national culture, networks, and individual influence in a multinational management team

13
National Culture, Networks, and Individual Influence in a Multinational Management Team Author(s): Jane E. Salk and Mary Yoko Brannen Source: The Academy of Management Journal, Vol. 43, No. 2 (Apr., 2000), pp. 191-202 Published by: Academy of Management Stable URL: http://www.jstor.org/stable/1556376 Accessed: 16/09/2009 00:48 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://dv1litvip.jstor.org/action/showPublisher?publisherCode=aom. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that promotes the discovery and use of these resources. For more information about JSTOR, please contact [email protected]. Academy of Management is collaborating with JSTOR to digitize, preserve and extend access to The Academy of Management Journal. http://www.jstor.org

Upload: satria

Post on 16-Nov-2014

664 views

Category:

Documents


0 download

DESCRIPTION

Research Artikel

TRANSCRIPT

Page 1: National Culture, Networks, And Individual Influence in a Multinational Management Team

National Culture, Networks, and Individual Influence in a Multinational Management TeamAuthor(s): Jane E. Salk and Mary Yoko BrannenSource: The Academy of Management Journal, Vol. 43, No. 2 (Apr., 2000), pp. 191-202Published by: Academy of ManagementStable URL: http://www.jstor.org/stable/1556376Accessed: 16/09/2009 00:48

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available athttp://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unlessyou have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and youmay use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained athttp://dv1litvip.jstor.org/action/showPublisher?publisherCode=aom.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printedpage of such transmission.

JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with thescholarly community to preserve their work and the materials they rely upon, and to build a common research platform thatpromotes the discovery and use of these resources. For more information about JSTOR, please contact [email protected].

Academy of Management is collaborating with JSTOR to digitize, preserve and extend access to The Academyof Management Journal.

http://www.jstor.org

Page 2: National Culture, Networks, And Individual Influence in a Multinational Management Team

c Academy of Management Journal 2000, Vol. 43, No. 2, 191-202.

RESEARCH NOTES

NATIONAL CULTURE, NETWORKS, AND INDIVIDUAL INFLUENCE IN A MULTINATIONAL MANAGEMENT TEAM

JANE E. SALK ESSEC

MARY YOKO BRANNEN San Jose State University

and University of Michigan

Individual influence is thought to shape team performance. However, empirical stud- ies of its potential determinants in multicultural teams, including national culture, are lacking. A network study of the management team of a 50-50 German-Japanese inter- national joint venture revealed multiple significant determinants, with advice central- ity the most closely associated with influence. National culture, though not statistically significant, remained important in explaining patterns of relationships in this team, but its role was far less direct and deterministic than suggested by prior research.

International joint ventures (IJVs) often use multi- national teams comprising managers from the parent companies. Cross-cultural and IJV studies often iden- tify cultural differences in such teams as causing many difficulties, including conflict, misunderstand- ing, and poor performance (Bivens & Lowell, 1966; Killing, 1983; Shenkar & Zeira, 1992). However, high- performing IJVs and organizations with multinational management teams, including ABB, Shell, Unilever, and ICL-Fujitsu (now merged), suggest that cultural diversity does not necessarily lead to poor perfor- mance. Cultural diversity might even confer an ad- vantage by giving managers a broader range of per- spectives for managing complex cultural systems (Boyacigiller & Adler, 1991). The purpose of this study was to enable a better understanding of indi- vidual influence and intergroup interaction in a mul- tinational team setting. Specifically, we attempted to answer the question, What are the determinants of individual influence in a well-functioning multina- tional management team?

Though networks researchers have paid limited at- tention to individual-level aspects of how social net- works affect influence and even less to multicultural

We are indebted to CIBER at the Fuqua School of Business, Duke University, and to the University of Michigan Business School for providing research sup- port. We would also like to thank Wayne Baker, Mark Fruin, Mark Mizruchi, and Anne Tsui for their comments and Jeffrey Johnson, David Krackhardt, and Ray Fried- man for offering technical advice early in the project.

networks, extant theory and methods developed in the social networks domain to study influence pro- vide a sound foundation for addressing this question. In incorporating theory and methods from this body of research, our goal was twofold: (1) to enrich cross- cultural and IJV theory and empirical knowledge con- cerning individual influence and its potential deter- minants, including national culture and (2) to extend a social networks perspective to multinational teams and organizations.

CONCEP'TUAL FOUNDATIONS

Our primary interest was in comparing the role of national origins to the roles of other potential deter- minants of influence in a multinational management team. However, the context of the team studied-a 50-50 shared management IJV (a joint venture with equal parent equity in which individuals from the parent firms are top managers)-has conceptual im- portance, reflecting partner intentions concerning is- sues of influence and control (Killing, 1983; Salk, 1996). IJVs use shared staffing to mobilize resources, combine knowledge, and balance influence from both parents to enhance performance (Salk, 1996). That said, some researchers see IJVs engaged in such prac- tices as particularly fertile ground for performance problems stemming from cultural differences (Kill- ing, 1983; Lane & Beamish, 1990; Salk, 1992, 1996). Lyles and Salk (1996) found higher average reported learning from foreign parents and significantly greater sensitivity to cultural conflict for 50-50 IJVs.

191

Page 3: National Culture, Networks, And Individual Influence in a Multinational Management Team

Academy of Management Journal

They reasoned that the effects of operational influ- ence and control by one group or the other are likely to be unclear in these IJVs. Shared management might open the door for more conflict because neither group has a formal mandate to unilaterally control opera- tions. At the same time, the lack of formal dominance by one group leaves more latitude for individual in- fluence patterns to emerge and for the acquisition and use of particular knowledge held by individuals.

Nationality, Social Identity, and Influence

Social identity theory (Tajfel & Turner, 1986) and social categorization theory (Turner, 1987), two re- search streams integrated into the broader cross- cultural literature by Erez and Earley (1993), both suggest that even superficial differences might re- sult in team members choosing national culture as a primary form of identity. According to social identity theory, individuals are primed by situa- tional and other cues to "enact" primary social identities as a basis for self-evaluation and en- hancement of self-images (Erez & Earley, 1993). A salient social identity leads to accentuation of in- traclass similarities and interclass differences (Erez & Earley, 1993: 151). Social identity theory re- searchers (Tajfel & Turner, 1986) have created lab- oratory experiments that produced in-group/out- group discrimination, even when subjects were assigned to groups at random. Social identity the- ory and social categorization theory suggest that race, accent, and situational factors might create and maintain in-group preferences in some types of interactions, despite convergence, or the develop- ment of complementary orientations among indi- viduals from different national cultures in how they work together. These preferences could in turn cause differences in individual influence to be as- sociated with nationality. However, it should be noted that research has yet to link enacted cultural identities to how individuals form and participate in social networks in organizations. It is important to recognize that someone could be anti-Semitic toward Jews as a social category, for example, yet still include many particular members of this social category in work and even friendship networks.

