Internet usage and cosmopolitanism in Europe: a multilevel analysis

ABSTRACT

Despite the transnational interconnected nature of the internet, cross-national comparisons in internet usage and their effects are still relatively scarce. Moreover, one of the core intrinsic properties that internet theorists have distinguished, the ability to increase democracy and ‘global understanding’ through its connectivity, has hardly been empirically studied. This paper examines how internet usage affects individuals’ openness to other cultures: cosmopolitanism. I analyze two manifestations of such openness: first, the cosmopolitan orientation toward other cultures in the broad sense; second, the interest in foreign cultural expressions. Using Eurobarometer data on 29 European countries, the results show that interactive internet practices are positively associated with openness to foreign culture. Buying culture online is positively related to interest in concrete expressions, but negatively to cosmopolitan orientation. Importantly, individual effects on cosmopolitan orientation are often moderated by the country people live in, whereas effects on interest in foreign expressions are more stable across Europe.

KEYWORDS:

• Internet usage
• culture
• cosmopolitanism
• cross-national
• web2.0
• downloading

Introduction

One of the most puzzling gaps in the study of the internet is the relative lack of attention to globalized usages or impacts. Despite the transnational interconnected nature of the medium, only few of the bulk of studies on the internet embark on cross-national comparisons. Some of these concern studies which analyze macro-level phenomena such as the digital divide (Guillén & Suárez, 2005; Norris, 2001; Notten, Peter, Kraaykamp, & Valkenburg, 2009); others focus on specific topics such as online safety (e.g., EU Kids Online: Livingstone, Kirwil, Ponte, & Staksrud, 2014; Lobe, Livingstone, Ólafsson, & Vodeb, 2011; Notten & Nikken, 2016) or the diffusion of internet jokes (Shifman, Levy, & Thelwall, 2014). Yet one of the core intrinsic properties that internet theorists have distinguished, the ability to increase democracy and ‘global understanding’ through its connectivity has hardly been empirically studied (Curran, Fenton, & Freedman, 2012; Zuckerman, 2013). One notable exception is the research by Norris and Inglehart (2009), who demonstrate that ‒ on a global scale ‒ internet usage is associated with lower levels of nationalism and higher adherence to tolerant modern values. An important conclusion of their study is, however, also that for further examination of how internet and other media impact democratic and cosmopolitan orientations, more specific information is needed on how people use the internet and to what kind of content they are exposed.

This paper aims to contribute to this ongoing debate by investigating how engagement in online downloading, buying, and interactivity affects individuals’ openness to other cultures. I analyze two manifestations of such openness, based on previous studies of cosmopolitanism: first, as an indication of a more ‘ordinary’ consumption-based cosmopolitanism, the interest in and consumption of foreign cultural expressions (Cappeliez & Johnston, 2013; Lamont & Aksartova, 2002; Rössel & Schroedter, 2015); second, the broader cultural cosmopolitan orientation that individuals show, which expresses a willingness to learn about the ‘Other’ as well as identifications beyond the nation state (e.g., Pichler, 2008, 2011). I examine these relationships for 29 European countries using Eurobarometer data. This approach not only enables me to examine how individual-level internet use is related to cultural openness, but also whether these effects are influenced by structural properties of the country in which one lives. Importantly, these data also provide the opportunity to rule out alternative explanations (e.g., general cultural consumption, international focus in work or one’s private life).

The relevance of this study lies in (a) investigating the cosmopolitan ‘promise’ of internet usage in-depth, (b) examining other forms of online activity than those that often have the focus, and (c) improving our understanding of how the impacts of internet usage can vary across countries. Digital media offer spaces and technologies that facilitate individuals to accumulate encounters with other cultures within the everyday lived experience (Sobré-Denton, 2015; Zuckerman, 2013). As such, I argue that this extends beyond internet usage to find information on politics or other types of hard news (e.g., Ekström & Östman, 2013). Among the most common usages are entertainment – such as purchasing music, films and other content – commerce, and social activities (Blank & Groselj, 2014; Hargittai & Walejko, 2008). And while these usages may not always influence political or civic engagement, they might just open windows to the world in the sense that they enable individuals to access entertainment products from other parts of the world, or communicate to foreign persons.

