Culture and the Structure of the International Hyperlink Network

Authors

  • George A. Barnett,

    1. Department of Communication
      School of Informatics
      State University of New York at Buffalo
    Search for more papers by this author
    • George A. Barnett is Professor of Communication at the State University of New York at Buffalo. His current research focuses on the international telecommunications network and its role on social, cultural, and economic development and the process of globalization.

      Address: Department of Communication, School of Informatics, State University of New York at Buffalo, Buffalo, New York 14260 USA

  • Eunjung Sung

    1. Department of Communication
      School of Informatics
      State University of New York at Buffalo
    Search for more papers by this author
    • Eunjung Sung (Ph.D.) is a researcher in the department of Communication at the State University of New York at Buffalo. Her research interests include new media effects, diffusion of innovation, intercultural/national communication, and globalization.

      Address: Department of Communication, School of Informatics, State University of New York at Buffalo, Buffalo, New York 14260 USA


Abstract

This article explores the relationship between national culture and the structure of the international Internet linkage. World System Theory suggests that economic relations within nations are the primary organizing principles of international communication. However, recent research suggests that other factors may also impact the process. Using hyperlink data from Barnett and Park (in press), the current research examines the role of culture as an organizing mechanism of the Internet. The results indicated that national culture is significantly related to network centrality and its overall structure. The limitations of this research and the implications of the findings for globalization, a structural theory of international communication, and cultural convergence are discussed.

Introduction

Culture is a group’s collective meaning system and includes its values, attitudes, beliefs, customs, and thoughts. Intercultural communication may be defined as the exchange of information between well-defined groups with significantly different cultures. Globalization is “the process of strengthening the worldwide social relations which link distant localities in such a way that local events are shaped by circumstances at other places in the world” (Giddens, 1990, p. 64). One potential consequence of globalization is cultural homogenization due to the exchange of information among people from different cultural groups.

Traditionally, World System Theory has ignored the exchange of information among nations. Still, the literature from World System Theory (Chase-Dunn & Grimes, 1995; Wallerstein, 1976) on the antecedent determinants of international interaction suggests that the economic relations among nations represent the primary organizing mechanism of international communication. However, recent research (Barnett, 1999, 2001, 2002; Barnett & Choi, 1995; Galtung, 1993; Huntington, 1996) indicates that cultural factors, such as language and religion, play a significant role in the process.

This article examines the relationship between culture and the international Internet as expressed by the pattern of hyperlinks among nations. One attribute of the network composed of the nations of the world linked together by the Internet is centrality. It may be defined as the number of links or the social distance required to reach all the other components in a network. Centrality is of particular significance because World Systems Theory (Barnett, Jacobson, Choi, & Sun-Miller, 1996; Chase-Dunn & Grimes, 1995; Wallerstein, 1976) argues that international interaction is organized as a center to periphery structure. Therefore, it may be worthwhile to examine how national culture is related to international Internet flows and their role in the construction of the structure of intercultural communication.

Support for the notion that Internet behavior varies with national culture and level of economic development can be found in Wellman and Haythornthwaite (2002), who report studies from the United States, Canada, Great Britain (Anderson & Tracey, 2002), Germany (Wagner, Pischner, & Haisken-DeNew, 2002), and Japan (Miyata, 2002). Chen, Boase, and Wellman (2002), using data from a National Geographic Web survey, compare how people in different parts of the world use the Internet. They found that in spite of regional differences in the characteristics of users, the Internet is used in similar ways worldwide. Frequent users have a more positive sense of online community.

Theoretical background

Culture and Intercultural Communication

Culture consists of habits and tendencies to act in certain ways, but not actions themselves. Rather, it is composed of language patterns, values, beliefs, customs, and thought patterns. Goodenough (1964, p. 36) defines culture, not as things or behavior, but rather as “the forms of things that people have in mind, their models for perceiving, relating, and otherwise interpreting them.”Geertz (1973) treats culture as an ordered system of meanings and symbols, in which social interaction takes place and develops.

Culture is also a socially shared activity, and therefore a property of a group rather than an individual (Nieberg, 1973). It is normative and may best be represented as a measure of central tendency of the group mind (Durkheim, 1938). It does not derive from the internal conditions of individuals, but rather from society’s social conventions. Durkheim (1953, pp. 25-26) calls these shared cognitions “collective representations.”

