Culturally Responsive Global Leadership Development

Conceptualizing and measuring cultures and

their consequences: a comparative review of

GLOBE’s and Hofstede’s approaches

Mansour Javidan1, Robert J House2, Peter W Dorfman3, Paul J Hanges4

and Mary Sully de Luque1

1The Garvin Center for Cultures and Languages of International Management, Thunderbird, The

Garvin School of International Management,

Glendale, USA; 2Wharton School of

Management, University of Pennsylvania, Philadelphia, USA; 3College of Business

Administration and Economics, New Mexico

State University, Las Cruces, USA; 4Department

of Psychology, University of Maryland, College Park, USA

Correspondence: Mansour Javidan, Professor and Director, The Garvin Center for Cultures and Languages of International Management, Thunderbird, The Garvin School of International Management, 15249 N. 59th Avenue, Glendale, AZ, USA. Tel: þ1 602 978 7013; Fax: þ1 602 843 6143; E-mail: [email protected]

Received: 4 November 2004 Revised: 7 March 2006 Accepted: 29 July 2006

Abstract This paper explains why GLOBE used a set of cultural values and practices to

measure national cultures. We show why there is no theoretical or empirical

basis for Hofstede’s criticism that GLOBE measures of values are too abstract or

for his contention that national and organizational cultures are phenomena of different order. We also show why Hofstede has a limited understanding of the

relationship between national wealth and culture. Furthermore, we explain

why Hofstede’s reanalysis of the GLOBE data is inappropriate and produces incomprehensible results. We also show the validity of managerial samples in

studying leadership. Finally, we explain why Hofstede’s claim that GLOBE

instruments reflect researchers psycho-logic reveals ignorance of psychometric methodologies designed to ensure scale reliability and construct validity.

Journal of International Business Studies (2006) 37, 897–914.

doi:10.1057/palgrave.jibs.8400234

Keywords: GLOBE; cross cultural management; national culture; cross cultural leader- ship; international management

Introduction In his critique of GLOBE in this issue, Geert Hofstede makes several important points that need further explanation and clarification (Hofstede, 2006). We are thankful to JIBS and Kwok Leung for the opportunity to respond to Hofstede’s criticisms. We are grateful to Geert Hofstede for his review of the GLOBE book. As a result of space limitations, it is impossible for this paper to address every issue raised in Hofstede’s critique. Instead, we respond to his main claims and concerns.

GLOBE is a large-scale program involving over 160 researchers from many parts of the world and a support staff of three administrators. Its objectives have been described in detail in the recently published book (House et al., 2004), which received the 2005 M. Scott Myers Award for Applied Research in the Workplace conferred by the Society for Industrial and Organizational Psychology. Our hope is that Hofstede’s critique and this rejoinder will be helpful to those with a scholarly interest in the complex issues in cross-cultural research.

Is Hofstede’s work action research based? Hofstede conducted a consulting project for IBM and later decided to reinterpret his findings in terms of how IBM employees in

Journal of International Business Studies (2006) 37, 897–914 & 2006 Academy of International Business All rights reserved 0047-2506 $30.00

www.jibs.net

different countries responded to the survey ques- tions. Contrary to his assertion, his work is not action research because action research involves a spiral of steps including fact finding, planning, action steps, evaluation, amended plans, and further action until planned change is achieved (Lewin, 1948). Hofstede’s work lacked most of these steps.

Since the publication of his book in 1980, Hofstede has provided advice on how to conduct rigorous cross-cultural research. For example: ‘To achieve good construct validity, therefore, we need both good measurements and good theory’ (Hofstede, 2001: 4). GLOBE heeded such advice, developing strong theory and rigorous measure- ments. Early in the research project, GLOBE proposed an integrated theory (House et al., 2004: 17) developed from the central proposition that attributes defining a specified culture are predictive of leadership styles and organizational practices in that culture. Further, the theory predicted that selected aspects of cultural practices will account for the economic competitiveness of nations as well as the physical and psychological well-being of their members.

As our research was theory driven, we first specified the general nature of the constructs we wanted to measure before writing items or devel- oping GLOBE scales. This critical step determines how the items should be written as well as the kinds of statistical analysis that need to be performed to assess the adequacy of the scales (Kozlowski and Klein, 2000). All major GLOBE constructs (societal culture, organizational culture, and culturally endorsed implicit leadership theory) are what multilevel researchers call convergent– emergent constructs (Kozlowski and Klein, 2000). These constructs are convergent because the responses from people within organizations or societies are believed to center about a single value usually represented by scale means. They are called emergent because, even though the origins of these constructs are a function of the cognition, affect, and personality of the survey respondents, the properties of these constructs are actually manifested at an aggregated level of analysis (i.e., the aggregated level for GLOBE was either the organizational or societal level). The GLOBE book (House et al., 2004) provides strong evidence of the scales’ psychometric properties. GLOBE instruments can be downloaded through its website, http://www.thunderbird.edu/wwwfiles/ ms/globe.

Is Hofstede’s research decentered, and is GLOBE US centered? In his critique, Hofstede (2006) suggests that his work is decentered, and contends that GLOBE is US centered. We disagree. His work is based on a consulting project that he and his European colleagues conducted for IBM in the 1960s. It is likely that, as a client, IBM – the dominant US- based corporation at the time – had very specific interests in hiring the consultants. The survey questionnaire was designed very much based on IBM’s needs and interests, which can hardly be labeled as non-US centered in the 1960s. Thus, the information collected was centered on IBM needs.

Hofstede suggests that a major decentering step in his research was the inclusion of a fifth dimension called Long- versus Short-Term Orienta- tion based on Bond and colleagues’ work with the Chinese culture (Bond and Chi, 1997; Chinese Culture Connection, 1987). Although we agree that adding this new dimension is a step towards decentering, we also see this as a reflection of the shortcoming of Hofstede’s pure empiricism. The IBM survey did not include any items related to this dimension because presumably IBM had no interest in it. Therefore, using the IBM data as the basis for discovery, Hofstede’s work did not include this dimension. Such an incremental approach of adding to the list of dimensions is due to the limitations of his original design and begs the question: what other dimensions are missing because IBM was not interested in them?

The claim that GLOBE reflects a US ‘hegemonic’ research effort is baseless. GLOBE consists of over 160 scholars from 62 cultures, referred to as country co-investigators (CCIs). The CCIs took a direct role in designing every aspect of the program. In August 1993, they met at the University of Calgary in Canada. Among the many topics of discussion, they discussed and agreed on the definition of leadership that would drive the GLOBE project. They also established the next steps in the research project. Within their respective countries, CCIs conducted focus groups and individual interviews with managers to discuss their views on outstand- ing leadership. They sent reports to Robert House on their findings. Professor House spent 6 months in the Netherlands working with a group of Dutch researchers writing items for culture and leadership. After the items were written, they were sent to all the CCIs for their feedback. CCIs reported on face validity, understandability, and relevance of the culture and leadership items in their cultures.

Conceptualizing and measuring cultures and their consequences Mansour Javidan et al

898

Journal of International Business Studies

They also reported on all concerns they had about using any particular item in their culture. Many items were reworded and some were deleted as a result of this process. Following the completion of this stage, the normal process of psychometric assessment was performed. The results of this phase were again sent to CCIs for translation and back- translation, as well as for their feedback on the usability of GLOBE items in their culture. Two rounds of pilot tests were conducted in several countries to empirically verify the instruments.

To summarize, GLOBE decided that it is time to move beyond Hofstede’s approach and to design constructs and scales that are more comprehensive, cross-culturally developed, theoretically sound, and empirically verifiable.

Why did GLOBE distinguish between cultural values and practices? Definitions of culture vary from the very inclusive (e.g., ‘culture is the human-made part of the environment’; Herskovitz, 1955) to highly focused (‘culture is a shared meaning system’; Shweder and LeVine, 1984: 110). GLOBE distinguished between cultural values and practices because of its view that national culture can be broadly defined as ‘values, beliefs, norms, and behavioral patterns of a national group’ (Leung et al., 2005).

The conventional wisdom on culture seems to reflect Hofstede’s notion of cultural onion (Hof- stede, 1980a, 2001). Hofstede (2001: 9) defines culture as ‘the collective programming of the mind that distinguishes the members of one group or category of people from another’. He views values as the invisible part of culture manifested through cultural practices, consisting of symbols, heroes, and rituals (2001: 10). He visualizes the relationship between culture, values, and practices as the ‘Onion Diagram’ (2001: 11), arguing that cultural values drive practices.

Researchers have shown that different values and value hierarchies are related to a variety of behaviors such as choice of medical specialty, choice of a university major, consumer purchases, cooperation and competition, and managerial behavior, among many others (Bianchi and Rosova, 1992; Schwartz, 1996; Bond and Chi, 1997; Barnea and Schwartz, 1998; Schwartz and Bardi, 2001; Smith et al., 2002).

There is general acceptance that the value-based framework for measuring cultures has been help- ful in deciphering cultures (Leung et al., 2002; Smith et al., 2002; Leung et al., 2005). We concurred

with this view, and incorporated the value-based approach in the GLOBE research program, but with a caveat. An in-depth review of the existing literature on culture and its relationship with behavior/practice shows that almost all recent researchers measure national or societal culture through a set of values reflecting what is important or desirable in any culture.