Cosmopolitanism and openness toward foreign culture

The rise of globalization and transnational flows of persons, consumer goods, and information has spurred the study of cosmopolitanism, as scholars pointed out how notions of identity and community are increasingly interwoven with both local and global conditions (Beck, 2002; Vertovec & Cohen, 2002). While the body of literature is large, numerous conceptualizations and definitions of cosmopolitanism have been put forward in the literature. Most studies emphasize the aspects of multiculturalism, trust, openness and tolerance toward others, and willingness to engage in discussion on different opinions, as key aspects of cosmopolitanism (Vertovec & Cohen, 2002). This conceptualization refers to the idea of a world community, in which persons share a certain state of mind that embodies the quest for universalism and the desire to transcend the particular (Mihelj, van Zoonen, & Vis, 2011; Pichler, 2008). As various scholars have exemplified, this manifests itself in acknowledging other viewpoints and accepting differences, in incorporating global affordances without giving up local affiliations, but not necessarily in the same way for ethical, political, and cultural contexts (Beck, 2002; Hannerz, 1992; Pichler, 2008). This understanding of cosmopolitanism with its emphasis on the universalist ideal stands somewhat in contrast to the empirical reality of who holds such views. Quite consistently cosmopolitan orientations are voiced most frequently by the higher educated and higher status strata of society (Pichler, 2008, 2011; Skrbis & Woodward, 2007). Also at the country level, inhabitants of richer countries are generally more cosmopolitan oriented (Norris & Inglehart, 2009; Pichler, 2011), although geographical exceptions have been found: in Europe, the Nordic countries, Britain and Ireland are less inclined to be positive about cosmopolitanism (Pichler, 2008).

Another line of research stresses ordinary, or contemporary, cosmopolitanism. Following the lead of Lamont and Aksartova (2002), these scholars see cosmopolitanism as part of a cultural repertoire which is enacted in everyday life (Cappeliez & Johnston, 2013; Skrbis & Woodward, 2007). In doings so, they explicitly criticize the tendency to view cosmopolitanism as either the exclusively multicultural idea (Lamont & Aksartova, 2002), as well as the experiences of the transnational elites who can afford to consume the most exquisite cultural offerings of global cities (cf. Sassen, 1991). In contrast, these authors thus consider how citizens apply cultural openness in their everyday lives, for instance by the decoupling of behavioral benefits (e.g., consumption of foreign music) and more complex attitudinal dilemmas (Skrbis & Woodward, 2007). This literature closely aligns to studies that take a more practical stance and examines taste preferences, more specifically openness to media products and culture from other parts of the world (e.g., Kuipers & De Kloet, 2009; Meuleman & Savage, 2013; Rössel & Schroedter, 2015). Again, the higher educated are more likely to engage in foreign culture (Meuleman & Savage, 2013; Rössel & Schroedter, 2015). What is more, transnational experiences and social capital positive affect preferences for foreign culture (Mau, Mewes, & Zimmmermann, 2008; Rössel & Schroedter, 2015). How internet usage affects these various forms of cosmopolitanism is still largely unknown.

Social effects of internet usage

Research into the social effects of internet usage has largely focused on interpersonal relations, communities, civic engagement, political engagement (e.g., Boulianne, 2015; Shah, Cho, Eveland, & Kwak, 2005), some of which have been collapsed under the concept of social capital (Neves, 2013). The main argument of these studies is that the low cost, high speed, personalization opportunities, and ubiquity of the internet create social affordances for establishing, maintaining and supplementing social ties. Individuals can thus extend their set of information resources by connecting to multiple networks and communities (‘networked individualism’) which may present them with different and more varied perspectives, opinions, and knowledge (Rainie & Wellman, 2012; Shah et al., 2005; Tseng & Hsieh, 2015). Exposure to alternative view points as well as interaction with others in these networks is often found to be beneficial for participation in socially relevant behaviors (e.g., in the political or civic domain) (Boulianne, 2015; Neves, 2013).

Potentially, this connectivity through information and communication technologies is global in nature. While it is true that a large part of the internet communication is with family, friends, and significant others in the immediate environment, there are also large opportunities for finding news and information from distant locations (e.g., online versions of foreign newspapers), joining special interest communities (e.g., political discussion forums, fan sites), or discovering previously unfamiliar entertainment (e.g., via YouTube). Scholars have signaled the ‘binding’ properties of the internet (in contrast to ‘bonding’ with existing ties): an awareness of and openness to novel experiences due to new social ties (Williams, 2007) or, at a more aggregated level, online word-of-mouth (Baek, 2015).