Collective representations do not derive from individual minds, but from the association of minds. That is, collective representations are formed during the process of social interaction. Without general agreement about the meaning of symbols and other communication rules, social interaction would be impossible. As members of social groups communicate, they negotiate the shared meanings of symbols. As a result, culture is external to the individual. Thus, in order to understand culture, one must take the aggregate into consideration. Put simply, culture can be described as the way of life of a people (Rosman & Rubel, 1995).

Consistent with the notion that culture is the set of shared collective cognitions, Hofstede (1991) defines culture as “the collective programming of the mind which distinguishes the members of one human group from another” (p. 25). He also emphasizes that culture is not a property of individuals, but of groups. Hofstede (1980) suggests that the relevant dimensions of culture should be identified and investigated when conducting international research. To examine national culture, Hofstede (1980) surveyed the values and perceptions in 53 countries and three multi-country regions: Arabia, West Africa, and East Africa. His data were collected from employee attitude surveys undertaken between 1967 and 1973 within IBM. Based on statistical analysis, he suggested that national cultures may be differentiated along four dimensions: power distance, collectivism vs. individualism, femininity vs. masculinity, and uncertainty avoidance.

Power distance (PDI) is “the extent to which the less powerful members of institutions and organizations within a country expect and accept that power is distributed unequally” (1991, p. 28). High power-distance societies are more autocratic. Low power-distance societies value equality, with a preference toward democratic processes (Hofstede, 1980, p.98).

Individualism (IND) pertains “to societies in which the ties between individuals are loose: [E]veryone is expected to look after himself or herself and his or her immediate family” (1991, p. 51). Collectivism“as its opposite pertains to societies in which people from birth onwards are integrated into strong, cohesive ingroups, which throughout people’s lifetime continue to protect them in exchange for unquestioning loyalty” (p. 51). In societies high in individualism, people look after their own interests, and value their independence. Societies low in individualism support group values and beliefs and seek collective interests (Hofstede, 1980, p. 214).

Masculinity (MAS) pertains “to societies in which social gender roles are clearly distinct. Femininity pertains to societies in which social gender roles overlap” (1991, pp. 82–83). Societies with high masculinity tend to admire qualities such as ambitiousness, achievement, money, performance, and assertiveness. In contrast, societies low in masculinity emphasize people, quality of life, helping others, preserving environment, and not drawing attention to oneself (Hofstede, 1980, p. 261).

Uncertainty avoidance (UAI) is a measure of the degree to which a given culture adapts to changes and copes with uncertainty and ambiguity. Cultures with high uncertainty avoidance tend to have a low tolerance for uncertainty, so these societies create rules and regulations. Cultures with low uncertainty avoidance tend to be less rule-oriented (Hofstede, 1980, p. 153).

Recently, Hofstede (1994) added a fifth dimension, long-term orientation, to differentiate cultures. This dimension reflects the extent to which a society has a pragmatic and future-oriented perspective rather than a conventional, historic, or short-term point of view. A culture with high long-term orientation values perseverance and thrift, prioritizes general purposes over individual interests, and orders relationships by status. A culture with a short-term orientation, on the other hand, focuses on quick results, and honors tradition, personal standpoints, social obligations, and people’s need to preserve “face.”

Many scholars have sought to validate Hofstede’s dimensions and to explore their theoretical and practical contributions. The dimensions have been used as a framework for cross-cultural inference and generalization (Au, 1999). Applications of the dimensions have been studied to examine conflict and negotiation (Lee & Rogan, 1991; Ohbuchi & Takahasji, 1994), compliance gaining and influence strategies (Sanborn, 1993), managerial decision-making (Vitell, Nwachuku, & Barnes, 1993), job and communication satisfaction (Bochner & Hesketh, 1994), and persuasion (Aaker & Maheswaran, 1997).

Maitland and Bauer (2001) analyzed global Internet diffusion using national level cultural variables. Their analysis demonstrated the feasibility of using quantitative measures of national cultural variables in a multivariate global study. The results indicated that English language use was a significant indicator in determining when countries first adopt the Internet, and that there is a positive correlation between both gender equality (MAS) and uncertainty avoidance (UAI) and rate of adoption. Thus, Maitland and Bauer’s findings suggest that culture may be one of the underlying factors that determine the structure of communication via the Internet.