Cultural Identities and Individual Orientations and Behavior

Cross-cultural research (Hall, 1983; Hampden- Turner & Trompenaars, 1993; Hofstede, 1984; Ronen & Shenkar, 1985; Triandis, 1989) has catalogued how basic assumptions, values, and behavioral norms vary across cultures. Culture has been defined as the

the members of a group (Schein, 1985) and as the "collective programming of the mind" (Hofstede, 1984: 21). National culture, as an often taken-for- granted product of primary socialization, is thought to be particularly potent, and its effects on individu- als are thought to be particularly resistant to change. However, Schein (1985) suggested that organizational cultures arise from specific historical events experi- enced by a group and organization as well as from the influence of individuals engaged in their routine in- teractions. Thus, within the same, relatively stable national culture context, organizational and group cultures can vary widely (Brannen, 1994).

Preliminary evidence suggests that certain as- pects of national culture represented in IJV teams, whether or not differences associated with them disappear, may at least become less salient over time. In a comparative study of multinational man- agement teams in three IJVs, Salk (1996) found that in-group identification according to national ori- gins was strong in the earliest months of the IJVs, but that other primary identity boundaries became more relevant by the end of the first year of opera- tions in two of the cases. Given that national cul- ture is deeply rooted in the socialization of manag- ers, it is plausible that nationality continued to affect team members' preferences and work behav- iors. Even if members accord less salience to na- tional identities over time, national identity-based intergroup dynamics could persist in the absence of conflict and performance difficulties.

Researchers have replicated basic differences among national cultures, taken as central tenden- cies of large populations. However, ethnographic studies of multinational teams and organizations have shown that individual expatriates do not nec- essarily reflect general population tendencies (Brannen, 1994; Schneider & Barsoux, 1997; Sumi- hara, 1992). Team members might be more homo- geneous than their respective national populations in terms of education, occupation, and socioeco- nomic subgroup memberships, while diverging in terms of generational, religious, geographic, or other subgroup affiliations within a national cul- ture. Multilayered participation in diverse sub- groups helps explain the observed variance within national cultures and why individuals from the same national cultural group import different nor- mative and behavioral expectations and orienta- tions into a team setting (Brannen, 1994). The lit- eratures on social identities and on culture led us to pose the following research questions:

Research Question 1. Will patterns of in-group bias be revealed, especially in terms of na-

assumptions, values, and artifacts that are shared by

192 April

tional culture, in multinational teams?

Page 4: National Culture, Networks, And Individual Influence in a Multinational Management Team

Salk and Brannen

Research Question 2. Will there be systematic differences between national subgroups in multinational teams in terms of individual in- fluence and the factors associated with indi- vidual influence?

It is possible that, despite attempts by interna- tional joint venture parents to avoid operational dominance by one group or another, one national group will be more influential. Research Question 2 is methodologically as well as substantively impor- tant as a diagnostic. For example, if the average scores on measures of explanatory variables for the members of two nationalities are systematically dif- ferent, separate models of influence for each sub- group might need to be developed.

Using a Japanese-German IJV as a Research Context

Cross-cultural research suggests specific differ- ences between Japanese and German managers if they are treated a priori as representative of aggre- gate cultural tendencies. Two differences refer- enced in this literature concern advice seeking and the general importance ascribed to socioemotional relationships. German managers are known for placing a high reliance on expertise and formal, individual responsibility, and Japanese managers are known for having a generalist and diffuse view of responsibility (Brannen & Salk, 2000; Hampden- Turner & Trompenaars, 1993). Moreover, Hall (1983: 174) suggested that establishing and main- taining interpersonal contact and harmony on an emotional level is vital for daily team functioning for Japanese people but not for Germans. Japanese managers spend much time interacting informally during and after work (Brannen, 1994).

Hampden-Turner and Trompenaars (1993) sug- gested that both Japanese and German managers were likely to share a group or organizational ori- entation rather than an individualistic orientation. However, for the Japanese the operative logic is wa (harmonious aesthetics of relationships), and for the Germans it is the rule-based logic of Konsens und Ordnung (consensus and orderliness; Hampden- Turner & Trompenaars, 1993: 101). These authors further noted that "the German or G-type personal- ity has a large, dark private space and a small outer public space, and even this is relatively inaccessi- ble, meaning that Germans are not easy to get to know" (1993: 223). Moreover, "[German] co-work- ers may have no social relations outside the work- place. Indeed, this is quite common" (1993: 224). Japanese managers might therefore apply a socio-

lationships, but German managers might follow the formal division of responsibility, establishing few if any socioemotional bonds with other employees. These ideas motivate a question concerning IJV teams composed of Japanese and German nationals:

Research Question 3. Will Japanese managers use a socioemotional logic in developing their task and advice networks, and will German man- agers use a logic of choice based on expertise?

Position, Orientation to Local Norms, Network Centrality, and Individual Influence

Social network research and theory have primar- ily evolved in uninational settings. But we believe that network concepts can be applied to enrich understanding of individual influence in IJV and other multicultural settings. Individual influence in an organization entails being able to induce oth- ers to do things and behave in ways that they might not otherwise (e.g., French & Raven, 1959; Kotter, 1985). Ibarra's (1993) network-based study of indi- vidual influence and innovation identified three constructs associated with individual influence: formal position, personal characteristics, and net- work centrality. Control over resources, rewards and punishments, authority to exercise influence in particular domains (legitimacy), and perceived expertise often adhere to formal positions in orga- nizations (French & Raven, 1959; Ibarra, 1993).