This approach to studying culture and its influ- ence on behavior reveals two important but untested assumptions. First, it assumes that mea- suring actors’ values is a robust way of measuring cultures. The conventional wisdom is that calculat- ing the respondents’ individual values is a sufficient measure of the collective’s culture. We call this the ecological values assumption, meaning that knowing the values of members of a culture is a sufficient way of knowing the culture.

Second, it assumes that the linkage between values and specific behaviors of actors can be generalized to the relationship between values and general behaviors of all members of the culture. As there is evidence that values are related to such practices as voting behavior or managerial beha- vior, when people in a society report that they value future orientation, then these people must also be practicing future orientation in their day-to-day activities. In other words, it is assumed that knowing values in a culture tells us about what actually happens in that culture. We call this the onion assumption, in reference to Hofstede’s Onion Diagram.

Both these assumptions are critical drivers of the existing literature. All major recent research, influenced mostly by Hofstede (1980a), accepts these assumptions without proper theoretical or empirical scrutiny. Rather than accepting these two assumptions as given, GLOBE decided not to accept the first assumption (ecological values assumption) and to verify the second (the onion assumption).

We approached the ecological values assumption from a different perspective. If learning about a collective’s values is a sufficient way of knowing its culture, then why do we use the word ‘culture’? Isn’t there more to culture than just a set of values (Schein, 1992)? We took a holistic view of culture as more than just a set of values, consisting rather of both values and actual ways in which members of a culture go about dealing with their collective challenges. We were sympathetic to a broader definition of culture proposed by Herskovitz (1948), who defined it as the ‘man-made part of the environment.’ Such an approach entails actual

Conceptualizing and measuring cultures and their consequences Mansour Javidan et al

899

Journal of International Business Studies

behavior as well as interpretations of behavior (Smith et al., 2002). In differentiating values and practices, we followed Hofstede et al.’s guidance:

Values items describe what the respondent feels ‘should be,’

practices items what she or he feels ‘is’. The distinction

between the two is present not only in the conception of

the researchers but also in the minds of the respondents.

(1990: 294)

Other researchers are also taking a broader view of culture. Leung, Bond, and their colleagues (2002; 2006) have recently introduced the notion of social axioms, which they define as general beliefs. Many of their items are comparable to GLOBE’s measures of general behavior in a society. Examples are: ‘Powerful people tend to exploit others’ and ‘The various social institutions are biased toward the rich’ (Leung and Bond, 2006).

The conventional approach to measuring cultures is through asking individual respondents about what is important to them as an individual and then aggregating the results at the culture level. Leung and Bond (2006) summarized the underlying premise for this approach:

For most people, life is not an aimless, mindless drift; their

actions and activities are conscious or unconscious mani-

festations of their responses to two fundamental questions:

What do they want to pursue in life and how do they pursue

those goals? The ‘what’ question has been extensively

researched under the rubric of values, the study of which

seeks to identify general goals that people regard as

important (e.g., Rokeach, 1973: 2).

GLOBE took a different approach. We followed Schein’s (1992) view of culture as a product of a collective’s attempts to address two sets of group issues: external adaptation and internal integra- tion. Culture evolves as a collective adapts to ongoing challenges surviving in the face of external threats and opportunities and managing relations among its members.

Instead of asking the respondents what is desir- able to them as individuals, we asked them to express their views on what is desirable in their societies. Our premise, unexplored until now, is that societal cultural dynamics go beyond mathe- matical averages of what is desirable to individual actors. Hofstede (2001: 17) describes it as follows:

Cultures are not king size individuals. They are wholes, and

their internal logic cannot be understood in the terms used

for the personality dynamics of individuals. Eco-logic differs

from individual psycho-logic.

Although we agree with Hofstede, we operationa- lized eco-logic as more than mathematical averages

of individuals’ responses. We used respondents as informants to report on the gestalt of their cultures, consisting of values and other elements. Further- more, we were sympathetic to Hofstede’s caution against using self-reports of individuals:

On top of this, we are all better observers of others than of

ourselves; but, as the experience with the LPC questionnaire

shows, in observing others we reveal something about

ourselves, too. Therefore, paper-and-pencil measures of

values through perceptions of third persons can be expected

to have greater behavioral validity than those based on self-

descriptions – this is what Fiedler proves extensively in the

case of LPC. (Hofstede, 2001: 9)

GLOBE tested the widespread onion assumption by constructing measures that ask respondents to report on how things are done in their societies. Societal cultural practices were operationalized in terms of isomorphic items to cultural values. In this way we were able to empirically assess the wide- spread but never tested assumption that cultural values drive cultural practices. Below is an example of a cultural value and a cultural practice question related to Power Distance.

Practices In this society, power is: (reverse-coded)

Values In this society, power should be: (reverse-coded)

In his critique of GLOBE, Hofstede (2006) states:

Asking ‘as is’ questions basically assumed that these people

were in a position to compare their society with other

societies. This assumption, I believe, is naı̈ve – it takes

international experience plus an unusually open mind to

produce anything like a credible comparison between one’s

own society and others. (p. 886)

We find this assertion perplexing. At no time did GLOBE ask the respondents to compare their societies with others. We simply asked the respon- dents to describe their societies as they are and as they should be. There was no reference to any other cultures. To sum up, GLOBE felt that it was time to take a more rigorous approach to understanding and measuring national cultures by separating values and practices.

Concentrated at the top Shared throughout the society

1 2 3 4 5 6 7

Concentrated at the top Shared throughout the society

1 2 3 4 5 6 7

Conceptualizing and measuring cultures and their consequences Mansour Javidan et al

900

Journal of International Business Studies

What is the empirical relationship between cultural values and practices? The onion assumption leads to the conclusion that we should expect a positive correlation between cultural values and practices. This is what we tested, but what we found is both counter-intuitive and counter to conventional wisdom. We discovered that, for seven cultural dimensions, cultural values and practices are negatively correlated. For one dimension, In-Group Collectivism, the correlation is insignificant, and for one, Gender Egalitarianism, it is significant and positive.

In his critique, Hofstede (2006) provides his explanation for the negative correlations:

My contribution to this understanding is to explain it from

the respondents’ inability to describe ‘practices’ in any

other way than by applying their ‘values’. (p. 886)

The negative correlations mean that ‘as is’ answers and

‘should be’ answers were not independent. When respon-

dents were asked to describe their society ‘as is’ this reflected

their ‘should be’ ideology. They tended to criticize their

society from an ideological point of view (from ‘things

are A but should rather be B’ to ‘things are B but should

rather be A’). (p. 886)

He goes on to explain the positive correlations in the case of two dimensions to ‘mean that in these cases the respondents tended to take the actual situation as their norm, which is a characteristic of a coherent culture’ (p. 886).

His basic point is that for some dimensions (i.e., those with negative correlations), the respondents have values opposed to practices, and in some cases they have values that are consistent with their practices (i.e., those with positive correlations). But this is no explanation. It is simply a restatement of the facts. The critical question, unanswered in his analysis, is why the correlation is negative for some and positive for other dimensions.

Hofstede also suggests that the wording of some questions may be too abstract and difficult for the respondents to understand. He used an item from Uncertainty Avoidance to illustrate the abstractness of the GLOBE items. However, further examination of specific GLOBE items does not support his point about abstractness. In the case of Institutional Collectivism, the negative correlation between values and practices is similar to the negative correlation between values and practices in Uncer- tainty Avoidance (�0.61 versus �0.62). Two of the items measuring the former dimension are: ‘In this society, leaders encourage group loyalty even if individual goals suffer’ and ‘The economic system

in this society is designed to maximize individual interest versus collective interest.’ It is difficult to argue that middle managers in participating coun- tries could not understand such questions.

A more plausible explanation for the negative correlations is that the onion assumption is too simplistic to be helpful. An in-depth examination of the relationship between cultural values and practices for Uncertainty Avoidance shows a non- linear relationship. For societies with Uncertainty Avoidance values scores below the GLOBE median (around a value score of 4.5), the higher the reported value score, the lower the reported practice score. Beyond the median point, the society’s value score has no relationship with its practice score. These results plainly cast doubt on the validity of Hofstede’s argument above and his onion assumption. To attribute such a relationship to bad items simply ignores the evidence.

Table 1 shows the relationship between cultural values and practices scores for all the nine dimen- sions. GLOBE countries are grouped into four quartiles based on their cultural practices scores. The table shows that the relationship between values and practices scores is dimension specific, but generally the most notable relationship between values and practices is manifested in societies with practice scores in the extreme regions, either high or low. For example, societies with Future Orientation practices scores in the fourth quartile (lowest) show the highest upward move in their aspirations. In contrast, societies with Assertiveness Orientation practices scores in the top quartile (highest) show the largest downward move in their aspirations. In short, the onion assumption is again in doubt. These findings constitute an important caution to researchers against assuming a simple linear relationship between values and practices.

