluster analysis for 53 countries and regions, using the s c o r e s on all four dimensions as variables. The program used was developed by F p r s t and Vopel (1977). The analysis was performed at Kiel University. Germany, thanks to the kind assistance of Professor Dr. Klaus Brockhoff.

70 Geert Hofstede (the Netherlands'

through Jamaica from Austria through Finland. A further breakdo\ra shows, lor example, a Belgium-France cluster (No. 6), three Asian clusters (Nos. 3, 8, and 9), an Anglo-Saxoi cluster (No. 11), and a Nordic-Dutch cluster (No. 13). Japan (No. 4) is the most different from any other country. It should be realized that the computer that produced this figure knows neither geography nor history. We see that, nevertheless, in most cases, geographically or historically close countries cluster together. This illustrates to what extent the mean an- swers of Hermes employees on certain paper-and-pencil ques tions do reflect basic cultural patterns.

Trends over Time

The fact that the Hermes data were measured twice, around 1968 and around 1972. allows some conclusions about world- wide shifts on the four dimensions during this period. The dimension showing the largest universal shift is individualism. An increase in individualism was found in all countries except Pakistan. The data suggest that the increase in individualism followed an increase in wealth, rather than the other way around. On the dimension of individualism, there was some reduction in the distance between extreme countries, so that we can speak of a certain convergence over time.

On the dimension of masculinity-femininity, there was, on the average, a shift toward the masculine side; but this hy no means aifected all countries. In general, the trend was for masculine countries to become more masculine and for femi- nine countries to become more feminine, so that there was a divergence rather than a convergence on this dimension over time.

oh the dimension of power distance, thejcrend was more c-cfinplex. In nearly all the countries. Hermes employees' preference for a more consultative or democratic manager increased: but only in the countries in which power distances .were already relatively low was this preference matched by a corresponding shift in perceived actual behavior of the man-

National Cultures in Four Dimensions 71

a g e r s . In countries with great power distances, there was a marked increase in employees' perceived fear of disagreeing with superiors. All in all, this dimension, too, showed diver- gence among countries rather than convergence.

On the dimension of uncertainty avoidance, only the question of s t r e s s (feeling nervous or tense at work) showed a distinct worldwide trend: an increase in the vast majority of countries. However, here again the tendency was toward divergence rath- e r than convergence of countries at the extremes.

It is possible (Hofstede, 1980. Chap. 8) to relate these vari- ous trends to the age groups of the respondents; it can be shown that the shifts affect all except sometimes the oldest (over 40 or 50) age groups and that there is no evidence for a generation effect in values that would move along like a wave over time.

Four y e a r s , of course, is a very short time for measuring worldwide value shifts. In Hofstede (1980. Pp. 367 fl.), evi- dence from other sources is used to speculate on longer-term trends. Thus, there seems to be little doubt that since 1965 there has been a worldwide trend toward increased s t r e s s , anxiety, intolerance, and other uncertainty-avoidance-related attitudes, which may be part of an oscillating movement with a wavelength between 25 and 40 y e a r s . In power distance, the longer-term trend is probably one of decrease, and in individu- alism, very clearly of increase; out for masculinity and femi- ninity, the shifts vary from one country to another.

Notes

1) The research project described In this paper was carried out in 1973-79 at the European Institute for Advanced Studies in Management at Brussels, Belgium, and continued in 1981 at the Institute for Research on Intercultural Cooperation (IRIC), Arnhem, the Netherlands.

2) Significance levels are indicated as follows: *** = 0.001 level; ** = 0.01 level: • = 0.05 level.

3) 0 . J. Stevens (1976) "Negotiation, Arbitration, Organiza-

72 Geert Hofstede (the Netherlands)

tion: Planned Intervention Styles in Three European Countries." Notes for draft paper. Fontainebleau, France: INSEAD.

References

B a r r y , H., Bacon, M. K., and Child, I. L. (1957) "A C r o s s - cultural Survey of Some Sex Differences in Socialization." Journal of Abnormal and Social Psychology, 55, 327-32.

B a s s , B. M., and Burger, P. C. (1979) Assessment of Man- agers: An International Comparison. New York: The F r e e P r e s s .

C r o z i e r , M. (1964) The Bureaucratic Phenomenon. Chicago: University of Chicago P r e s s .

Cyert, R. M., and March, J. G. (1963) A Behavioral Theory of the F i r m . Englewood Cliffs, N. J.: Prentice-Hall.

Etzioni, A. (1975) A Comparative Analysis of Complex O r - ganizations (Rev. ed.). New York: The Free P r e s s .