Social influence also arises from characteristics of an individual that induce identification and in- teraction with that individual (French & Raven, 1959; Ibarra, 1993; Shah, 1998). Interpersonal at- traction has been linked to demographic similarity between actors. Actors have multiple demographic identities, including age, race, gender, nationality, education, and, in IJVs, parent organizational affil- iation. Actors also can be similar or different in terms of unit or functional affiliation in an organi- zation (Shah, 1998). As discussed earlier, work in social identity theory and social categorization the- ory, as well as cross-cultural studies (Erez & Earley, 1993), proposes that individuals choose salient identities for social comparison that enhance self- identities and self-efficacy. This suggests that whether the nationality of managers is a relevant basis for attraction could be a matter of choice.

Another aspect of personal differences is individ- ual orientation toward locally established norms. Ac- ceptance as a member of a group in an organization entails being able to act in ways that are consistent with the normative expectations of other members (Van Maanen & Schein, 1979). Such behavior is

emotional choice criterion to different types of re-

2000 193

thought to be related to influence (e.g., Kotter, 1985).

Page 5: National Culture, Networks, And Individual Influence in a Multinational Management Team

194 Academy of Manage

Hall (1983) suggested that synchronization in critical domains such as decision making is vital but prob- lematic for multicultural teams. Over time all human groups develop a common set of precedents and ex- periences that become the foundation for a context- specific, or local, working culture (Brannen & Salk, 2000; Schein, 1985). Interviews and participant ob- servation revealed convergence around emergent lo- cal norms for decision making in a team comprising Japanese and German managers (Brannen & Salk, 2000). The local norms entailed including fewer man- agers in meetings than the Japanese preferred and front-loading discussions with the parent headquar- ters in Japan to speed things up. The Japanese and German managers described the norms surrounding emergent decision-making processes in the same way, but the Germans perceived the processes as slow, while the Japanese saw them as fast. Certain decision-making issues-specifically, the values of consensus, speed, and efficiency-surfaced in all the open-ended interviews we conducted (Brannen & Salk, 2000).

To unpack this group's culture, therefore, also re- quired taking into account how individual team members perceived and reacted to emergent norms. The degree to which individual members came to accept the acknowledged local norms about consen- sus seeking and fast and efficient decision making and the general investment they had made in adap- tation could all be seen as indicators of comfort and feelings of self-efficacy in this specific group setting.

The centrality of individuals in social networks provides a distinct basis for exercising influence (e.g., Brass & Burkhardt, 1993; Ibarra, 1993). Social networks reflect patterns of communication and exchange among members of an organizational group. Individuals who are central in networks can exert influence by being in a position to mediate flows of information and resources between others, because others seek them out to communicate or consult with and/or because they are connected to other powerful actors. Network theorists distin- guish between instrumental and socioemotional (or, in U.S. settings, friendship) networks; instru- mental networks can further be subdivided into task, advice, and information networks. Ibarra's (1993) results supported a positive link between network centrality and influence in organizational innovation. She also found that network centrality explained significant variance in influence over and above that explained by formal position and personal characteristics. The above discussion frames the following research questions:

Research Question 4a. Will formal position be related to individual influence?

,ment Journal April

Research Question 4b. Will the nationality of a manager be related to individual influence?

Research Question 4c. Will orientations to lo- cal, emergent norms (in this particular case, adaptation, a preference for consensus, and a belief that decision making is fast and efficient in a given setting) be related to individual in- fluence?

Research Question 4d. Will network centrality be related to individual influence? Research Question 4e. Will network centrality be more strongly associated with individual influence than other determinants, such as for- mal position, nationality, or orientation to lo- cal norms?

METHODS

Research Setting

We identified the organization that served as our research setting via contacts with the Japanese par- ent and had preliminary discussions with its co- directors. The Nutech IJV (a pseudonym), located in a small town in the Ruhr region of Germany, was established in 1990, and production commenced in late 1991. The Japanese parent, a medium-sized firm like the German parent, was motivated by the strategic need to have a manufacturing platform from which to service and expand a market pres- ence inside the European Community. A commit- ment to keep the plant open, despite changes in the competitive environment, motivated the German parent. Both parents were in the same industry and had had some prior contact via a licensing agree- ment in the 1970s and 1980s. Though a few of the parents' managers had some prior international ex- perience, Nutech was the first joint venture expe- rience for everyone concerned. All parent and IJV informants considered the IJV to be a strong per- former in terms of meeting its business plans and in having generally good working relationships among team members. Participant observation re- search conducted at the site before data collection for this study suggested that the few problems the IJV had experienced were related to getting the new production technology up to speed. Historically, manufacturing and R&D functions had played a particularly important role in the success of the IJV.

Ownership and governance were shared equally in the IJV. The Japanese and German parent each owned 47.5 percent of the IJV's equity. A Japanese trading company, which acted as a distributor of the IJV's products, took a 5 percent ownership share, but it had neither a voting position on the governance board nor

Page 6: National Culture, Networks, And Individual Influence in a Multinational Management Team

Salk and Brannen

a mandate for active oversight of IJV management. As of the middle of 1994, the IJV employed 216 individ- uals-145 wage earners and 71 salaried managers. Shared governance was confined to the top three levels of the organization. Of the 16 top team mem- bers, 5 were Japanese, 10 were German, and 1 was South African (a junior manager in quality control). A Japanese codirector and a German codirector shared the top position.

Of the 9 positions at the top two levels, 4 had Japanese incumbents, and of the 16 positions at the top three levels, 5 had Japanese incumbents. How- ever, the greater number of Germans in these posi- tions should not be viewed as identical with greater formal power. The Japanese occupied the top man- agement of the sales/marketing and technology areas, with two-thirds of the top positions in these two functions held by Japanese managers, and the Ger- mans dominated production and administrative functions. The preponderance of Germans in the lat- ter two functions reflected the need to operate effec- tively with a German workforce and in a German legal and business context. English was the business language of the IJV; only one of the Japanese manag- ers spoke more than a few words of German, and only one German manager (a junior team member, two levels below the top) spoke a little Japanese. The Japanese were somewhat older on the average (since age-based seniority is practiced in Japan), but the age ranges were quite wide on both sides.

Ten semistructured interviews (composed of open- ended questions and conducted in German, Japanese, and English), done seven months prior to collection of the network data, all pointed to cultural differences as an aspect of IJV experience that differed from in- terviewees' past management experiences. Content analysis suggested that some of these differences (dif- ferent degrees of autonomy practiced by the parents and different decision-making styles) had resulted in the creation of new IJV-specific ways of doing things, and other differences (such as the Japanese desire for informal interaction and the German tendency to sep- arate work and private relations) persisted (Brannen & Salk, 2000). This variation begged the question of how the evolved fabric of social relations, including influence, might be explained by nationality or other factors. Finally, the interviews established that prob- lems installing the equipment and ramping up output quality and quantity had been the major contingen- cies faced by the team in its first two years (Brannen & Salk, 2000).