A structural model of intercultural communication

Intercultural communication is the exchange of “cultural” information between two groups of people with significantly different cultures.1 While this definition is clearly circular, it can be clarified by specifying the meaning of its critical concepts. In other words, intercultural communication focuses on the exchange of information between two or more social systems embedded in a common environment. Communication results in the reduction of uncertainty about the future behavior of the other system through an increase in understanding of the other social group.

In the past, scholars have limited its study to the individual level. However, intercultural communication occurs on many levels (Smith, 1999), including via mediated communication. International organizations working throughout the world also link disconnected cultural groups, helping their members to understand the similarities and differences among groups. Intercultural communication is thus the exchange of information among well-defined groups with significantly different cultures.

To help understand the impact of culture on international communication, one may adopt the structural model of communication (Barnett & Lee, 2002), displayed in Figure 1. This model represents the process of intercultural/international communication as a sociogram or a communication network composed of two interacting groups (nations), each with its own culture. Individuals or other information sources (the media and other organizations) are represented as circles, and the communication flows as lines. Arrows indicate the direction of information flow. This system is composed of two groups, A and B, with porous boundaries. Generally, communication within the groups is relatively dense while communication between the groups is sparse (Yum, 1988).

Figure 1.

Structural model of intercultural communication.

Intercultural communication concerns the linkages between Groups A and B that involve individuals a, b, and c. These links also include the mass media (Korzenny, Ting-Toomey, & Schiff, 1992; Ware & Dupagne, 1994), because information that reduces uncertainty about groups A and B is communicated via the mass media, either print (Kim & Barnett, 1996) or electronic (Varis, 1984).

Also connecting the groups are international organizations that are not part of either group A or B, but rather part of global society transcending any single cultural group (Boli & Thomas, 1997; Meyer, Boli, Thomas, & Ramirez, 1997). These may be such organizations as the United Nations or the World Bank, whose members are the nations of the world (IGOs), nongovernmental issue-based organizations such as Amnesty International and Greenpeace (INGOs) (Boli & Thomas, 1997; Jacobson, 1979), or transnational corporations (Monge & Fulk, 1999; Walters, 1995). These organizations bring people from different nations together in a common forum.

Historically, linkages among different cultural groups have increased, resulting in globalization—the process of strengthening worldwide social relations that links distant localities in such a way that local events are shaped by circumstances at other places in the world (Giddens, 1990). That is, events occurring at one place reduce the uncertainty of the future behavior of groups at another location. The increase in transborder communication has led to the rapid global diffusion of values, ideas, opinions, and technologies, i.e., the underlying components of culture. Transborder communication has opened cultural boundaries and created a global community with an increasingly homogenous culture, particularly regarding political, economic, educational, and scientific activities (Beyer, 1994; Robertson, 1992).

Giddens (1990) argues that globalization is an inherent part of modernization. One consequence of modernization is an increase in time-space distanciation that renders physical distance less of a barrier to intergroup communication. This increase is due in part to innovations in transportation and telecommunications such as the Internet. Globalization stretches the boundaries of social interaction such that the connections among different social contexts or nations become networked across the globe. Thus, the communication between the two groups presented in Figure 1 may be generalized to all the separate nations of the world. The mass media and other communication technologies compress time and space, becoming a catalyst for globalization (Giddens, 1990; Robertson, 1990). As a result, McLuhan’s (1962) notion of the global village is becoming a reality.

Various forms of the Structural Model, also known as the Network Model, have been used to investigate intercultural communication (Smith, 1999; Weimann, 1989; Yum, 1984, 1988), intergroup communication (Kim, 1986), and international communication (Barnett, 1999, 2001).2 This research has recently been reviewed in depth by Barnett and Lee (2002). They argued that intercultural communication may be analyzed by using network analysis.

The present article explores the relationship between national culture and the structure of the Internet, operationalized as the international hyperlink network. Specifically, it examines the relationship between national culture measured with Hofstede’s dimensions and the centrality and the overall structure of the Internet flow using network analysis.

Based on the theoretical background discussed above, the following research questions are proposed:

RQ 1: What is the relationship between the Internet hyperlink network and Hofstede’s dimensions of national culture? Specifically, what is the relationship between network centrality and national culture? And, what is the relationship between the overall structure of the network and the dimensions of national culture?

RQ 2: In light of World System Theory, controlling for Gross Domestic Product, are Hofstede’s dimensions of culture significant predictors of the centrality of Internet hyperlink networks?