Cosmopolitan orientations or foreign cultural taste patterns tap into values that are deeply engrained in national identities and sociocultural stratifications, making them perhaps relatively more resistant to change (cf. Norris & Inglehart, 2009). Yet, the core argument from studies into social capital about interactivity triggering engagement, tolerance, and openness to alternative experiences seems also applicable here. Interpersonal talk and participation in online forums shape identity construction (e.g., Boyd, 2014). Economists and sociologists have pointed out the growing importance of social information cues online for influencing cultural preferences (Potts, 2014; Salganik & Watts, 2009). Digital media are increasingly used – often next to traditional face-to-face recommendations – to find new cultural products (Verboord, 2010; Nguyen, Dejean, & Moreau, 2014; Tepper & Hargittai, 2009). The rise of transnational social media (e.g., Facebook) and web2.0 applications (e.g., leaving reviews on Amazon, fan sites with discussion forums) has increased the possibilities of global flows (see Baek, 2015). Therefore, I expect that interactive online activities – such as interpersonal talk, and visiting websites or blogs where one is exposed to different opinions – will stimulate the interest in foreign cultural products (hypothesis 1a), and eventually also cosmopolitan orientations (hypothesis 1b).

Besides the increased possibilities to interact with other users, ‘web2.0 technologies’ have also created greater affordances for the acquisition of cultural content. The most controversial trend probably concerns the rise of MP3 files and the subsequent increase of (illegal) downloading music, films, and games (LaRose, Lai, Lange, Love, & Wu, 2005), but also the rapid rise of webstores (e.g., Amazon) implies new ways of finding mediated culture. How transformations in accessing content reshape preferences have hardly been empirically examined. One of the first and most influential predictive models is the concept of the Long Tail by Anderson (2006), who stated that the interconnectedness of the web increases the supply and creates easier access to all forms of mediated culture, annulling restrictions of space, time, and physical availability of the pre-internet age. As a result, he argues, more ‘varied’ consumption patterns will emerge in which ‘niche’ products will have larger shares, because on the web consumers have larger selections to choose from than in physical stores. In this view, downloading and buying via the internet would be positively related to diversity.

This argument resonates with economic theories which state that new technological affordances can alter the level of risk that individuals are willing to take to try out new things. Starting from the premise that any purchase of culture is risky since cultural products are experience goods which value cannot be known unless consumed (Caves, 2000), Information and Communication Technologies could lower such risk. Concretely, if mediated culture is readily available online without large costs, consumers can try out new content without being in danger of paying for an experience good that disappoints (Potts, 2014). Obviously, this last argument only concerns (free) downloading. The impact of online buying on openness to foreign culture is thus probably weaker. Also, effects on cosmopolitan orientations are in all likelihood weaker than for interest in concrete products, since downloading and buying do not imply exchanges of viewpoints, but rather an encoding/decoding model. Following these arguments, I hypothesize that downloading and online buying have a positive effect on consuming foreign culture (hypothesis 2a), and, to a lesser extent, on cosmopolitan orientation (hypothesis 2b).

Country differences

One of the aims of this paper is to advance the cross-national study of internet use. Digital divide research has demonstrated that there are considerable differences in the use of ICTs and, more specifically, internet connectivity and usage across countries (Norris & Inglehart, 2009; Notten et al., 2009). At the same time, cosmopolitanism is not equally distributed across countries because it is influenced by, among other things, affluence, connectedness to other countries, and value orientations (e.g., Pichler, 2008). Effects of internet use on cosmopolitanism are therefore expected to be dependent on the contexts in which individual users reside. Put differently, social affordances of the internet are probably shaped by the configuration of the broader social and media environment in which particular behaviors or dispositions are perhaps more stimulated than others (e.g., Lobe et al., 2011; Norris & Inglehart, 2009).