It is also important to note that it is incorrect to interpret the negative correlations between the practices and values scores as meaning that a score above the midpoint (i.e., 4) on one scale (e.g., practices) is associated with a score below the midpoint (i.e., 4) on the other scale (e.g., values). Respondents from virtually all societies report a higher value score on Performance Orientation than their practices scores. The GLOBE average ‘values score’ on this dimension is 5.94 and the average ‘practice score’ is 4.10, yet there is a �0.28 correlation between Performance Orientation values and practices. The negative correlation between practices and values occurs because for

Conceptualizing and measuring cultures and their consequences Mansour Javidan et al

901

Journal of International Business Studies

societies with higher practices scores the increment desired is smaller than it is for those with societies with low practices scores. This finding occurs for four of the GLOBE dimensions: Performance Orientation, Future Orientation, Humane Orienta- tion, and Power Distance (for the latter in a reverse sense because societies prefer less Power Distance).

Yet it is intriguing to consider why the relation- ship between values and practices is so complex. The extant explanation for the general relationship between practices and values is that people behave in a particular way because they hold particular views on how things should be. Our findings show the opposite relationship: people may hold views

on what should be based on what they observe in action. We propose a deprivation hypothesis for this negative relationship. Societies that are reported to be relatively highly performance oriented want to be even more performance oriented, but the increment is not nearly as much as for those societies that are reported to be less performance oriented: the latter desire a much higher level of achievement. Additional research is clearly needed to determine why there is such a complex relation- ship between values and practices (House et al., 2004). Finally, our findings further lead us to conclude that the ecological values assumption is also in doubt. Studying a culture’s values is only

Table 1 Average practice (as is) and values (should be) scores for societies in different quartiles arranged by as is scores

Quartile 1 Quartile 2 Quartile 3 Quartile 4 Interpretation

Uncertainty avoidance 5.01 4.30 3.89 3.46 Societies in 1st quartile of practices have desire for

less UA, 4th quartile societies desire more UA3.99 4.72 4.83 4.95

�1.02 0.42 0.94 1.49

Future orientation 4.46 3.99 3.67 3.27 Societies in 1st quartile of practices have less strong

upward aspirations for FO than societies in 4th

quartile

5.12 5.53 5.65 5.63

0.66 1.54 1.98 2.36

Power distance 5.80 5.47 5.18 4.89 All societies have downward aspirations for PD,

strongest is for 1st quartile. Asymmetrical

relationships exist among quartiles

2.69 2.64 3.09 2.19

�3.11 �2.83 �2.09 �2.70

Institutional 4.78 4.42 4.06 3.73 Societies in 4th quartile of practices have strongest

upward aspirations for ICcollectivism 4.48 4.48 4.77 5.21

�0.30 0.06 0.71 1.48

Humane 4.71 4.23 3.90 3.51 Societies in lowest quartiles have strongest upward

aspirations for HOorientation 5.32 5.40 5.46 5.53

0.61 1.17 1.56 2.02

Performance 4.62 4.24 3.97 3.56 All societies aspire to be more PO. Those in 4th

quartile have strongest upward aspirations for PO.orientation 5.79 5.95 5.96 6.08

1.17 1.71 1.99 2.52

In-group collectivism 5.87 5.55 5.04 4.04 Societies in lower quartiles have upward aspirations

for in-group collectivism, but societies in 1st

quartile are content with current levels

5.72 5.79 5.55 5.62

�0.15 0.24 0.51 1.58

Gender egalitarianism 3.83 3.54 3.21 2.90 All Societies desire more GE, 1st quartile societies

have least strongest upward aspirations for GE

while those in 4th quartile have most upward

aspirations

4.72 4.52 4.61 4.21

0.89 0.98 1.40 1.31

Assertiveness 4.63 4.27 3.99 3.66 Societies in highest quartiles have moderate

downward aspirations for Assertiveness3.51 3.65 4.14 3.99

�1.12 �0.62 0.15 0.33

Top line in each row¼average As Is, middle line¼average Should Be, bottom line¼difference between average Should Be and average As Is).

Conceptualizing and measuring cultures and their consequences Mansour Javidan et al

902

Journal of International Business Studies

one, but not necessarily the most robust, way of measuring a culture.

Another important consequence of GLOBE’s values and practices distinction is the finding that each is related to important, but distinct, phenom- ena. That is, reported cultural practices (but not values) are associated with a large variety of societal phenomena such as economic health, national competitiveness, societal health, life expectancy, and the Human Development Index. Put simply, reported cultural practices are predictive of societal phenomena.

On the other hand, we showed that reported cultural values and not practices are associated with reported attributes of outstanding leadership across GLOBE countries. For example, we demonstrate that the country’s reported score on Power Distance values is a positive predictor of the desirability of self-protective leadership and a negative predictor of the desirability of charismatic/value-based lea- dership (House et al., 2004: 45). In other words, leaders’ reported effectiveness is associated with the society’s cultural values and aspirations, but the society’s effectiveness is associated with its cul- tural practices. This new and previously unknown phenomenon would have been undiscovered had we followed the conventional wisdom of only measuring a culture’s values. Hofstede is surprisingly quiet about this important finding. In short, there are compelling theoretical and empirical reasons for developing a more fine- grained understanding of cultures and their consequences.

Are GLOBE measures of values too abstract? In his critique, Hofstede (2006) suggests that ‘The items are formulated at a high level of abstraction, rather far removed from the respondents’ daily concerns. ’ He further referred to GLOBE values as ‘ideology’ (p. 886). Whether an item is too abstract is a judgment call. Hofstede does not suggest any criteria for making such a judgment. We explained earlier the logic of using the respondents as informants about their societies. It is obvious that questions asked for these purposes will be at a higher level of abstraction than questions asking respondents whether they plan to be with the same employer in the next five years (a Hofstede item). But the question is not useless or invalid because it is more abstract. Values are designed to be ‘broad and nonspecific’ (Hofstede et al., 1990: 291). Hofstede defined values as ‘a broad tendency to prefer certain states of affairs over others’ (1980a:

19) and referred to values as ‘abstract preferences’ (Hofstede, 1998: 484). As explained earlier, GLOBE was meticulous in designing the questionnaires to ensure that the items are reliable and valid, as well as to justify aggregation at the society or organiza- tion level. Hofstede may think GLOBE values are too abstract, but that is his personal judgment, with no conceptual or empirical support.

As to Hofstede’s point that GLOBE measured ideologies through its ‘should-be’ items, our view is that introducing yet another label is not helpful. There is no shortage of labels in the literature, and adding another concept without clarity or depth adds to the confusion. We defined values as other researchers have done, namely that which is desirable (e.g., Rokeach, 1968, 1973; Hofstede et al., 1990; Hofstede, 1998; Schwartz and Bardi, 2001; Smith et al., 2002). Kluckhohn’s (1951) definition is:

A value is a conception, explicit or implicit, distinctive of an

individual or characteristic of a group, of the desirable

which influences the selection from available modes,

means, and ends of action. (p 395)

Hofstede (2001, 2006) distinguishes between values as desired and desirable. He suggests that values as desired are more closely related to behavior than values as desirable, and are stronger predictors of behavior. He claims that his dimensions are measures of desired values. Such a distinction is not empirically supported. The available research on values that we were able to examine defines values as desirable. We could find only two researchers advocating such a distinction: Hofstede, and Levitin (1973). Related to this, Hofstede (2001: 8) suggests that ‘Values as desired are measured by such words as important/unimportant, expressing activity as well as evaluation’. In contrast, Schwartz and his colleagues define values as desirable, and use the same words important/unimportant to mea- sure them (e.g., Schwartz, 1992, 1996; Schwartz and Bardi, 2001). So, this issue is ignored by the extant literature. Furthermore, Hofstede’s claim that his values are measures of desired, thus more predictive of practices, was not supported in the only study where he and his colleagues directly linked the two (Hofstede et al., 1990). They reached the conclusion that ‘values correlated with other values but rarely with practices (as we had already found in an earlier factor analysis of individual scores)’ (p 298).

To summarize, there is no empirical or theoretical justification for the assertion that GLOBE measures of values are too abstract or that the distinction

Conceptualizing and measuring cultures and their consequences Mansour Javidan et al

903

Journal of International Business Studies

between values as desirable versus values as desired is meaningful.

Is it appropriate to measure organizational culture and national culture using isomorphic scales? Hofstede’s (2006) critique, based on a single study of a small number of organizations (Hofstede et al., 1990), suggests that organizational cultures are different phenomena from national cultures. He claims that national cultures can be measured only through a set of values and organizational cultures are best measured through a set of practices. He thus criticizes GLOBE’s approach to measuring national and organizational cultures using mea- sures of practices and values on similar dimensions.

A complete review of Hofstede et al.’s (1990) research is beyond the scope of this paper, but we shall take a brief look. He and a team of researchers interviewed and surveyed employees and managers in 20 units of ten Dutch and Danish organizations. They used ANOVA to compare the scores of organizational units on ten values questions from the original IBM study and 18 (out of a total of 54) practices items developed based on their interviews at the ten organizations. They showed that all but one of the values items and all the 18 practices questions differentiated among the 20 units (Hof- stede et al., 1990: 296, Table 1). They then reached two conclusions: first, ‘for organizations, questions on practices explain almost twice as much variance as questions on values’ (Hofstede et al., 1990: 297); second, ‘Organizational culture differences are thus composed of other elements than those that make up national culture differences’ (Hofstede et al., 1990: 312). They further claimed: ‘We believe that national cultures and organizational cultures are phenomena of different order: using the term ‘‘cultures’’ for both is, in fact, somewhat mislea- dingy’ (Hofstede et al., 1990: 313).