Barnett and Sung (2003) previously examined the relationship between Hofstede’s dimensions of culture and the structure of the international hyperlink network. Using data from 1998 for the links among the OECD countries, they found a significant correlation between individualism and network centrality. The correlation remained significant after controlling for Gross Domestic Product (GDP). Also, the overall structure of the hyperlink network was significantly related to individualism (dimension 1), masculinity (dimension 2), and power distance (dimension 3). These findings are broadly consistent with Maitland and Bauer’s (2001) findings that certain dimensions of culture impact Internet behavior.

Barnett and Sung (2003) have been criticized for their sample, which contained only 24 economically developed nations. The present research replicates the earlier research with more recent data (2003) collected on a broader sample (N = 47) including key Internet-using nations that are not OECD members, such as China, India, Russia, South Africa, Israel, Brazil, and Argentina.

Methods

The Data

National Culture

Hofstede (1991) argues that conducting cross-cultural comparisons within a single transnational organization represents an ideal situation for identifying differences in national value systems because the subjects are similar on all attributes except nationality. Through statistical analysis he found that the sample differed from country to country on four dimensions: social inequality, including the relation to authority (power distance); the relation between the individual and the group (collectivism vs. individualism); concepts of masculinity and femininity (femininity vs. masculinity); and ways of dealing with uncertainty (uncertainty avoidance). For the present study, National Culture was operationalized using Hofstede’s (1991) measures of the four dimensions: 1) collectivism vs. individualism, 2) femininity vs. masculinity, 3) uncertainty avoidance, and 4) power distance.

Communication Structure

The structure of the international Internet for 2003 (see Barnett & Park, in press) is operationalized as a communication network, as suggested by Barnett and Lee (2002). The data consist of international hyperlinks as reported by Barnett and Park (in press). The numbers of bilateral interdomain hyperlinks among nations were obtained from a search using Alta Vista. The number of interdomain hyperlinks embedded in Web sites between all TLDs (top level domains, such as .ca for Canada) of 47 nations including all OECD member countries except Poland, and six gTLDs (generic top level domains, .com, .net, .edu, .mil, .org and .gov), were gathered on January 30, 2003. Together, these TLDs represent approximately 98% of Internet traffic (Internet Software Consortium, 2002).

Data collection was accomplished through a script written in Python. The search algorithm was simply:

domain: .xx AND link: .yy.

For example, the command domain: .ca AND link: .uk resulted in the number of hyperlinks from Canadian Web sites that had links to websites in the UK. The pattern of directional hyperlinks of more than 356 million links was examined.

There has been considerable criticism of the use of Alta Vista for the study of the World Wide Web. It provides only a limited coverage of the Web (Bar-Ilan, 2001), searching only 550 million out of 2.12 billion Web pages (Barabási, 2002). These pages are primarily in English (Thelwall, Tang, & Price, 2003). Rousseau (1999) and Thelwall (2000) reported that the search engine has uneven coverage of Web pages and provides irregular results, although more recently Thelwall (2001) reports that Alta Vista has become more stable. Snyder and Rosenbaum (1999) also found inconsistencies in the results of Alta Vista searches. Thus, there may be systematic error in tracing hyperlinks using AltaVista that may significantly affect the results; it is very difficult to estimate this error. Moreover, the results are of questionable reliability due to Alta Vista’s instability. However, these problems may be due in part to the dynamic nature of the web (Leydesdorff & Curran, 2000).

There is consensus that, “we would recommend not using Alta Vista for informetric research on the web, unless one needs a unique feature of this particular search engine” (Rousseau, 1999, p. 8). The present study employs a unique feature of Alta Vista: its ability to identify informational relationships between a diverse range of nodes such as countries (Ciolek, 2001). Alta Vista is the only search engine that traces incoming and outgoing links between Web sites. Further, Ingwersen (1998) is confident that national searches with Alta Vista are reliable. Since the data are aggregated to the national level and this search does not look for particular Web pages, the problems of uneven coverage and the lack of stability are less severe. Finally, no systematic bias of the procedure is apparent.

Because no single TLD totally represent the U.S., .edu, .mil, .us, and .gov were combined to designate the U.S. (.usa). The other gTLDs, .com, .org, .int and .net, were excluded from this grouping because access to these gTLDs is not exclusively American. Since this article focuses on international hyperlinks and these gTLDs do not represent nations, they were excluded from the analysis.