Data Collection and Measures

A questionnaire was designed to follow up on the

gave this questionnaire, which was written in En- glish, to all 16 top team members during a visit in May 1995. It was administered in the form of a structured interview with hard text accompani- ment for the informants to follow and mark an- swers on. Since one author was fluent in Japanese and one was fluent in German, informants could choose whether to give their answers immediately or to ask for clarification in English or their native language. We had agreed on prepared answers to informant questions in Japanese and Germans to control for consistency in questionnaire execution.

Network data and measures. In our study, as in many network studies (e.g., Ibarra, 1993; Padgett & Ansell, 1993; Shah, 1998), data from all members of a bounded population were collected and ana- lyzed. We determined the boundaries of the team to be studied in two ways: first, by asking the co- directors of the targeted IJV to define these bound- aries (they named the first three hierarchical levels) and, second, by leaving a blank space at the bottom of each question about networks to allow infor- mants to add missing individuals. One member of the third level was eliminated because of the nature of his role and other team members' reporting next to no interaction with him. Only a few people in- serted additional names, and no name appeared more than twice, further confirming that the boundary established for the top team was valid. Thus, we had data from 100 percent of the team's members (16 individuals).

The network data collected covered three net- work types: task-related, advice-related, and pri- vate. We based our network questions on Ibarra (1993), making modifications as needed. To mea- sure task-related communication, we asked "Who do you talk to regularly about business-related mat- ters? Place an X by those you talk to more than once a week on the average." To measure advice-related communication, we asked "Who do you go to for advice when you have a work-related problem or a decision you have to make?" For private commu- nication, we asked "Who have you met with pri- vately outside of Nutech? Place an X by the people who you meet privately at least once a month on average." Team members could check as many names as they wished for each question and add names, as noted above. We used the term "private" rather than the term "friendship" because we pre- dicted that the Japanese managers might tend to call everyone their friend, while the Germans might feel a question about friends to be an invasion of privacy. The level of personal affect that justifies calling someone a friend in the United States might not be high enough for German respondents to

interviews and observation described above. We

2000 195

make a similar attribution, as friendship implies a

Page 7: National Culture, Networks, And Individual Influence in a Multinational Management Team

Academy of Management Journal

deeper and more enduring commitment for Ger- mans than it does for Americans (Lewin, 1936). Hence, we viewed an individual's choosing to see a fellow team member privately, outside of working hours, as an appropriate threshold defining the elective, socioemotional network, capturing the es- sence of what friendship indexes in a U.S. context.

Of the several different ways of measuring central- ity (Bonacich, 1987; Freeman, 1979), a key construct for our analysis, two were of particular conceptual interest. The first was in-degree centrality, a measure of how many nominations by others an individual receives, and thus an indicator of deference or status (Wasserman & Faust, 1994). The second was be- tweenness centrality, a measure of the degree to which an individual is in a position to mediate flows of information between others. Ties were not sym- metricized because lack of reciprocity can indicate status differences (Krackhart, 1990), and reciproca- tion rates for the three networks studied here were relatively low (48-66%). Some past empirical stud- ies have found different types of centrality to be highly correlated with one another (e.g., Ibarra, 1993). In our data, the correlation for the task-related com- munication network between in-degree and between- ness centrality was .70 (p

- .003); for advice-related communication, it was .76 (p - .001); and for private (socioemotional) communication, it was .63 (p -

.011). Thus, we did not combine centrality measures and report results using in-degree centrality, though betweenness centrality produced results by and large similar to those we report. Network data also were used to calculate density, the proportion of actual dyadic relationships relative to potential ones.

Other independent variables. We included two measures of formal position. For the ordinal vari- able level, the codirectors were assigned a 3; those who reported directly to them received a 2; and members of the third team level received a 1. Func- tion had five categories. Because manufacturing and R&D were especially important in the early history of the IJV, we created a dummy variable, assigning those in the administrative and sales functions a 0 and those in manufacturing and R&D a 1. Nationality was confirmed during the admin- istration of the questionnaire, with Japanese coded 0 and German coded 1. Orientation to local norms was measured by three variables, decisions, prefers consensus, and adapt, each tapped by a question- naire item measured on a seven-point scale ranging from 1, "strongly disagree," to 7, "strongly agree": "Decision-making by Nutech management is fast and efficient" (decisions). "I prefer a decision-mak- ing process that allows sufficient time for a consen-

been necessary to change my way of working to be effective in this organization" (adapt).

Dependent variable. To measure influence, we asked this question: "Please rate the amount of influence that the following people have at Nutech." A list of team members (plus blank spaces for nominating others) followed. After each name, a 1-7 scale was presented, and respondents were requested to assign an influence rating to each in- dividual, with 1 for "little influence in the IJV" and 7 for "highly influential in the IJV." We then cal- culated the average rating for each individual, elim- inating the self-assigned ratings.

Control variable. Tenure was used as a control variable because establishing ties and influence in a group should take time. It was measured as the number of months between the time a manager joined the IJV and data collection.

Data Analysis

Analyses performed here employed SAS (Hatcher & Stepanski, 1994) and UCINET IV (Borgatti, Everett, & Freeman, 1992). We used UCINET IV to analyze a priori block models and to calculate network densi- ties and centrality scores. SAS was used to calculate Spearman correlation coefficients, t-tests, and ordi- nary least squares (OLS) regression models.1

We examined the level of general cohesion and integration of the team by calculating reachability, defined as the distance between actors, using the number of edges in the shortest path (Wasserman & Faust, 1994). We sought evidence of in-group bias using the Gibbs and Martin index of industrial diver- sification (Baker, 1992: 410). The Gibbs and Martin index gives the probability (H) that two randomly selected actors are from different groups. It is calcu- lated as H = [1 - pk][NI(N - 1)], where p is the proportion of actors in category k and N is the size of the total number of individuals. Multiplying the in- dex by N/(N - 1) corrects for selecting the same individual twice. In an ideal network, where choices are not based on in-group preference, the percentage of ties between groups should be greater than or equal to the Gibbs and Martin index, and a percentage lower than the index indicates discrimination in fa- vor of in-group interaction. We also used a priori

1 There has been some concern that network data vio- late a basic assumption of OLS regression, namely, that the observations are not independent when centrality scores are used. OLS regression is an accepted method for analyzing such data, however (for a discussion of this

sus to be reached" (prefers consensus). And "It has

196 April

point, see Krackhardt [1990]).