We reanalyzed their data to verify their first claim that organizational practices explain twice as much variance at the organizational level as do values. As a result of space limitations, the details of our reanalysis are not provided here, but will be made available upon request. Our findings show that Hofstede et al.’s (1990) first claim is based on faulty interpretation of the F-ratios. Contrary to their conclusion, there is no evidence that practices explain more variance than values at the organiza- tional level of analysis.

The second conclusion is also unjustified because they showed in their own analysis that national

values (operationalized through the IBM value items) are indeed differentiated across organiza- tional units at P-values o0.001. Hofstede’s defini- tion of culture is that part of a collective which distinguishes it from other collectives. He and his colleagues show that their measures of national culture do differentiate among organizational units, but they then reach a conclusion opposite to their own findings.

The conclusion of our reanalysis is that both claims made by Hofstede et al. (1990) are wrong, and there is no justification for suggesting that GLOBE should not use similar measures of practices and values to measure organizational and national cultures. Hofstede and his colleagues (Hofstede et al., 1990) are not supported in their claim that cultural practices are limited to the domain of organizations, and that values have no place in organizational culture. The notion that ‘national cultures and organizational cultures are phenom- ena of different order’ (Hofstede et al., 1990: 313) has no theoretical or empirical basis. As the authors pointed out: ‘The source of our research data, twenty organizational units in two North-West European countries, is obviously far too limited to claim any universality for the model’ (Hofstede et al., 1990: 313). We show here that their analyses are also dubious.

Is Hofstede’s reanalysis of the GLOBE measures meaningful? Hofstede (2006) reanalyzed the GLOBE items, and claims he found five factors, suggesting to him that ‘as far as possible (these results) resembled the original Hofstede model’. We patently disagree. There are two major issues of concern regarding Hofstede’s analysis: one is statistical in nature, the other concerns the meaningfulness and interpret- ability of the factors.

As requested, we provided Hofstede with the GLOBE item data averaged to the society level of analysis. He first conducted an exploratory princi- pal components analysis on the entire set of GLOBE culture items (i.e., 78 items). Hofstede justifies this analysis by claiming that, even though there are more items than countries, the sample size for this factor analysis is a function of the number of people (n415,000 observations) who completed the sur- vey as opposed to the number of societies used to create the society-level correlation matrix used in his factor analysis. Unfortunately, Hofstede pro- vided no support for this claim: thus it is dubious at best.

Conceptualizing and measuring cultures and their consequences Mansour Javidan et al

904

Journal of International Business Studies

Statisticians continue to assert that the appropri- ateness of a societal-level factor analysis is a function of the number of societies (i.e., 62 groups), not the number of individuals comprising the groups. For instance, Muthén and co-workers (Muthén, 1989, 1994; Muthén and Satorra, 1995) examined this issue when they developed the multilevel confirmatory factor analysis procedure. They clearly specified that the aggregated covar- iance matrix is a function of the number of societies, or, more precisely, the number of societies minus one. Thus there is no support for Hofstede’s claim that the societal-level factor analysis is a function of the number of respondents.

What is the consequence of performing the factor analysis in this manner? Widely regarded statisti- cians (Gorsuch, 1974; Everitt, 1975; Cattell, 1978; Nunnally, 1978) have been warning about the dangers of conducting factor analysis (or any other analyses) in situations in which there are more items than observations (e.g., lack of stability of factor structure). This is exactly the situation created by Hofstede’s analysis of 78 items with 62 societies. The usual recommendation is that there should be at least three to ten observations for each variable in a factor analysis (Everitt, 1975; Cattell, 1978; Nunnally, 1978). Indeed, Nunnally (1978) was so against conducting analyses when there were fewer observations than items that he declared it was one of the best ways to fool oneself with factor analysis.

For this reason, the GLOBE scales were developed and refined over multiple phases using different research methods, as well as multiple quantitative data collection periods. Specifically, after construct- ing items for each proposed cultural dimension, the first phase consisted of qualitative analyses in which various techniques such as item review by representatives from each involved country, Q-sorts, and translation/back-translation were con- ducted to assess any problematic items in our survey. An item was defined as problematic if it was not meaningful in a particular culture or if it had different interpretations across cultures. This qualitative information revealed important initial information about the measurement equivalence of the GLOBE scales. Further, the Q-sorts of the items into their respective dimensions provided initial confirmation of our cultural dimensions and provided initial information with which to struc- ture the quantitative data analysis.

We then collected three rounds of quantitative data to develop the GLOBE scales. The first two data

collection efforts were pilot studies. The final data collection was used to test the hypotheses of the GLOBE study (referred to as the Phase 2 data in House et al., 2004) and it provided ample evidence of the excellent psychometric properties of the GLOBE scales.

Although statisticians would not support Hof- stede’s factor analysis of the GLOBE items, it seems Hofstede’s exploratory analysis actually provides evidence supportive of the GLOBE scales. When he conducted his exploratory factor analysis on the aggregated GLOBE items, Hofstede indicates 18 factors were extracted. In other words, even if one disregards the extensive qualitative and quantita- tive pilot studies, and simply performs a completely data-driven exploratory factor analysis procedure, one still extracts 18 factors, the exact number of GLOBE culture scales. Nevertheless, given the inappropriate item-to-observation ratio, we did not focus on interpreting the rotated factor solu- tion. Quite simply, the rotated factor structure is too unstable to interpret meaningfully. Rather, we refer the reader to the multilevel confirmatory factor analyses we performed (Hanges and Dickson, 2004) that fit well with the qualitative and quantitative pilot studies scale development pro- cess. This more sophisticated factor-analytic proce- dure is consistent with the nested structure of the final GLOBE data set and provides empirical evidence for the appropriateness of the culture factor structures at the societal level of analysis.

The final analysis conducted by Hofstede was a second-order factor analysis in which the GLOBE culture scales were analyzed to determine whether higher-order factors could summarize the themes among the scales. Hofstede found that five factors emerged from this second-order factor analysis. He then correlated each of these five second-order GLOBE scales with the first-order culture scales that he developed in the 1980s.

There are no major technical issues with con- ducting a second-order factor analysis on the GLOBE scales. The problem with this second-order factor analysis is that these factors do not make sense. GLOBE decided to focus our discussion of societal culture on the first-order practices and values scales for the following reason: appropriate labels simply don’t exist for the five second-order factors, as is reflected in the factor structure reported by Hofstede (2006). For instance, the first factor that accounts for the largest amount of variance combines Uncertainty Avoidance, Future Orientation, In-Group Collectivism, Performance

Conceptualizing and measuring cultures and their consequences Mansour Javidan et al

905

Journal of International Business Studies

Orientation, and Gender Egalitarianism. It requires mental gymnastics to provide a meaningful label for this factor. Some of these cultural dimensions might be meaningfully related to each other, but there is no meaningful higher-order construct discussed in the cross-cultural literature to label this factor. Hofstede, however, claims he can interpret these second-order factors. He believes that they simply reflect his original scales. To convince the reader, Hofstede conducts a number of interpretative-gymnastic maneuvers to fit the obtained GLOBE second-order factor structure to his original five scales. A closer examination of Hofstede’s analyses on GLOBE scales reveals that he is actually comparing apples with oranges. Specifi- cally, he conducted a second-order factor analysis on the GLOBE scales but used his original first-order cultural scales to identify the meaning of the second-order GLOBE factors.

We conducted a similar second-order analysis on Hofstede’s scales. We performed an exploratory ecological principal components analysis on Hof- stede’s five culture scales with a varimax rotation. We used the listwise deletion approach to handle missing data, which resulted in 26 countries being used in this second-order factor analysis. Finally, the Kaiser (1960) ‘eigenvalue greater than 1’ rule was used to determine the number of factors to retain.

The results reveal that only two factors were retained (accounting for approximately 66% of the societal-level variance among the original Hofstede scales). In other words, conducting the same type of analysis with the Hofstede scales as he did with the GLOBE scales, we find that there are not five distinct Hofstede scales; rather there appear to be only two. The first factor is composed of individualism (negative loading), power distance, and long-term orientation. The second factor is composed of masculinity and uncertainty avoidance.1 Although the reader could try to label these two Hofstede culture factors as something meaningful, we do not think that it is a useful exercise, and we suspect that Hofstede would agree. In fact, Hofstede indicates that, despite the correlations among his culture scales, they show sufficient differences in the prediction of other variables such that the five cultural variables solution should be retained. We agree with his logic. As reported in the GLOBE book (House et al., 2004), there is substantial evidence for the differ- ential predictive ability of the GLOBE scales too. So, using Hofstede’s logic for keeping his five factors,

the societal GLOBE dimensions should be kept intact and not combined into incomprehensible second-order factors.