Barnett and Park (in press) report the reliability of the hyperlink data based on two searches conducted eight weeks apart. The data sets correlated .624, indicating that the hyperlink data are only somewhat reliable. An examination of the discrepancies between the measurements revealed that Indonesia’s pattern of linkage differed the most. This is consistent with Smith (1999), who found that Indonesian data were distorted due to the retrieval of noise pages. The domain .id contained many sites other than Indonesia’s. The reliability with Indonesia removed was .785.3

To determine the validity of the search procedures, Barnett and Park (in press) examined the first 10 identified sites of 15 randomly selected TDLs (a total of 150 sites) for accuracy. This analysis showed that 93.3% of the Web sites identified were correctly categorized. Only 15 sites did not have the correct TLD. The least accurate TLD was Indonesia (.id), where only seven of the ten sites were Indonesian. Smith (1999) also found this problem with Indonesia. One possible reason for this was that .id is also the lower level domain name for the state of Idaho. Alta Vista searches may result in false positives in those instances where the top level and a lower level domain share the same label. This is especially problematic for certain states (.ca for Canada and California, .de for Germany and Delaware, .il for Israel and Illinois, .in for India and Indiana, .co for Columbia and Colorado, and .ar for Argentina and Arkansas) or when nations use the generic TLD names to designate those functions in their country (e.g., .edu.au to indicate an educational host in Australia).

Network Analysis

Communication structure may be examined through network analysis. Network analysis is a set of research procedures for identifying structures in social systems based on the relations among the system’s components rather than the attributes of individuals (Rogers & Kincaid, 1981). The method may be generalized to describe the patterns of communication among nations. This article uses the descriptions of the relations among nation states based on the frequency of communication mediated through Internet hyperlinks.

The basic network data set is an N x N matrix S, where N equals the number of nodes in the analysis. A node is the unit of analysis. It may be an individual or higher level component, such as an organization or a nation out of which the system is composed. Each cell, sij, indicates the strength of the relationship between nodes i and j. In communication research, this relationship is generally the frequency of communication among the nodes. The frequency may be restricted to a particular topic, or communication channel (the Internet). S is symmetrical (sij= sji) when one is not concerned with directionality. In those instances, when the source and receiver of the information are differentiated, S is asymmetrical (sij≠ sji).

Given its form, a number of different mathematical or statistical methods may be applied to S to describe the structure of the network. In this article, three measures of structure are employed. One is a measure of centrality. Centrality is the mean number of links or the social distance (the inverse of the frequency of communication) required to reach all other countries in a network, such that the lower the value the more central the nation. Three indicators of centrality used in this research are indegree, outdegree, and Bonacich’s eigenvector measure (Bonacich, 1972). The Bonacich measure is appropriate in those instances where the network is completely interconnected and the strengths of the links given in real numbers. To determine the relationship between centrality and culture, the measures of centrality will be correlated with Hofstede’s dimensions of national culture.

Centrality is of particular significance because World Systems Theory (Barnett et al., 1996; Chase-Dunn & Grimes, 1995; Wallerstein, 1976) argues that international interaction is structured along a center to periphery dimension. Peripheral societies specialize in the production and export of labor-intensive, low-wage, low-technology goods desired by more central nations. In return, the core produces capital-intensive, high-wage, high-technology goods to export to the periphery.

Traditionally, World System Theory has ignored the exchange of information among nations. It has only recently been discussed in these terms (Barnett et al., 1996; Chase-Dunn & Hall, 1994). The theory argues that economics is an antecedent determinant of international interaction and that economic relations are the primary organizing principles of international communication. Thus, according to World Systems Theory, centrality should be more strongly related to economics (GDP per capita) than to national culture.

Multidimensional scaling was employed to describe the overall of structure of the network.4 It describes the underlying structure of the international system based upon the patterns of hyperlinks among nations. Barnett (2002) and Barnett et al. (2001) used multidimensional scaling to examine the international telecommunications network for 1999 and the Internet for 1998. The MDS of the Barnett and Park data was obtained using the metric multidimensional scaling algorithm from UCINET 6 (Borgatti, Everett, & Freeman, 2002). This resulted in four dimensions for the Internet, accounting for 42% of the variance in the network. The individual countries’ loadings on these dimensions were correlated with Hofstede’s four dimensions of national culture to determine the relation between culture and international communication.