Page 8: National Culture, Networks, And Individual Influence in a Multinational Management Team

Salk and Brannen

block models based on nationality (German or Japa- nese) to test for in-group bias.

RESULTS

Subgroup Bias and Differences

In-group favoritism. Research Question 1 asks whether our data will reveal patterns of in-group bias, especially in terms of national culture. Results obtained with the Gibbs and Martin index indicate such bias for the task-related and advice-related networks and perhaps for the private networks. For the studied team, composed of five Japanese and ten German members, the value of the index is .48. Thus, whenever the index measure was greater than the proportion of out-group ties in a network, there was evidence of in-group bias. For the task- related network, the proportion of out-group ties was .41; for the advice-related network, it was .38; and for the private relations network, it was .48, the cutoff. The sparseness of German private relations probably accounts for this last result.

Subgroup differences. Question 2 asks whether there will be systematic differences between na- tional subgroups on multinational teams in indi- vidual influence and associated factors. Private in- degree centrality differed (Japanese, x = 45.33, s.d. = 7.30; German, x = 20.00, s.d. = 17.21; t = 3.99; p - .002), but there were no differences in other centrality measures, influence ratings, formal posi- tion, and tenure for the national subgroups. The sparseness of private ties and differences between the Germans and Japanese (explained at greater length in the next section) drove the private in- degree centrality difference.

For measures of orientation to local norms, only the degree to which individuals reported a need to adapt their way of working to be effective in the IJV differed (Japanese, x = 5.80, s.d. = 1.30; German, x = 3.90, s.d. = 1.85; t = 2.30, p - .04). This finding is not surprising, since the Japanese were expatri- ates and the Germans were operating in their home country. Lack of significant differences between the German and Japanese members in preference for decision-making speed, coupled with variance within subgroups, suggests that individual re- sponses to these questions did not directly reflect national cultural norms and preferences.

Reachability analyses (Wasserman & Faust, 1994) further suggested that the team was reasonably well integrated. For task-related communication, all in- dividuals could reach one another with a maxi- mum distance of 2. For advice network relations, the maximum distance was 3. The analysis of the

story. Here we found many pairs for whom the minimum number of connecting links was great; for six members (all Germans), the distance was infinite for all row distances. As suggested in the literature review, the Germans might be expected to make a strict distinction between their work and private relations. The extremely low number of pri- vate relationships reported corroborates this idea. In contrast, looking at private relations for the five Japanese team members, we saw a maximum dis- tance of 2; this finding suggests that the Japanese managers engaged in rich, multilayered personal and professional interactions.

The logic underlying the networks. Forming a priori blocks based on nationality allowed us to a,nalyze patterns of interaction within and between subgroups (see Table 1 for reduced block matrixes) and thus address Research Question 3. This ques- tion asks whether Japanese managers will employ a socioemotional logic in developing their task and advice networks, while Germans employ a logic based on expertise. For task-related communica- tion, the average in-group densities (.70 for the Germans and .90 for the Japanese) were higher than the out-group densities (.50 for the Germans and .66 for the Japanese). If the overall average density of .67 is taken as a cutoff, then only the in-group densities are higher, though it should be noted that the number of relationships (ties) with Germans referenced by Japanese informants barely missed this cutoff. Thus, for the task-related networks, there is no evidence for different logics. Both the German and Japanese managers preferred to have in-group members in their task networks. This find-

TABLE 1 Reduced Block Matrixes: Within- and Between-

Block Average Densities for the German and Japanese Subgroupsa

Networkb Japanese German

Task-related communication Japanese .95 .66 German .50 .70

Advice-related communication Japanese .70 .24 German .48 .49

Private communication Japanese .85 .28 German .22 .11

a For the Japanese subgroup, n = 5; for the German subgroup, n = 10.

b Respective averages are .67, task-related; .45, advice-related; private network, on the other hand, told a different and .25, private.

2000 197

Page 9: National Culture, Networks, And Individual Influence in a Multinational Management Team

Academy of Management Journal

ing also supports the existence of in-group bias (Research Question 1).

For the advice networks, we found that the Japa- nese exhibited high average in-group density (.70) and low out-group density (.24), thus showing a very high degree of reliance for advice on the in-group. By contrast, the Germans showed almost no difference in their propensity to use in-group and out-group mem- bers for advice (out-group, .48; in-group, .49). Only the Japanese out-group nominations of advice rela- tionships (.24) were below the overall advice network average of .45. The reduced block model for private relations further illuminates these results. For this team, the overall average is .25, a low value that was, as noted earlier, driven by the infrequency of private ties among the Germans (out-group, .22; in-group, .11); by contrast, the Japanese average in-group den- sity was .85, and the out-group density was .28. Thus, these results suggest that the Japanese managers used a socioemotional logic in forming their advice net- works, while the Germans used a logic based on per- ceived expertise.

Research Questions 4a-4e: Determinants of Influence

Table 2 presents means, standard deviations, and Spearman correlation coefficients. Table 3 presents the results of regression analyses used to test Re- search Questions 4a-4e. Only a few explanatory variables are highly correlated: Advice centrality is highly correlated with the speed and efficiency of decision making (.71) and a preference for consen- sus (.67). These variables (decisions and prefers consensus) were highly correlated (.83). We in- cluded data from the one South African team mem-

ber in calculating influence. However, we calcu- lated the regression equations outlined below including only the German and Japanese managers (N = 15). This exclusion did not alter our findings.

Association of formal position with influence (Research Question 4a). A model with function was not significant. Model 1 (Table 3) is a look at the amount of variance explained by level. With tenure controlled, level explained 41 percent of the variance in influence ratings and was significant at the .05 level. Since the highest management level was coded 3 and the lowest 1, the coefficient (.58) is consistent with a prediction that higher formal level is associated with higher influence. Hence, a positive answer to Research Question 4a is sup- ported for level but not for function.