Rather than assessing the meaning of the GLOBE scales as Hofstede (2006) does in his critique, we performed a multilevel confirmatory factor analysis to test the construct validity of our scales. We believe that the Hofstede procedure of re-factor- analyzing, and then re-factor-analyzing again the original GLOBE scales creates numerous interpret- ability problems, which does not clarify the issue of relationships between the GLOBE scales and his. Instead, we conducted an analysis that can be referred to as a multilevel, multitrait, multimethod confirmatory factor analysis (CFA). It is a multilevel analysis because GLOBE scales were completed by multiple individuals nested within each society (e.g., multiple individuals reporting about their society). Multilevel CFA produces unbiased esti- mates of the societal-level factor structure because it statistically accounts for the within-group factor structure to estimate the between-group factor structure (Hanges and Dickson, 2006). It is a multitrait analysis because we simultaneously examined the factor structure of the nine cultural practices scales. It is a multimethod analysis because we examined the extent to which the unobtrusive cultural practice scales (Gupta et al., 2004) based on Culturegrams (Brigham Young University, 1999), loaded on the same factors as the GLOBE self-report measures. For a more detailed explana- tion of our unobtrusive measures, the reader is referred to the GLOBE book (House et al., 2004).

The analysis adequately fit the data (CFI¼0.94, RMSEA¼0.015). The GLOBE scales and their asso- ciated unobtrusive measures significantly loaded on their a priori specified cultural dimension. The correlations among the cultural dimensions show that, even though the latent cultural practice dimensions are correlated, these dimensions are also distinct such that they separate into the nine dimensions at the societal level of analysis.

To examine the relationship between these latent dimensions and the Hofstede scales, we performed the analysis again but this time introduced the five Hofstede scales. In this new analysis, we allowed these scales to be correlated with each of our nine latent cultural practice dimensions. Once again, the model adequately fitted the data (CFI¼0.94, RMSEA¼0.02). Table 2 shows the correlations among the nine GLOBE cultural dimensions and the (original) five Hofstede scales. From this table, it is clear that there are some meaningful connections

Conceptualizing and measuring cultures and their consequences Mansour Javidan et al

906

Journal of International Business Studies

between the Hofstede dimensions and some of the cultural constructs (e.g., Power Distance). More importantly, however, is that the significant rela- tionships are scattered, with some significant relationships appearing with constructs that are conceptually distinct (e.g., Hofstede Long-Term Orientation with latent construct of Gender Egali- tarianism) or with multiple Hofstede scales simul- taneously being related to the same cultural dimension (e.g., three Hofstede scales were signifi- cantly related to the latent Power Distance dimen- sion). Moreover, for two of the cultural dimensions (i.e., performance orientation, assertiveness orien- tation), there are no significant correlations with any of the Hofstede dimensions. Finally, for institutional collectivism practices, there were no correlations with the Hofstede scales greater than 0.40.

In other words, the overriding conclusion from these analyses is that it is impossible to support the claim that the GLOBE dimensions only replicate the five cultural dimensions presented in Hofstede’s original study. The findings further show that the nine GLOBE societal dimensions of practices show strong construct validity, as evidenced by their loading on the factor dimensions along with unobtrusive measures.

Is national wealth an antecedent of culture? Or its consequence? Or both? In his critique of GLOBE in this issue, Hofstede (2006) suggests that many measures of national culture are correlated with national wealth. He interprets the correlation to mean that culture is affected by economic factors, stating ‘Wealth sup- ports individualism, but it also relates to other dimensions’ (p. 885). His analysis focuses on the

consequences of economic prosperity, arguing that economic wealth will have an impact on a society’s culture. For example, he claims that higher levels of economic prosperity lead to lower levels of power distance (Hofstede, 1980a).

Hofstede’s interpretation is only a partial view of this relationship. It is legitimate to ask ‘What are the consequences of economic wealth?’ as he does. However, another critical question, ignored by Hofstede, but pursued by many sociologists, histor- ians, and institutional economists, is ‘What drives economic prosperity?’ (Hunt, 2000).

Max Weber (1864–1920) was perhaps the first scholar to examine this question. He theorized that the work ethic associated with Protestantism was largely responsible for the differences in wealth between northern and southern European coun- tries (Weber, 1904/1930). Many scholars have explored the impact of national culture on national economy (e.g., Hofstede, 1980a, b; Landes, 1998; Harrison and Huntington, 2000).

Banfield (1958) provided a cultural explanation of poverty and authoritarianism in Southern Italy. Myrdal (1968), in a ten-year study of South Asia, concluded that cultural factors are the principal obstacles to modernization. Kahn (1979) and Hofstede and Bond (1988) argued that the impress- ive economic growth experienced in South East Asian countries during the 1965–1985 period was due mainly to the region’s Confucian cultural roots. Having worked in the US Agency for International Development for years, Harrison (1992: 1) asks the question: ‘Why do some nations and ethnic groups do better than others?’ And his answer is: ‘The overriding significance of culture is the paramount lesson I have learned in my thirty years of work on political, economic, and social development’.

Table 2 Societal-level correlations between cultural latent constructs and Hofstede measures

Hofstede’s culture scales

Power distance Masculinity Individualism Uncertainty avoidance Long-term-orientation

Uncertainty avoidance �0.70* �0.04 �0.20 �0.42* �0.07

Institutional collectivism �0.31* �0.13 �0.31* �0.38* 0.24

Family collectivism 0.54* 0.44* �0.82* �0.05 �0.20

Assertiveness �0.06 0.29 0.04 0.13 �0.08

Performance orientation 0.05 0.26 �0.13 �0.18 0.05

Future orientation �0.39* 0.15 �0.18 �0.58* 0.03

Humane orientation 0.19* �0.04 �0.07 �0.48* 0.28

Gender egalitarianism �0.02 �0.24 0.05 �0.27 �0.58*

Power distance 0.78* 0.40* �0.11 0.33* �0.09

*Significant at Pp0.01.

Conceptualizing and measuring cultures and their consequences Mansour Javidan et al

907

Journal of International Business Studies

Landes (2000: 2), in a review of the evidence, concluded that ‘Max Weber was right. If we learn anything from the history of economic develop- ment, it is that culture makes almost all the difference’. Porter (2000: 14) agrees:

Attitudes, values, and beliefs that are sometimes collectively

referred to as ‘culture’ play an unquestioned role in human

behavior and progress. This is evident to me from working

in nations, states, regions, inner cities, and companies at

widely varying stages of development.

Different explanations are provided as to why and how culture affects economic development. Lind- say (2000) suggests that culture’s role is in shaping individuals’ thoughts about risk, reward, and opportunity, and the way they think about pro- gress. Redding (2005), however, argues that culture underlies the institutions, which in turn underlie business systems (North, 1990). Regardless of the theoretical dynamic, there is sufficient evidence to suggest that culture plays an important role in shaping the economic progress of a society. There- fore the relation between national culture and economic success can be co-evolutionary (Redding, 2005). As Javidan (2004: 117) put it:

A performance oriented society can prosper and as a result

can better educate its people who will contribute more to

their societies. The relationships among wealth, national

culture, and other archival variables are so intertwined that

they cannot be easily isolated, and cause and effect

relationships, although intuitively appealing, are hard to

verify empirically.

GLOBE showed that 12 out of its 18 values and practices dimensions are correlated with national wealth. Hofstede’s conclusion is: ‘In Hofstede (2001) I have argued that differences in values that can be accounted for by economic factors don’t need to be explained by cultural factors’ (p. 885). Such an approach is valid only if, as Hofstede (2006) believes, economic factors drive cultural factors. But if, as we have shown, cultural factors also drive economic factors, then what is associated with economic factors may in fact have cultural roots. Therefore, in controlling for economic factors, we are taking out more than the effect of national wealth; we are taking out a large chunk of cultural elements as well. That is why we disagree with his approach to controlling for economic variables.

Are middle managers an appropriate group to report on leadership? In his critique of GLOBE, Hofstede (2006) chides GLOBE for using managers in its research on

leadership: ‘Measuring leadership from survey answers by leaders is, in my eyes, a debatable approach. If you want to find out about the quality of a product, do you ask the producer or the consumers?’ (p. 884)

Our response to his question is that you actually need to ask both the consumers and the producers to know about the quality of the product. Each side provides a different but important perspective. We chose middle managers because in our view they are both producers and consumers of leadership. They report to top management and therefore experience leadership as recipients. They also manage lower-level employees and thus have experience with behaving as leaders. GLOBE pro- vides the psychometric properties of its scales, where 21 primary leadership subscales were devel- oped in a factor analysis of 112 leadership attributes and were then subjected to multilevel confirmatory factor analysis. Using a variety of statistics to assess within-organization or within-societal agreement (e.g., r(wg), ICC(1) and multilevel-confirmatory analysis), results indicated substantial support for the leadership subscales (see Hanges and Dickson, 2004). These 21 leadership subscales, in turn, were subject to a second-order maximum likelihood exploratory factor analysis. We refer to the six second-order factors as the dimensions of CLT (Culturally Endorsed Leadership Theory). The relia- bility of the scales designed to measure these global CLT dimensions are quite acceptable (i.e., average internal consistency reliability¼0.84; average inter- rater reliability¼0.95).

GLOBE asked the participating managers ques- tions regarding outstanding leadership, and did not ask them to do self-assessment as asserted by Hofstede in his critique in this issue. For each leadership attribute such as ‘Diplomatic’ (defined as skilled at interpersonal relations, tactful), respon- dents were asked to rate on a seven-point scale whether ‘this behavior or characteristic greatly contributes to (or inhibits) a person from being an outstanding leader’. Our interest was to explore the Implicit Leadership Theory (ILT) across cultures.