To determine the overall relation between a nation’s GDP and national culture, and the centrality of Internet structure, multiple regression analysis was performed.

Results

Hyperlink Network

The 2003 hyperlink network is similar to that of 1998 (Barnett et al., 2001). The network is completely interconnected and therefore has a density of 1.0. Overall, according to the Bonacich measure, the U.S. is the most central country, followed by Australia, the U.K., China, Japan, Canada, and Germany. China emerged as a central node in international hyperlinks. Most peripheral in the network are Uruguay, Luxemburg, U.A.E., Thailand, Slovakia, and Romania. When the direction of link is considered, the United States is the most central in terms of in-degree, having the most links to its Web sites. It is followed by Indonesia, India, Italy, and France. On this indicator, Uruguay, U.A.E., and the Czech Republic are the most peripheral countries. Germany is most central in out-degree; it connects to the most Web sites outside the country. It is followed by the U.K., United States and Australia. Indonesia, U.A.E., and India are the most peripheral on this measure. The three measures of centrality for the 47 nations in the hyperlink network are presented in Table 1. Also presented are the means and standard deviations of the centrality measures, which provide an indication of the heterogeneity of these network indicators. Figures 2, 3, and 4 provide graphic representations of the distributions of the three measures of centrality.

Table 1.  International Internet centrality
 HYPERLINK
 Out-degreeIn-degreeBonacich
1jp4903376.01258347.014.080
2uk13199222.03158211.021.778
3usa12870134.015604977.0130.548
4ca3095233.03093532.012.294
5de21057460.01654674.012.020
6au5426344.02560601.032.425
7nl1727226.02519543.02.533
8fr3902700.04810245.06.407
9fi1075524.02417304.02.886
10se1300896.03158267.03.719
11it3449312.04839254.08.813
12tw2326265.01054423.05.999
13no1325859.04071733.05.246
14es1220562.02509513.04.403
15dk975896.0946539.01.401
16be1262965.02314083.02.371
17br1531697.02602113.03.879
18kr2073988.01183832.04.183
19ch2644175.02785815.04.561
20nz810539.0451855.02.664
21at1549444.03818536.06.052
22mx856764.0539578.01.842
23ru5486270.0892447.08.671
24za307440.01195380.01.518
25il558149.01071716.03.398
26ar1023911.01052573.02.532
27cz3566929.0394164.04.246
28sg516712.0476974.01.612
29hu960292.01016499.01.072
30hk454343.0401484.01.454
31gr536174.02256500.03.530
32tr691906.01727296.02.438
33pt501710.01324461.01.602
34my189410.03662336.04.497
35ie422803.03132264.04.756
36cn4039781.01400609.016.360
37is148033.03603710.05.284
38in84940.06764779.09.313
39id78134.06854038.05.058
40lu102764.0494286.00.518
41cl335857.01042648.01.803
42th153190.0745072.00.752
43ee398799.01100667.01.556
44sk359634.0971465.00.867
45ro262954.0836885.00.869
46uy316368.082910.00.244
47ae84552.0312499.00.556
Descriptive Statistics
 Mean2343971.02343971.07.970
 Std Dev3896501.52515188.019.026
Figure 2.

Centrality distribution—out-degree.

Figure 3.

Centrality distribution—in-degree.

Figure 4.

Centrality distribution—Bonacich measure.

Centrality and Culture

Table 2 presents the correlations between centrality in the Internet network and Hofstede’s dimensions of national culture. Individualism is significantly related to the centrality of hyperlink networks (indegree, r= .407, p < .01, N = 41; outdegree, r= .318, p < .05, N = 41; Bonacich, r= .357, p < .05, N = 41). All other relations between network centrality and the dimensions of culture are not significant.

Table 2.  Correlations between culture and hyperlink networks
 UAIMASPDIIND
  1. N = 42, * p < .05, ** p < .01.