Nationality with influence (Research Question 4b). Nationality was not a significant predictor of influence, and no model is reported. The tentative answer to Research Question 4b is negative.

Orientation to local norms (Research Question 4c). A model with adaptation (not reported) was not significant. Table 3 also summarizes regression results for prefers consensus and decisions, both of which resulted in significant models. Prefers con- sensus (model 2) had an adjusted multiple squared correlation (R2) of .26, a value lower than that for decisions (model 3), which was .53. Moreover, pre- fers consensus was only marginally significant (p < .10), and decisions was significant (p < .01). Thus, the results support a positive answer to Research Question 4c, but the results for decisions were stronger than those for prefers consensus.

Centrality and influence (Research Question 4d) and the relative explanatory power of centrality and other factors (Research Question 4e). Examin-

TABLE 2 Means, Standard Deviations, and Correlationsa

Variable Mean s.d. 1 2 3 4 5 6 7 8 9 10

1. Influence 4.81 1.23 2. Level 2.47 0.74 .43 3. Function 0.47 0.52 -.17 .28 4. Nationality 0.67 0.49 -.23 .52* .38 5. Adapt 4.53 1.89 .19 -.56* -.13 -.50* 6. Prefers consensus 5.00 1.51 .50* -.37 .25 -.39 .48* 7. Decisions 5.20 1.08 .63* -.48t -.03 -.43 .45* .83*** 8. Advice centrality 43.55 20.91 .76*** -.33 .20 -.20 .06 .67** .71** 9. Task-related centrality 98.22 28.78 .55* .16 .09 .18 -.34 .31 .20 .60*

10. Private centrality 28.44 18.93 .53* -.13 -.09 -.65*** .19 .51t .54* .43 .33 11. Tenure 45.00 12.24 .03 -.05 .30 -.08 .03 .03 -.05 -.01 -.24 .02

aN= 15. tp < .10 * p < .05

** p < .01 *** p < .001

April 198

Page 10: National Culture, Networks, And Individual Influence in a Multinational Management Team

Salk and Brannen

TABLE 3 Results of Regression Analyses for Individual

Influencea

Variable Model 1 Model 2 Model 3 Model 4

Level .58 Prefers .46*

consensus Decisions .66** Advice centrality .70** Tenure .40+ .34 .43* .20

R2 .50 .37 .60 .61 Adjusted R2 .41 .26 .53 .54

Probability .02 .06 .00 .00

a N = 15. All coefficients are standardized. tp < .10 * p < .05

** p < .01

ing the equations with centrality regressed on influ- ence, we saw that only the regression with advice centrality (model 4) was strongly significant (adjusted R2 = .54, p - .004). The models with task-related and private centrality (not reported) were only moder- ately significant. Thus, the data support a positive response to Research Question 4d only for advice centrality. Comparing the previously reported regres- sions, plus examining the correlations of the explan- atory variables with influence from Table 2, gives a largely convergent result for Research Question 4e, which asks if network centrality will be a stronger determinant of influence than the other variables. The correlation of advice centrality with influence is by far the most significant (.76, p < .001), and the regression model with this variable, in which we controlled for tenure, has the most explained vari- ance. However, it must be noted that the differences in variance explained between the regression with advice centrality and that with decisions (model 3, Table 3) are not significant. Significant correlations among explanatory variables (decisions with prefer- ence for consensus, advice centrality, and level), plus the small number of observations, created a barrier to interpreting a full model combining these variables.

Hence, although our results did not allow us to reject Research Question 4e, it is noteworthy that a positive answer is true only for advice centrality. The regression results also show strong predictive power for decisions and level. This finding is corroborated by the correlation of decisions with influence, though the correlation for level just misses the cutoff for a significant correlation with influence. These results are consistent with the prediction that all three fac- tors-formal position, orientation to local norms, and

DISCUSSION

National Culture's Impact

Our results reveal an apparent conundrum. On the one hand, pursuing Research Questions 1, 2, and 4b, we found high reachability, no systematic differences between national subgroups on the vast majority of indicators, and no significance for na- tionality as an explanatory variable for influence. These findings converge to portray an integrated team that has avoided or overcome the legacy of having two very different national cultures repre- sented among its members. However, results on the Gibbs-Martin index and the block models assessing the advice and private networks, which we used to address Research Questions 1 and 3, point to a general pattern of in-group preference and strik- ingly different Japanese and German advice-seek- ing patterns. Taken together, what do our findings say about culture's impact on individual influence in this team?

First, our findings should be interpreted in the context of this particular joint venture, which is a high-performance team context. Informants from both inside and outside the IJV who knew it well felt that it was an effective team and organization. This view received corroboration from our qual- itative observations made seven months before administering the network questionnaire and from questions asked about IJV performance in that questionnaire. The German mean rating of IJV performance (1, poor, to 7, very high) was 5.7 (s.d. = .48), and the Japanese mean rating was 6.2 (s.d. = .45).

That said, we believe our findings indicate that culture is important but that its role is far more complex than past research and theory suggest. The few indicators examined for Research Question 2 that did reveal differences between the German and Japanese team members, the findings on advice- seeking patterns (Research Question 3), and some of the correlations between indicators predicting influence and nationality, all help to unravel this conundrum. The far greater propensity of the Jap- anese to form socioemotional bonds with other managers-particularly other Japanese people, but also Germans-and the different advice-seeking patterns were consistent with predictions derived from the cross-national literature about Germans and Japanese (Hall, 1983; Hampden-Turner & Trompenaars, 1989; Hofstede, 1984). Moreover, the correlations we found between nationality and a belief that decision making was fast and efficient (-.39) and between nationality and a preference for consensus (-.43) were relatively high (but both

centrality-contribute to explaining influence.

199 2000

were higher for the Japanese team members), and

Page 11: National Culture, Networks, And Individual Influence in a Multinational Management Team

Academy of Management Journal

they narrowly missed the p < .10 cutoff for signif- icance. Higher average reported adaptation by the Japanese suggests that the location of the IJV in Germany posed a greater challenge for them, though we could not determine whether this differ- ence was due to their being expatriates or to chal- lenges internal to the IJV.

Nevertheless, there was enough variance in indi- vidual responses to suggest that it is the volition to accept and adapt to local, emergent norms-a stretch for both groups-rather than national sub- group-based preferences, that contributes to being influential. The managers who were more influen- tial tended to be those who had adopted more flex- ible interpretations of "fast and efficient" decision making in this particular context. Analyses for Re- search Question 1 suggest that the subgroup an- chors are still present.