According to ILT, individuals hold a set of beliefs about the kinds of attributes, personality character- istics, skills, and behaviors that contribute to or impede outstanding leadership. These belief sys- tems are assumed to affect the way an individual accepts and responds to others as leaders (Lord and Maher, 1991). A major assertion of ILT is that leadership is in the ‘eye of the beholder’. That is, leadership is a social label given to individuals if

Conceptualizing and measuring cultures and their consequences Mansour Javidan et al

908

Journal of International Business Studies

either (a) their personality, attributes, and beha- viors sufficiently match the observer’s beliefs about leaders or (b) the observer attributes group success or failure to the activities of perceived leaders (Lord and Maher, 1991).

Shaw (1990) theorized that cultural values would affect the content, structure, and automaticity of leadership prototypes. Through several recent small-scale studies, researchers have supported Shaw’s (1990) hypothesis that culture influences ILTs (e.g., Gerstner and Day, 1994; Offerman et al., 1994). Our intent was to test the notion that the perception of outstanding leadership in each culture is influenced by the culturally derived implicit leadership theory. Such a test requires an appraisal of leadership in general rather than administering a self-assessment or reporting on a specific leader.

To close, the usefulness of the GLOBE leadership scales rests not only in their strong psychometric properties, but also in their ability to highlight implicit leadership theories across the world. Our findings show:

(a) substantial agreement about outstanding lea- dership requirements among respondents with- in a society;

(b) differences across societies and societal clusters; and

(c) significant relationships between the global CLT dimensions and dimensions of societal cultural values.

For instance, the cultural values of Performance Orientation and Gender Egalitarianism are asso- ciated with the desirability of participative leader- ship across societies (Dorfman et al., 2004).

Do Hofstede’s culture dimensions represent the respondents’ eco-logic, and do GLOBE’s culture dimensions represent the researchers’ psycho-logic? The essence of Hofstede’s criticism is that GLOBE items are figments of the researchers’ US-based imagination without understanding the worldview of the respondents, whereas Hofstede’s items were reflective of the respondents’ eco-logic.

To respond to this criticism, consider Table 3, which is a brief comparison of Hofstede and GLOBE research. As shown in this table, GLOBE has taken every step possible to design a truly cross-cultural instrument that can help us decipher the eco-logic of the respondents in each country. The unusually complex and complicated instrument design and

analytic procedures were designed to ensure that what GLOBE measured is valid and reliable mea- sures of the respondents’ eco-logic across cultures. As explained earlier, the culture questionnaire items were designed to measure truly ecological phenomena rather than just averaging individual phenomena. In contrast, Hofstede’s work lacked many of the critical steps required for a rigorous cross-cultural research project. The fact that his work lacks many of the steps that GLOBE took is not surprising, given that his work dates back to the 1960s. What is surprising is his assertion that, despite all the evidence, GLOBE scales are driven by the researchers’ US-centric psycho-logic without any roots in the respondents’ eco-logic.

One possible explanation for his inexplicable conclusion is that, starting with his 1980 book, he has shown a surprising disregard for the discipline of statistics and psychometrics. Here are a few important examples. As explained in Table 3, he did not provide the required psychometric properties and the evidence for aggregability in his original dimensions. As a result, he may have committed an ecological fallacy, as his dimensions were not tested for aggregability and are the averages of individual responses in each nation. Without a test for aggregability, we cannot know whether or not he committed an ecological fallacy. In discussing the relationship between cultural dimensions and national wealth, he interpreted the correlation coefficient to mean that national wealth drives culture. In examining the relation- ship between organizational and national cultures, he conducted faulty statistical analyses. Finally, in his critique in this issue (Hofstede, 2006), in justifying his inappropriate factor analysis of GLOBE’s 78 item by 56 country matrix, he claims that the problem with a flat matrix where the number of cases is smaller than the number of variables is not an issue. Making such claims, which are drastically different from the views of main- stream statisticians, without any evidence or sup- port shows a disregard for the discipline of statistics. He also disregards all the psychometric procedures used in GLOBE, especially the state-of- the-art methodologies to ensure reliability, diver- gent and convergent validity and aggregability, to conclude that GLOBE scales are figments of the research team’s imagination and are not under- stood by the respondents.

In short, there is overwhelming evidence that GLOBE instruments are ecologically focused, have nothing to do with the researchers’ psycho-logic,

Conceptualizing and measuring cultures and their consequences Mansour Javidan et al

909

Journal of International Business Studies

and pass very rigorous tests of validity, reliability, and aggregability across cultures.

So what does all this mean to cross-cultural researchers? We have responded to Hofstede’s main criticisms of GLOBE. These are important matters that deserve sufficient theoretical and empirical attention. In concluding the paper, we wish to make a few broader comments and suggest new avenues for research. To his well-deserved credit, Geert Hof- stede is among the pioneers of research in cross- cultural psychology and international manage- ment. However, the overwhelming influence of his 1980 book, based on his IBM consulting project, has perhaps made it too easy for other researchers to use his culture dimensions and associated country scores in an uncritical manner. This

presents two associated problems. First, it perpe- tuates a prevalent and false sense of confidence that all dimensions of national culture have been discovered. GLOBE has proved otherwise. Second, without presenting evidence, researchers tend to assume that national samples obtained in a cross- cultural project mirror the country characteristics obtained by Hofstede some 35 years ago. The result of the ease, simplicity, and the prevalent use of his constructs and associated country scores seems to be a tradition that can only be called Hofstedeian hegemony. No single researcher or research team, including GLOBE, should own the cross-cultural research field!

Cross-cultural researchers now have more options and opportunities, and should use them. They can continue to use the Hofstede five dimension model of cultures, or they may wish to choose a more

Table 3 Comparison of GLOBE and Hofstede research

GLOBE research program Hofstede research

Purpose To design and implement a multiphase and multimethod

program to examine the relationship between national

culture, leadership effectiveness, and societal phenomena

To conduct a post hoc interpretation of the

findings of a survey on employee morale

(1980a; 42)

Instrument

design and data

analysis

Theory-driven constructs

A total of over 160 researchers from 62 societies involved in

the research. They were directly involved in research design

starting 1993. They conducted individual and focus group

interviews with managers in their own countries

All CCIs received questionnaire items and provided reports

on their face validity, understandability and relevance in their

own cultures

Items were edited on the basis of these reports and new items

were added

The final draft of the items went through a very rigorous

psychometric process for instrument design

The surviving instruments were translated and back-

translated in each country

Pilot tests were conducted in several countries to empirically

verify the cultural dimensions

Common source error was controlled for in the research

design

Rigorous statistical procedures to verify that the scales are

aggregable, unidimensional and reliable, and to ensure cross-

cultural differences

State-of-the-art statistical techniques (HLM) used to test a

priori hypotheses, showing that the culture-to-leadership

relationships existed at organizational or societal level, not

individual level

Rigorous statistical evidence for relationship between societal

and organizational culture

Multimethod-multitrait analysis and multilevel confirmatory

factor analysis to establish construct validity

Consulting project to address the needs of a

dominant US-based corporation in the 1960s

(1980: 40) with a distinct corporate identity

(1980: 41)

Questionnaire items generated by a team of

6 European researchers to cover the issues

that were of concern to the company

identified through interviews with

employees (1980: 42)

Questionnaire items were focused on what

was relevant to each person, rather than their

society

Questionnaires were translated into local

languages without back-translation (1980:

45)

Ambiguous psychometric instrument design

process

Unclear properties on established

psychometric requirements

Emphasis on cross-country comparison

without evidence for within-country

aggregability (1980: 41)

Conceptualizing and measuring cultures and their consequences Mansour Javidan et al

910

Journal of International Business Studies

complex approach such as ours that is anchored in strong theory and empiricism. In a nutshell, GLOBE produced a set of nine dimensions, each measured twice, isometrically, as practices and respective values.2 Our advice regarding which particular cultural dimensions to focus on is that it depends entirely on the research question of interest. We have shown in the GLOBE book that values are more related to some things (such as conceptions of effective leadership) and practices are more relevant for others (such as societal phenomena). We hope that our work energizes other researchers to dig deeper and provide a more complex understanding of the role of different types of dimensions and the different roles of values and practices in regard to various phenomena. Nonetheless, the use of GLOBE findings and measures is not the only alternative. Several distinguished scholars have pursued a variety of large-scale research projects, and they should be used when the situation is appropriate.

GLOBE has found that culture also influences the type of leadership qualities that are perceived to lead to outstanding leadership. Both universally desired and universally refuted qualities were uncovered in our research. Most intriguing from a cross-cultural perspective, however, is finding that certain aspects of leadership, such as humane and/ or participative leadership, are culturally contin- gent. So many questions remain unanswered. For instance, what are the behavioral manifestations of such desirable leadership qualities such as integrity, charisma, and team orientation? One would expect that there are at least subtle differences between the enactment of each in countries as varied as China, Egypt, Brazil, and the United States (Javidan et al., 2006). One conundrum of leadership related to cultural expectations is how leaders can defy cultural imperatives in their quest for organiza- tional and cultural change. When, where, with what kind of subordinates, and under what condi- tions can leaders implement significant organiza- tional change? The manner in which leaders influence others across many cultures is an area of research currently being explored with CEOs in the next phase of the GLOBE project.