UAI1.000 
MAS.1021.000 
PDI.227.0011.000 
IND−.244.037−.672**1.000
D1.166−.110.045−.103
D2−.191.025.136−.030
D3−.509**.110−.175.174
D4.224−.220−.242.338*
Indegree−.104.287−.248.407**
Outdegree−.132.187−.063.318*
Bonacich−.177.202−.136.357*

Table 2 presents the correlations among the four dimensions of culture and the four dimensions describing the structure of the Internet hyperlinks. The results indicate that a country’s location on the third dimension of the hyperlink network is significantly related to its uncertainty avoidance (UAI) (r=−.509, p < .01), and a country’s location on the fourth dimension of the Internet hyperlinks is significantly related to its individualism (IND) (r= .338, p < .05). It should be noted that the third dimension accounted for only 5.4%, and the fourth, 5.1%, of the variance in the network.

To evaluate the impact of national culture on centrality in the network independent of economics, multiple regressions were performed with GDP and individualism as the independent variables and the three measures of centrality as the dependent variable. Table 3 summarizes the results of the regression analysis. GDP is a significant indicator of the three measures of centrality of Internet hyperlink networks (indegree, R = .69, F= 17, 635, p < .000; outdegree, R = .49, F= 4.78 p < .05; Bonacich, F= 12.27, p < .000). However, when controlling for GDP, the relationship between centrality and culture defined by the individualism dimension becomes nonsignificant, with only indegree approaching the .05 level.

Table 3.  Multiple regression predicting network centrality, GDP, and individualism
Dependent VariablesIndependent VariablesRR SquareAdjusted R SquareUnstandardized CoefficientsStandardized Coefficients
BStd.ErrorBetatSig.
  1. N = 40, * p < .05, ** p < .01.

Bonacich CentralityGDP.626.392.36019.6474.825.5364.072.000**
Individualism.180.115.2071.571.124
In-degreeGDP.694.481.4544319191.9897501.06.5864.812.000**
Individualism42607.06921341.232.2431.996.053
Out-degreeGDP.448.201.1591709760.1784354.94.3292.180.036*
Individualism27885.29818650.786.2261.495.143

Conclusion and Discussion

This examination of the relationship between national culture and the structure of the Internet hyperlink network produced three major findings. The first research question asked if there was a relationship between the Internet hyperlink network structure and Hofstede’s dimensions of national culture. Only individualism is significantly related to the three measures of centrality of hyperlink networks. The more central a country is in international Internet flows, the more individualistic its culture. Using the results of multidimensional scaling to represent the overall structure of the networks, one finds that the third dimension of the Internet is significantly related to uncertainty avoidance and the fourth to individualism. These results are somewhat at odds with Maitland and Bauer (2001), who found that gender equality (MAS) and uncertainty avoidance (UAI) predicted rate of Internet adoption. The current study shows that culture, in particular the dimensions of individualism (IND) and uncertainty avoidance (UAI), are factors determining the structure of the Internet. Gender equality was not a significant predictor. However, it should be noted that Maitland and Bauer (2001) focused on adoption, and the present research focused on the structure of hyperlink flows.

The results corresponding to the second research question indicate that the total GDP is a significant indicator of all three measures of centrality of Internet hyperlink network. As suggested by World Systems Theory, the economy rather than culture is the primary determinant of the structure of international hyperlink flows. This finding is in opposition to Barnett and Sung (2003), who found a strong relationship between culture and the organization of the Internet. The difference between the two studies may be attributable to a number of factors. One factor may be the improvement of the sample to include a broader range of nations. Second, there may have been significant changes in the Internet between 1998 and 2003. Third is the lack of validity in either one of the data sets. Fourth is the lack of validity of Hofstede’s dimensions as either general cultural indicators or for the description of contemporary culture due to globalization. However, when controlling for economics, the relation between individualism and centrality merely approaches significance for indegree. Thus the results suggest that national culture is only a minor organizing factor of international Internet flow when compared to the nation’s economy.

Globalization has been a focus of intercultural communication research since the late 1980s (Hamelink, 1990), pervading academic, commercial, and political discourse. Media technologies such as satellites and optical fiber have made the world a smaller place. The word “global” generally has a positive ring. It connotes values such as one world, unity, familiarity, and sharing. Since the use of the concept “global” as a descriptive term lacks precision and relevance, it would be more useful to apply the concept globalization to a set of processes (Hamelink, 1990, p. 382).