A similar interpretation can be made for advice centrality, the most potent predictor of influence. Although the block models for the advice net- work displayed subgroup-specific general pat- terns, there were no significant differences in average advice centrality between subgroups, and individual ratings within subgroups varied sub- stantially. The two managers with the highest advice centrality were in the areas most closely connected to the technological core of the orga- nization and in the third and second levels of the hierarchy: the production manager, a German, and the R&D manager, a Japanese. Both of these managers were seen as highly competent (exper- tise criterion), and both were socially well con- nected; indeed, this German manager had more private nominations from both groups than other German managers. Though our attempt to look at this pattern by making function a dummy vari- able with 1 equal to the technology core had nonsignificant results, we cannot rule out the idea that two different logics of choice-exper- tise-based for Germans and socioemotional for Japanese-played roles in individual advice centrality.

Determinants of Influence

An unexpected result was that only advice central- ity showed a significant association with influence and that it had the highest values of any explanatory variable. We have already noted that the two manag- ers with the highest advice centrality ratings were in the technical core of the organization. The codirectors had only moderately high ratings for this type of centrality. This suggests that centrality in the advice network was related to criticality (having expertise

ing the organization). It is possible to have an impor- tant formal role and to be central in task-related com- munication or in private interactions and yet be seen as less critical in helping an organization respond to technological and other sources of uncertainty. Thus, the codirectors appeared to play a highly central role in terms of general flows of task-related information and social cohesion (as indicated by the high central- ity in private networks of the Japanese codirector), but at the same time they played a less critical role in managing the most problematic contingencies, which were related to the production technology.

We already posited that nationality was not sig- nificant because the dummy variable approach can- not reveal the more subtle patterns and relation- ships outlined above. Consistent with past research, position and orientation to local norm variables were significant. We can tentatively con- clude that all three sets of factors-formal position, individual orientations, and centrality-contribute to influence. This view is consistent with views that network centrality is a conceptually distinct predictor of influence (Ibarra, 1993).

Implications and Limitations

Future research. Network methods and concepts appear to be a fruitful way of enriching understand- ing of culture's roles in multinational teams and or- ganizations. Our research reveals the multifaceted impact of national origins on the IJV team we studied. It further suggests that wide gaps between the cul- tures of team members do not doom a team or orga- nization to suffer poor performance or other develop- mental pathologies. The variance in the orientation and influence scores found within the German and Japanese subgroups of the IJV team indicate that there is danger in assuming that, a priori, members of a small face-to-face group embody the values and norms imputed to the large populations from which they came. Our results therefore suggest the desirabil- ity of collecting individual responses about orienta- tions toward both national cultural norms and values and emergent local norms.

The data indicate that basic differences can per- sist, even in teams that lack the conflict and per- formance problems noted by Lane and Beamish (1990) and Salk (1996). Our data suggest that dif- ferences themselves do not cause problems; rather, it is how a team's context and individual team members' orientations to local (team) norms chan- nel these differences. This observation, along with our results concerning advice centrality, support Doz's (1996) contention that the context and pro- cess dimensions of IJV teams deserve more atten-

closely associated with key contingencies confront-

200 April

tion. Consistent with past work on sources of

Page 12: National Culture, Networks, And Individual Influence in a Multinational Management Team

Salk and Brannen

power and influence (Kotter, 1985; Pfeffer & Salan- cik, 1978), our results indicate that particular un- certainties and contingencies confronted by an or- ganization or group shape which individual attributes, like expertise, affect centrality and influ- ence. Moreover, it suggests that the types of cen- trality that will be most closely associated with influence can vary across organizations and, poten- tially, over time as well. This idea might be most important for young groups and organizations and those in dynamic environments. Although we would expect that our general finding that there are multiple causes of influence will be replicated in studying other teams, we believe that some of the present findings concerning advice centrality stem from the particular contingencies surrounding technology and production that this team had to deal with. In other settings, centrality in other net- works could be more important.

Many network studies have been conducted in single organizations in uninational settings; it is therefore not surprising that the role of historical context and cultural origins has not received much attention. Though our findings concerning the de- terminants of influence parallel those of Ibarra (1993), more studies need to be conducted in non- U.S. and multicultural settings to confirm the gen- eralizability of social network theory pertaining to influence. Without our intimate knowledge of the historical context of the IJV we studied, we would find our results concerning national culture and advice centrality difficult to explain; this observa- tion suggests that the underspecification of context is a lacuna in the networks literature that could be redressed by our approach of combining partici- pant-observation research with network question- naire instruments.

Practice. Though more research is needed to es- tablish the generalizability of our findings, strong performance, shared 50-50 equity, and shared man- agement staffing creating a multicultural team are not unique to this IJV. Thus, we can draw some tentative implications for practitioners who design and manage joint venture and other multicultural teams. How to bring knowledge and managerial resources from parent organizations to bear on IJV operations is a key dilemma in forming and design- ing international joint ventures. If a goal is to create a conduit whereby parent company knowledge and expertise can influence an IJV (Lyles & Salk, 1996), our study suggests that paying careful attention when choosing which firm will staff which formal position might be a necessary, but by no means sufficient, condition. Our results further imply that trying to anticipate key contingencies for an IJV

managers with appropriate expertise and interper- sonal skills in positions related to key contingen- cies are vital means of establishing and maintaining influence. Expertise and interpersonal adaptability may be even more important conduits for influence and knowledge than formal authority.

Limitations. Stemming from a single case study of a small IJV team population, our statistical re- sults are descriptive rather than predictive. More- over, our results' generalizability to theory and their external generalizability can only be estab- lished by further research. In addition, our mea- sures of the normative orientations of members were based on one-item scales. Even though they yielded significant results, the validity and reliabil- ity of these measures are unknown.

CONCLUSION

This study is a first step toward unpacking the role played by national culture in determining in- dividual influence in multinational teams. It sug- gests why the causal logic underpinning past theo- rizing about national culture in multicultural settings has been incomplete. Cross-cultural and IJV research have tended to be bogged down in the dummy variable mind-set concerning culture. By focusing on micro processes, we have shown the limitations of this mind-set for the present case. Focusing on micro processes helps to illuminate how IJV and other multinational teams cope with their complex cultural legacies.