Based on the literature, I focus on four social contextual characteristics to further interpret the relationship between internet usage and openness to foreign culture. First, the diffusion of internet usage indicates how common it is to use the internet in a country, which in all likelihood stimulates associated behaviors and dispositions (as discussed in Section, ‘Cosmopolitanism and openness toward foreign culture’) throughout society. Thus, I expect that interest in foreign culture and cosmopolitan orientations are more present in societies with a higher level of internet penetration (hypotheses 3a and 3b). Second, previous studies both into cosmopolitanism and the digital divide have shown that internet use is larger in richer countries due to the available resources to invest in infrastructure and education (e.g., Notten et al., 2009). Therefore, I expect a positive effect of prosperity on interest in foreign culture (Hypothesis 4a), and cosmopolitan orientations (Hypothesis 4b).

A third factor is language. One of the possible barriers of internet becoming a transnational medium that stimulates cosmopolitanism is insufficient foreign-language proficiency (e.g., Hermeking, 2005). Due to the early dominance of English on the internet, particularly English language skills are important if one wants to engage with foreign media and culture. It thus expected that higher English proficiency leads to more interest in foreign culture (Hypothesis 5a) and more cosmopolitan orientations (Hypothesis 5b).

Finally, media scholars have shown how value orientations impact media use and its effects (e.g., Gong, Stump, & Li, 2013; Norris & Inglehart, 2009). Particularly, the work of Hofstede (2001) has been influential in this respect. Among the dimensions in value orientations he distinguished, individualism – defined as a preference for a loosely knit social framework in which individuals are expected to take care of only themselves and their immediate families – appears to be most important one in relation to the internet (Hermeking, 2005). One of the core properties that has been ascribed to the internet is a more individualized way of using media, as expressed in Rainie and Wellman’s (2012) ‘network individualism’. It is expected that higher levels of individualism lead to more interest in foreign culture (Hypothesis 6a) and more cosmopolitan orientations (Hypothesis 6b).

For all four contextual factors, it could be that they do not (only) have a direct influence on openness to other cultures, but also via the individual level. That is, individual effects of internet usage could be moderated by country characteristics: becoming stronger or weaker depending on the context. However, since I do have clear expectations beforehand, I do not formulate hypotheses.

Method

To test these hypotheses, I use data from the Eurobarometer 67.1 survey conducted in February–March 2007.¹ I analyze 26,246 respondents in 29 countries or country parts (the Eurobarometer distinguishes East Germany and Northern Ireland as separate units). These countries (parts) are the following: Germany (West and East), France, Netherlands, Belgium, Luxembourg, Austria, Denmark, Sweden, Finland, the UK, Northern Ireland, Ireland, Spain, Portugal, Italy, Greece; Cyprus, Malta, Slovenia, Poland, the Czech Republic, Slovakia, Hungary, Bulgaria, Romania, Estonia, Latvia, and Lithuania. I use the weight included in the Eurobarometer data set to prevent that small countries have a disproportionally large influence on the results. At the individual level, one selection was made: respondents who claimed they finished their school after 30 were removed.

Dependent variables

Both dependent variables were constructed using categorical Principal Component Analysis (CATPCA) to allow for a more inductive approach. CATPCA is used in particular because many variables in the Eurobarometer data set have a nominal or ordinal measurement level.

In the first CATPCA, I included indicators of consumption of and being interested in foreign culture: four items probing whether or not the respondent (1) often watches foreign tv/movies, (2) sometimes reads foreign newspapers, (3) enjoys reading foreign books in original language, and (4) enjoys eating foreign cuisine, and two items asking whether the respondent is interested in arts and culture in (5) other European countries and (6) in the rest of the world (scale from 1 to 4). Based upon the component loadings, I interpret the first dimension as the respondent’s interest in foreign culture (see Table 1). The object scores of this dimension were saved and rescaled between 0 and 10.

Table 1. Outcomes CATPCA interest in foreign culture.