Barnett (2001) describes the current structure of international telecommunications based on its patterns of use and how it has changed since the late 1970s, demonstrating that globalization is taking place. He discusses the implications of these patterns for the development of a universal culture, suggesting that cultural convergence results from all forms of communication, including mediated cultural information exchanged among various cultural groups. He states:

Over the last two decades, the frequency of interaction among the nations of the world has increased steadily. While there is regionalization due to physical and cultural (linguistic) barriers, today, the world consists of a single integrated network of nations centered about North America and Western Europe. One potential consequence of globalization is the cultural homogenization or the convergence of the indigenous cultures of the world into a universal culture. (p. 23)

The globalization-localization dialectic suggests that globalization involves the linking of locals to the wider world while localization incorporates trends of globalization. As a result, cultures could be developing hybrid characteristics (Lemish, 1998; Pieterse, 1995). Over time, with information exchange among people from different cultural groups, one potential consequence is cultural homogenization, the convergence of the indigenous cultures of the world into a universal culture. Thus, the dimensions of culture based on research such as Hofstede’s conducted almost forty years ago might fail to describe the patterns of communication among current Internet users. Barnett (2001) finds that the current structure of the world’s communication system is organized along the lines of regional groupings of nations, generally with similar cultures. However, no regional groupings based upon ties of cultural cohesion appeared in the sample of only 47 nations.

World System theory argues that economic relations among nations are the primary organizing principle of international communication. The present results indicate that while national culture may be inadequate to explain the complexities of international communication, it does play a significant but minor role in describing, predicting, and explaining international Internet flows.

One weakness in this research is that the network members constitute a potentially biased sample. They exclude the countries of Eastern Europe and Africa because IBM, Hofstede’s setting for the research that revealed the dimension of culture, had no operations in these areas in the 1970s. However, the data from these excluded regions suggest that Eastern Europe has strong ties to Germany (Barnett & Choi, 1995) and Russia (Barnett, 2001), and the Africa nations to their former colonial power whose official language they share (Barnett & Choi, 1995). Barnett and Sung (2003) examined the role of culture as an organizing mechanism of the Internet and international telecommunications; however, their research was criticized due to the small sample using only OECD countries. In this respect, the present article makes a stronger contribution to knowledge about the relationship between culture and the international communication.

Future research needs to examine the changes in network flows through time with a greater number of countries. Certainly, as Barnett (2001) suggests, there is a reciprocal relationship between communication structure and culture; future research should analyze this relationship using a model containing more channels of intercultural communication, such as migration, air traffic, student flow, and the exchange of cultural products (the news media). In this way, a more precise understanding of the relationship between culture and international information flows could be gained.

In sum, this article has examined the relationship between culture and the international Internet as expressed by the pattern of hyperlink networks. The results support the notion that national culture is significantly related to the structure of the Internet. Individualism is strongly related to network centrality, and both individualism and uncertainty avoidance are related to the overall structure of the network.

Notes

  • 1

    Groups’ cultures may be considered significantly different in the statistical sense. Operationally, Barnett (1988) describes the procedures for the precise measurement of culture consistent with the theoretical orientation presented in this article. Lee and Barnett (1997) provide an example of their application to determine if two cultures are significantly different.

  • 2

    For a technical introduction to network analysis see Rogers and Kincaid (1981) or Wasserman and Faust (1994). Smith (1999) provides a detailed discussion of the technical terms and their mathematical operationalizations for intercultural communication.

  • 3

    Due to the lack of independence among the cells, a QAP correlation was calculated using UCINET (Borgatti et al., 2002). QAP is analogous to the Pearson correlation. However, its significance is calculated through a model that generates a distribution for each network and the likelihood that the correlation between the two matrices (networks) occurred by chance.

  • 4

    Since the sign of the loadings in multidimensional scaling is arbitrary, the sign of the correlations among the dimensions describing the structure of the communication network and the dimensions of culture are also arbitrary. As a result, the reader need not interpret the direction of the relationship. For this article, it is sufficient to note only that a relationship exists.

About the Authors

  1. George A. Barnett is Professor of Communication at the State University of New York at Buffalo. His current research focuses on the international telecommunications network and its role on social, cultural, and economic development and the process of globalization.Address: Department of Communication, School of Informatics, State University of New York at Buffalo, Buffalo, New York 14260 USA

  2. Eunjung Sung (Ph.D.) is a researcher in the department of Communication at the State University of New York at Buffalo. Her research interests include new media effects, diffusion of innovation, intercultural/national communication, and globalization.Address: Department of Communication, School of Informatics, State University of New York at Buffalo, Buffalo, New York 14260 USA

Ancillary