REFERENCES

Baker, W. 1992. The network organization in theory and practice. In N. Nohria & R. G. Eccles (Eds.), Networks and organizations: Structure, form and action: 397- 429. Boston: Harvard Business School Press.

Bivens, K. K., & Lovell, E. B. 1966. Joint ventures with foreign partners. New York: The Conference Board.

Bonacich, P. 1987. Power and centrality: A family of measures. American Journal of Sociology, 92: 1170-1182.

Borgatti, S. P., Everett, M. G., & Freeman, L. C. 1992. UCINET IV (version 1). Columbia, SC: Analytic Technologies.

Boyacigiller, N., & Adler, N. J. 1991. The parochial dino- saur: The organizational sciences in a global context. Academy of Management Review, 16: 1-32.

Brannen, M. Y. 1994. Your next boss is Japanese: Nego- tiating cultural change at a western Massachusetts paper plant. Unpublished doctoral dissertation, University of Massachusetts, Amherst.

(such as technology implementation) and placing

2000 201

Brannen, M. Y., & Salk, J. E. 2000. Partnering across

Page 13: National Culture, Networks, And Individual Influence in a Multinational Management Team

Academy of Management Journal

borders: Negotiating organizational culture in a

German-Japanese joint venture. Human Relations, 52: 451-487.

Brass, D. J., & Burkhardt, M. E. 1993. Potential power and power use: An investigation of structure and behavior. Academy of Management Journal, 36: 441-470.

Doz, Y. 1996. The evolution of cooperation in strategic alliances. Strategic Management Journal, 17: 55-83.

Erez, M., & Earley, P. C. 1993. Culture, self-identity and work. New York: Oxford University Press.

Freeman, L. C. 1979. Centrality in social networks. Con- ceptual clarification. Social Networks, 1: 215-239.

French, J. R. P., Jr., & Raven, B. 1959. The bases of social power. In D. Cartwright (Ed.), Studies in socialpower: 150-167. Ann Arbor: University of Michigan Press.

Hall, E. T. 1983. The dance of life. New York: Anchor/ Doubleday Books.

Hampden-Turner, C., & Trompenaars, F. 1993. The seven cultures of capitalism. New York: Doubleday.

Hatcher, L., & Stepanski, E. J. 1994. A step-by-step ap- proach to using the SAS system for univariate and multivariate statistics. Cary, NC: SAS Institute.

Hofstede, G. 1984. Culture's consequences. Beverly Hills, CA: Sage.

Ibarra, H. 1993. Network centrality, power, and innova- tion involvement: Determinants of technical and ad- ministrative roles. Academy of Management Jour- nal, 36: 471-501.

Killing, J. P. 1983. Strategies for joint venture success. London: Croom Helm.

Kotter, J. 1985. Power and influence. New York: Free Press.

Krackhardt, D. 1990. Assessing the political landscape: Structure, cognition and power in organizations. Administrative Science Quarterly, 35: 342-369.

Lane, H. W., & Beamish, P. W. 1990. Cross-cultural cooperative behavior in joint ventures in LDCs. Management International Review, 30 (special is- sue): 87-102.

Lewin, K. 1948. Some social-psychological differences between the United States and Germany. In K. Lewin (Ed.), Resolving social conflicts: 3-33. New York: Harper Brothers.

Lyles, M. A., & Salk, J. E. 1996. Knowledge acquisition from foreign parents in international joint ventures: An empirical investigation in the Hungarian context. Journal of International Business Studies, 27: 1-27.

Padgett, J. F., & Ansell, C. K. 1993. Robust action and the rise of the Medici, 1400-1434. American Journal of Sociology, 98: 1259-1319.

Pfeffer, J., & Salancik, G. 1978. The external control of organizations. New York: Harper & Row.

Ronen, S., & Shenkar, O. 1985. Clustering countries on

attitudinal dimensions: A review and synthesis. Academy of Management Review, 10: 435-454.

Salk, J. E. 1996. Partners and other strangers: Cultural boundaries and cross-cultural encounters in interna- tional joint venture teams. International Studies of Management and Organization, 26(4): 48-72.

Schein, E. H. 1985. Organizational culture and leader- ship (1st ed.). San Francisco: Jossey-Bass.

Schneider, S. C., & Barsoux, J-C. 1997. Managing across cultures. Hemel Hempstead, England: Prentice-Hall Europe.

Shah, P. P. 1998. Who are employees' social referents? Using a network perspective to determine referent others. Academy of Management Journal, 41: 249- 268.

Shenkar, O., & Zeira, Y. 1992. Role conflict and role ambiguity of chief executive officers in international joint ventures. Journal of International Business Studies, 23: 55-75.

Sumihara, N. 1992. A case study of structuration in a bicultural work organization: A study of a Japanese- owned and -managed corporation in the U.S.A. Un- published doctoral dissertation, New York University.

Tajfel, H., & Turner, J. C. 1986. The social identity theory of group behavior. In S. Worchel & W. G. Austin (Eds.), Psychology of intergroup relations: 7-24. Chicago: Nelson-Hall.

Triandis, H. 1989. Cross-cultural studies of individual- ism and collectivism. In J. Berman (Ed.), Nebraska symposium on motivation: 41-133. Lincoln: Univer- sity of Nebraska Press.

Turner, J. C. 1987. Rediscovering the social group. Oxford: Basil Blackwell.

Van Maanen, J., & Schein, E. H. 1979. Toward a theory of organizational socialization. In B. Staw (Ed.), Research in organizational behavior, vol. 1: 209- 264. Greenwich, CT: JAI Press.

Wasserman, S., & Faust, K. 1994. Social network anal- ysis: Methods and applications. Cambridge, En- gland: Cambridge University Press.

Jane E. Salk is a professor of strategy and management at the ESSEC Graduate School of Management in France. She holds a Ph.D. in organization studies from the Mas- sachusetts Institute of Technology. Her research, pub- lishing, teaching, and consulting are in the areas of in- ternational business and strategy and the design and implementation of cross-border alliances and mergers and acquisitions.

Mary Yoko Brannen is an associate professor of inter- national business at San Jose State University and an associate professor of executive education at the Uni- versity of Michigan Business School. She received her Ph.D. in organizational behavior, with a minor in an- thropology, from the University of Massachusetts at Amherst.

202 April