F o r s t , H. T . , and Vogel, F. (1977) Hierarchisch-agglomera- tive Klassifikation von Merkmalstraegern bzw. Merkma- len. Kiel, FRG: Institut fuer Betriebswirtschaft.

French, J . R. P . , and Raven, B. (1959) "The Bases of Social Power." In D. Cartwright (Ed.), Studies in Social Power.

• Ann Arbor, Mich.: ISR. Pp. 150-67. Gordon, L. V. (1976) Survey of Interpersonal Values (Rev,)

Chicago: Science Research Associates. Hofstede, G. (1977) "Cultural Elements in the Exercise of

Power." In Y. H. Poortinga (Ed.), Basic Problems in ' Cross-cultural Psychology'. Amsterdam: Swets and Z e i t - linger. Pp. 317-28.

Hofstede, G. (1979) "Hierarchical Power Distance in Forty Countries." In C. J. Lammers and D. J. Hickson (Eds.), Organizations Alike and Unlike: International and I n t e r - institutional Studies in the Sociology of Organizations. London: Routledge & Kegan Paul. Pp. 97-119.

Hofstede, G. (1980) Culture's Consequences: International Differences in Work-related Values. Beverly Hills, CalLE.: Sage Publications.

National Cultures in Four Dimensions 73

Hofstede, G. (1983) "Dimensions of National Cultures in Fifty Countries and Three Regions." In J. B . Deregowski, S. Dziurawiec, and R. C. Annis (Eds.), Expiscations in C r o s s - cultural Psychology. L i s s e , the Netherlands: Swets and Zeitlinger.

H s u , F . L. K. (1971) "Psychosocial Homeostasis and Jen: Conceptual Tools for Advancing Psychological Anthro- pology." American Anthropologist, 72, 23—44.

I n k e l e s , A., and L e v i n s o n , D . J . (1969) "National Character: The Study of Modal Personality and Sociocultural S y s - t e m s . " In G. Llndzey and E. Aronson (Eds.), The Hand- book of Social Psychology, Vol. 4. Reading, Mass.: Addison-Wesley.

Lynn, R., and Hampson, S. L. (1975) "National Differences in E x t r a v e r s i o n and Neuroticism." British Journal of Social and Clinical Psychology, ]A, 223-40.

Maslow, A. H. (1970) Motivation and Personality (2nd ed.). New York:- Harper & Row.

McClelland, D. C. (1961) The Achieving Society. Princeton, N. J.: Van Nostrand.

Mead, M. (1962) Male and Female: A Study of the Sexes in a Changing World. Harmondsworth: Penguin.

M e r t o n , R. K. (1968) Social Theory and Social Structure (3rd ed.). New York: The F r e e P r e s s .

M u l d e r , M. (1976) "Reduction of Power Differences in P r a c - tice: The Power-Distance Reduction Theory and Its Ap- plications." In G. Hofstede and M. S. Kassem (Eds.), European Contributions to Organization Theory. Assen, the Netherlands: Van Gorcum. Pp. 7 9 - 9 4 .

Mulder, M. (1977) The Daily Power Game. Leiden: Martinus Nijhoff.

O r n a u e r , H., Wiberg, H., Sicinski, A., and Galtung, J . (Eds.) (1976) Images of the World in the Year 2000: A C o m p a r a -

' tive Ten-Nation Study. The Hague: Mouton. P a x e t o , V. (1976) [1896] Sociological Writings. Selected and

introduced by S. E. Finer. Oxford: Basil Blackwell. P a r s o n s , T . , and Shils, E. A. (1951) Toward a General Theory

74 Geert Hofstede (the Netherlands)

of Action. Cambridge, Mass.: Harvard University P r e s s . Przeworski, A., and Teune, H. (1970) The Logic of Compara-

tive Social Inquiry. New York: Wiley. Pugh, D. S. (1976) "The 'Aston' Approach to the Study of Or-

ganizations." InG. Hofstede and M, S. Kassem (Eds.), European Contributions to Organization Theory. Assen, the Netherlands: Van Gorcum. Pp. 62-78.

Pugh, D. S., and Hickson, D. J. (1976) Organizational Sti-uc- ture in Its Context: The Aston Programme, I. London: Saxon House.

Robinson, W. S. (1950) "Ecological Correlations and the Be- havior of Individuals." American Sociological Review, 15, 351-57.

Tonnies, F. (1963) [1887] Community and Society. New York: Harper & Row.

Weber, M. (1970) [1921] Essays in Sociology. Edited by H. H. Gerth andC. W. Mills. London: Routledge L Kegan Paul.