To conclude, we do not think it is hyperbole to suggest that the future of our planet depends on better understanding and acceptance among peo- ples of differing cultures. Hofstede’s research was a good start in understanding the cultural dynamics among nations. GLOBE researchers continue in this tradition. Yet increased cultural contact among

nations through globalization has not yet provided for desired stability among nations. Why? Perhaps we need to know more about when two cultures come in contact, which culture dimensions are key in that relationship. Undoubtedly, some are more important than others, but which and under what conditions? It seems to us that the cultural dynamics of cultural contact goes beyond the objective measures of each culture, as the impor- tance of culture may lie in the subjective percep- tions by those from another culture. While disturbing, we may find that increased cultural contact without a greater balance of wealth among nations could exacerbate conflict. This brings us to the complex unbundling of the culture and wealth relationship, one that GLOBE and many others have explored. The same unbundling problem occurs for relationships between culture and many elements of the human condition including socie- tal health and human development. Here we need to discover how different cultural dimensions interact as well as understanding the relative importance of each as they contribute to many societal outcomes. Are we not asking about the impact of various bundles of cultural dimensions in order to shed the most light on the real cultural fabric of a nation? Let the cross-cultural research continue!

Acknowledgements We thankfully acknowledge Vipin Gupta’s contribution to the discussion of the relationship between values and practices and his development of Table 1.

Notes 1It should be noted that we also conducted the

factor analysis using a pairwise deletion to determine whether our results were somehow due to the dropping of data from countries that did not have complete observations. Remarkably, the results with pairwise deletion were highly consistent with our previous results. Two factors were extracted. The only difference between our two sets of factor analyses is that masculinity loaded on both factors in this new pairwise deletion factor analysis.

2Hofstede, in his critique, refers to Miller’s (1994) work, which suggests that the right number of information pieces is limited to seven. Hofstede should be aware that Miller was concerned with short-term immediate memory, which is not related to research on national cultures. But it is true that we have a larger number of cultural dimensions than Hofstede’s original four (and now five) dimensions.

Conceptualizing and measuring cultures and their consequences Mansour Javidan et al

911

Journal of International Business Studies

References Banfield, E.C. (1958) The Moral Basis of a Backward Society, The

Free Press: New York. Barnea, M. and Schwartz, S.H. (1998) ‘Values and voting’,

Political Psychology 19(1): 17–40. Bianchi, G. and Rosova, V. (1992) ‘Environment As a Value:

Intraindividual, Interindividual and Intercultural Differences’, In: H. Svodoba (ed.) Culture, Nature, Landscape, International Association of Landscape Ecology: Zdar nad Sazavou, Czech Republic, pp: 37–45.

Bond, M.H. and Chi, V.M-Y. (1997) ‘Values and moral behavior in Mainland China’, Psychologia 40(4): 251–264.

Bond, M., Leung, K., Tong, K., Reimel de Carrasquel, S., Murakami, F., Yamaguchi, S., Bierbrauer, G., Singelis, T., Broer, M., Boen, F., Lambert, S., Ferreira, M., Noels, K., Van Bavel, J., Safdar, S., Zhang, J., Chen, L., Solcova, I. and Stetovska, I. (2004) ‘Culture-level dimensions of social axioms and their correlates across 41 cultures’, Journal of Cross Cultural Psychology 35(5): 548–570.

Brigham Young University (1999) Culturegrams (1988–1999 edn. 2 vol.) Brigham Young University: Provo, UT.

Cattell, R.B. (1978) The Scientific Use of Factor Analysis, Plenum Press: New York.

Chinese Culture Connection (1987) ‘Chinese values and the search for culture-free dimensions of culture’, Journal of Cross- Cultural Psychology 18(2): 143–164.

Dorfman, P., Hanges, P. and Brodbeck, F. (2004) ‘Leadership and Culture Variation: The Identification of Culturally Endorsed Leadership Profiles’, In: R.J. House, P.J. Hanges, M. Javidan, P.W. Dorfman and V. Gupta (eds.) Leadership, Culture, and Organizations: The GLOBE Study of 62 Societies, Sage: Thousand Oaks, CA, pp: 669–719.

Everitt, B.S. (1975) ‘Multivariate analysis: the need for data, and other problems’, British Journal of Psychiatry 126 (March): 237– 240.

Gerstner, C.R. and Day, D.V. (1994) ‘Cross-cultural comparison of leadership prototypes’, Leadership Quarterly 5(2): 121–134.

GLOBE website: http://www.thunderbird.edu/wwwfiles/ms/ globe.

Gorsuch, R.L. (1974) Factor Analysis, W.B. Saunders: Philadel- phia, PA.

Gupta, V., Sully de Luque, M. and House, R. (2004) ‘Multisource Construct Validity of GLOBE Scales’, In: R.J. House, P.J. Hanges, M. Javidan, P.W. Dorfman and V. Gupta (eds.) Leadership, Culture, and Organizations: The GLOBE Study of 62 Societies, Sage: Thousand Oaks, CA, pp: 152–177.

Hanges, P.J. and Dickson, M.W. (2004) ‘The Development and Validation of the GLOBE Culture and Leadership Scales’, In: R.J. House, P.J. Hanges, M. Javidan, P.W. Dorfman and V. Gupta (eds.) Leadership, Culture, and Organizations: The GLOBE Study of 62 Societies, Sage: Thousand Oaks, CA, pp: 122–151.

Hanges, P.J. and Dickson, M.W. (2006) ‘Agitation over aggregation: clarifying the development of and the nature of the GLOBE scales’, The Leadership Quarterly doi:10.1016/ j.leaqua.2006.06.004.

Harrison, L.E. (1992) ‘Wealth of Nations: Culture and Economic Growth’, In: S.D. Hunt (2000) (ed.) A General Theory of Competition: Resources, Competences, Productivity, Economic Growth (Marketing for a New Century), Sage, pp: 205–238.

Harrison, L.E. and Huntington, S.P. (2000) Culture Matters, Basic Books: New York, NY.

Herskovitz, M.J. (1948) Man and His Work: The Discipline of Cultural Anthropology, Knopf: New York.

Herskovitz, M.J. (1955) Cultural Anthropology, Knopf: New York. Hofstede, G. (1980a) Culture’s Consequences: International

Differences in Work-Related Values, Sage: London. Hofstede, G. (1980b) ‘Wealth of Nations: Culture and Economic

Growth’, In: S.D. Hunt (2000) (ed.) A General Theory of Competition: Resources, Competences, Productivity, Economic Growth (Marketing for a New Century), Sage Publications, pp: 205–238.

Hofstede, G. (1998) ‘Attitudes, values and organizational culture: disentangling the concepts’, Organization Studies 19(3): 477–492.

Hofstede, G. (2001) Culture’s Consequences: Comparing Values, Behaviors, Institutions and Organizations Across Nations, 2nd edn. Sage: Thousand Oaks, CA.

Hofstede, G. (2006) ‘What did GLOBE really measure? Researchers’ Minds Versus Respondents’ Minds’, Journal of International Business Studies 37(6): 882–896.

Hofstede, G. and Bond, M.H. (1988) ‘The Confucius connection: from cultural roots to economic growth’, Organization Dynamics 16(4): 5–21.

Hofstede, G., Neuijen, B., Ohayv, D.D. and Sanders, G. (1990) ‘Measuring organizational cultures: a qualitative and quanti- tative study across twenty cases’, Administrative Science Quarterly 35(2): 286–316.

House, R.J., Hanges, P.J., Javidan, M., Dorfman, P. and Gupta, V. (2004) Culture, Leadership, and Organizations: The GLOBE Study of 62 Societies, Sage Publications: Thousand Oaks, CA.

Hunt, S. (2000) A General Theory of Competition: Resources, Competences, Productivity, Economic Growth (Marketing for a New Century), Sage Publications: Thousand Oaks, CA.

Javidan, M. (2004) ‘Performance Orientation As a Cultural Dimension’, In: R.J. House, P.J. Hanges, M. Javidan, P.W. Dorfman and V. Gupta (eds.) Leadership, Culture, and Organizations: The GLOBE Study of 62 Societies, Sage: Thousand Oaks, CA, pp: 239–276.

Javidan, M., Dorfman, P., Sully de Luque, M. and House, R.J. (2006) ‘In the eye of the beholder: cross cultural lessons in leadership from project GLOBE’, Academy of Management Perspective 20(1): 67–90.

Kahn, H. (1979) World Economic Development: 1979 and Beyond, Croom Helm: Boulder, CO.

Kaiser, H.F. (1960) ‘The application of electronic computers to factor analysis’, Educational and Psychological Measurement 20(1): 141–151.

Kluckhohn, F.R. (1951) ‘Values and Value-Orientations in the Theory of Action: An Exploration in Definition and Classifica- tion’, In: T. Parsons and E. A. Shils (eds.) Toward a General Theory of Action, Harvard University Press: Cambridge, MA, pp: 388–433.

Kozlowski, S.W.J. and Klein, K.J. (2000) ‘A Multi-Level Approach to Theory and Research in Organizations: Contextual, Tem- poral, and Emergent Processes’, In: K.J. Klein and S.W.J. Kozlowksi (eds.) Multilevel Theory, Research, and Methods in Organizations: Foundations, Extensions, and New Directions, Jossey-Bass: San Francisco, CA, pp: 3–90.

Landes, D. (1998) The Wealth and Poverty of Nations: Why Some Are So Rich and Some Are So Poor, Norton: New York.

Landes, D. (2000) ‘Culture Makes Almost All the Difference’, In: L.E. Harrison and S.P. Huntington (eds.) Culture Matters: How Values Shape Human Progress, Basic Books: New York, NY, pp: 2–13.

Leung, K. and Bond, M.H. (2006) ‘Psycho-logic vs. ecologic: Insights from social axiom dimensions’, In F. van de Vijver and D. van Hemert (eds.) Individuals and Cultures in Multi-level Analysis, Lawrence Erlbaum: Mahwah, NJ.

Leung, K., Bhagat, R.S., Buchan, N.R., Erez, M. and Gibson, C.B. (2005) ‘Culture and international business: recent advances and their implications for future research’, Journal of Interna- tional Business Studies 36(4): 357–378.

Leung, K., Bond, M.H., Reimel de Carrasquel, S., Muñoz, C., Hernández, M., Murakami, F., Yamaguchi, S., Bierbrauer, G. and Singelis, T.M. (2002) ‘Social axioms: the search for universal dimensions of general beliefs about how the world functions’, Journal of Cross-Cultural Psychology 33(3): 286–302.

Levitin, T. (1973) ‘Values’, In: J.P. Robinson and P.R. Shaver (eds.) Measures of Social Psychological Attitudes, Institute for Social Research: Ann Arbor, MI, pp: 489–502.

Conceptualizing and measuring cultures and their consequences Mansour Javidan et al

912

Journal of International Business Studies

Lewin, K. (1948) Resolving Social Conflicts: Selected Papers on Group Dynamics, G.W. Lewin (ed.), Harper & Row: New York.

Lindsay, S. (2000) ‘Culture, Mental Models, and National Prosperity’, In: L.E. Harrison and S.P. Huntington (eds.) Culture Matters: How Values Shape Human Progress, Basic Books: New York, NY, pp: 282–295.

Lord, R.G. and Maher, K.J. (1991) Leadership and Information Processing: Linking Perceptions and Performance, Unwin- Hyman: Boston, MA.

Miller, G.A. (1994) ‘The magical number seven, plus or minus two: some limits on our capacity for processing information’, Psychological Review 101(2): 313–352.

Muthén, B.O. (1989) ‘Latent variable modeling in heteroge- neous populations’, Psychometrika 54(June): 581–597.

Muthén, B.O. (1994) ‘Multilevel covariance structure analysis’, Sociological Methods and Research 22(3): 376–398.

Muthén, B.O. and Satorra, A. (1995) ‘Complex Sample Data in Structural Equation Modeling’, In: P.V. Marsden (ed.) Socio- logical Methodology, Vol. 25, American Sociological Associa- tion/Blackwell Publishers: Washington, DC, pp: 267–316.

Myrdal, G. (1968) Asian Drama: An Inquiry into the Poverty of Nations, 3 vols, Penguin: Harmondsworth.

North, D.C. (1990) Institutions, Institutional Change, and Economic Performance, University of Cambridge Press: Cambridge, UK.

Nunnally, J.C. (1978) Psychometric Theory, 2nd edn. NY: McGraw-Hill: New York.

Offerman, L.R., Kennedy Jr, J.K. and Wirtz, P.W. (1994) ‘Implicit leadership theories: content, structure, and generalizability’, Leadership Quarterly 5(1): 43–58.

Porter, M.E. (2000) ‘Attitudes, Values, Beliefs, and the Micro- economics of Prosperity’, In: L.E. Harrison and S.P. Huntington (eds.) Culture Matters: How Values Shape Human Progress, Basic Books: New York, NY, pp: 14–28.

Redding, G. (2005) ‘The thick description and comparison of societal systems of capitalism’, Journal of International Business Studies 36(2): 123–155.

Rokeach, M. (1968) Beliefs, Attitudes, and Values, Jossey-Bass: Washington, DC.

Rokeach, M. (1973) The Nature of Human Values, Free Press: Glencoe, IL.

Schein, E.H. (1992) Organizational Culture and Leadership, 2nd edn. Jossey-Bass: San Francisco.

Schwartz, S.H. (1992) ‘Universals in the Content and Structure of Values: Theoretical Advances and Empirical Tests in 20 Countries’,’, in M.P. Zanna (ed.) Advances in experimental social psychology, Vol. 25: Academic Press: New York, pp: 1–65.

Schwartz, S.H. (1996) ‘Value Priorities and Behavior: Applying of Theory and Integrated Value Systems’, In: C. Seligman, J.M. Olson and M.P. Zanna (eds.) The Psychology of Values: the Ontario Symposium, Vol. 8, Lawrence Erlbaum: Hillsdale, NJ, pp: 1–24.

Schwartz, S.H. and Bardi, A. (2001) ‘Value hierarchies across cultures: taking a similarities perspective’, Journal of Cross- Cultural Psychology 32(3): 268–290.

Shaw, J.B. (1990) ‘A Cognitive categorization model for the study of intercultural management’, Academy of Management Review 15(4): 626–645.

Shweder, R.A and Levine, R.A. (1984) Culture Theory: Essays on Mind, Self and Emotion, Cambridge University Press: New York.

Smith, P.B., Peterson, M.F. and Schwartz, S.H. (2002) ‘Cultural values, source of guidance, and their relevance to managerial behavior: A 47-nation study’, Journal of Cross Cultural Psychology 33(2): 188–208.

Weber, M. (1930) The Protestant Ethic and the Spirit of Capitalism (T. Parson, trans.) Scribner (original work published 1904): New York.

About the authors Mansour Javidan received his MBA and PhD degrees from the Carlson School at the University

of Minnesota. He is Professor and Director of the Garvin Center for the Cultures and Languages of International Management at Thunderbird, the Garvin School of International Management in Arizona. He is on the board of directors of the GLOBE Foundation. His publications have appeared in such journals as Strategic Management Journal, Academy of Management Executive, Leader- ship Quarterly, Management International Review, Organizational Dynamics, Journal of Applied Beha- vioral Sciences, Human Relations, Journal of World Business, and Journal of Organizational Change Management. He is the Editor, Global Leadership, of the Journal of World Business. He was on the editorial board of the Academy of Management Executive for 2001–2004. He was recently elected a Fellow of the Pan Pacific Business Association.

Robert J. House holds the Joseph Frank Bernstein endowed chair of Organizational Studies at the Wharton School of Management of the University of Pennsylvania. He has published more than 130 journal articles and has received five awards for outstanding publications. His major research inter- ests include the role of personality traits and motives as they relate to effective leadership and organizational performance, and the implications of cross-cultural variation for this research. He is the Founding Principal Investigator of GLOBE and has managed the GLOBE research program since 1992.

Peter W. Dorfman is a full Professor in the Department of Management, New Mexico State University. His Masters and PhD degrees are from the University of Maryland. His articles on leader- ship, cross-cultural management, and employee discrimination have appeared in Leadership Quar- terly, Journal of Applied Psychology, Academy of Management Journal, Academy of Management Review, Journal of Management, Advances in International Comparative Management, and Advances in Global Leadership, among others. Dr Dorfman’s current research involves investigating the impact of cultural influences on managerial behavior and leadership styles. He has been a co-principal investigator of the decade-long Global Leadership and Organizational Behavioral Effectiveness (GLOBE) Research Project. As part of GLOBE, he has been a co-country investigator for Mexico and a member of the GLOBE coordinating team for overall coordination of the project, and is now an executive committee member.

Conceptualizing and measuring cultures and their consequences Mansour Javidan et al

913

Journal of International Business Studies

Paul Hanges received his PhD in industrial/orga- nizational psychology from the University of Akron. He is a full professor and director of the Industrial/Organizational Psychology program at the University of Maryland. His current research interests focus on cultural influences on leadership and cognition, justice, fairness, and employee selection, as well as research methodology. His publications have appeared in such journals as Advances in Global Leadership, Applied Psychological Measurement, Applied Psychology: An International Review, Journal of Applied Psychology, Psychological Bulletin, and The Leadership Quarterly. Paul is on the editorial board of the Journal of Applied Psychology and The Leadership Quarterly. He was a co-principal investigator of the Global Leadership and Organi-

zational Behavioral Effectiveness (GLOBE) Research Project from 1993 to 2004 and he is currently on the board of directors of the GLOBE Foundation. Paul is a fellow of the American Psychological Association and the Society for Industrial/Organi- zational Psychology.

Mary Sully de Luque (PhD, University of Nebraska) is an Assistant Professor of Management at Thun- derbird, The Garvin School of International Man- agement, and is a Research Fellow in the Center for Cultures and Languages. Her research includes the influences of culture in the areas of leadership, feedback processes in the work environment, and human resource management. She serves as the research director of the GLOBE Phase Three Project.

Accepted by Kwok Leung, Deputy Editor-in-Chief, 29 July 2006. This paper has been with the author for one revision.

Conceptualizing and measuring cultures and their consequences Mansour Javidan et al

914

Journal of International Business Studies