	Dimension 1	Dimension 2
Often watching foreign tv/movies	.587	−.437
Sometimes reading foreign newspapers	.630	−.444
Enjoying reading foreign books in original language	.600	−.441
Enjoying eating foreign cuisine	.543	−.230
Interested in arts and culture in other European countries	.717	.646
Interested in arts and culture in the rest of the world	.720	.643
Cronbach’s Alpha	.706	.382

The second CATPCA included 14 variables probing issues of identity, feelings of belonging, and perceptions of Europe and its culture. These variables include three stand-alone questions about being interested in meeting people in Europe (scale 1 to 4), willingness to learn a new language (categories: don’t know; no, not willing; no, already sufficient command, yes, willing), and how respondents see themselves in terms of a future European citizenship (in the near future do you see yourself as ‘don’t know’, ‘[nationality] only’, ‘[nationality] and European’, ‘European and [nationality]’, and ‘European only’). Five variables measure the level of attachment to (a) ‘your city/town/village’, (b) ‘your region’, (c) ‘your country’, (d) ‘Europe’, and (e) ‘the world’ (scale 1 ‘not at all’ to 4 ‘very’).

Three variables indicate the perception of importance of culture and cultural exchanges, measured via three questions (importance for EU and feeling European, importance for developing tolerance in the world, Europe with its culture being able to contribute to greater tolerance in the world) with answering categories from 1 ‘totally disagree’ to 5 ‘totally agree’. Finally I include three variables that measure the perception of European culture. These variables are based on a factor analysis of eight statements (factor loadings higher than .40) and signal: (a) belief in a common European culture (eigenvalue 2.28), (b) skepticism of a common European culture (eigenvalue 1.31), and (c) belief that globalization will have a positive effect for European culture (eigenvalue 1.02).² The variables are computed via the mean scores on the contributing items.

The outcomes of the CATPCA show two dimensions (see Table 2): one expressing a more cosmopolitan cultural orientation, and one expressing a more local cultural orientation. Note that cosmopolitan orientation does not exclude attachment to local and national contexts, in line with cosmopolitan theory. I thus used this first dimension and label it cosmopolitan orientation. Again, the object scores were saved and transformed them into new variables running from 0 to 10 by setting the minimum score at 0, dividing by the maximum score and multiplying by 10.

Table 2. Outcome CATPCA cosmopolitan orientation.

	Dimension 1	Dimension 2
Interested in meeting people in Europe	.506	−.357
Willingness to learn a new language	.368	−.460
Future European citizenship	.411	−.381
Attachment to city/town/village	.251	.794
Attachment to region	.296	.802
Attachment to country	.349	.688
Attachment to Europe	.627	.127
Attachment to the world	.518	−.012
Cultural exchanges: feel more European	.695	−.134
Cultural exchanges: developing tolerance in the world	.671	−.170
Cultural exchanges: European contribution to more tolerance	.666	−.058
European culture: belief in common European culture	.530	.053
European culture: skepticism in common European culture	−.103	.118
European culture: belief positive effect globalization	.139	−.097
Cronbach’s Alpha	.738	.614

Independent variables at the individual level

Our primary explanatory variables concern the internet usage of individuals. There are four measurements. First, I use the question on how often someone used the internet apart from their professional activity to measure the frequency of usage (five categories ranging from ‘never’ to ‘every day’). Second, I counted how many ‘download’ activities one reported to do (from three possible activities: downloading free music, downloading free movies or TV programs, exchanging files). Third, I registered whether or not someone said to buy cultural products (such as books, CDs, theater tickets) online. Fourth, the number of ‘web2.0’ activities were counted – that is: activities which signal an interactive usage of the internet. The Eurobarometer data set contained five of such activities: creating a website, making phone calls, communicating using a webcam, visiting chat rooms/forum, and visiting blogs. All variables are rescaled between 0 and 1.

Control variables

First I take the standard social background characteristics into account. Age is measured in absolute years. Sex is coded as male = 0 and female = 1. Educational attainment is operationalized as the number of years of education. Since the Eurobarometer only asks at which age someone finished school, this was coded as follows: all respondents who state they finished school before age 11 are recoded as 0, age 11–30 are recoded between 1 and 20 years of education, and those who state they finished school after age 30 are removed. I also control for the size of the location where the respondent lives, since persons in more urban areas have been found to be more cosmopolitan (Pichler, 2008, p. 1121). In the Eurobarometer, this is measured in three categories: (a) rural area/village, (b) small/medium town, and (c) large town/city.

Finally I consider income. Since there is no direct measure of income available in the Eurobarometer, I created an index based on indirect measures. First, I multiplied the question on what ‒ in the eyes of the respondent ‒ is the lowest net monthly income that his/her household ‘would need to have in order to make ends meet’, with the perceived relative height of the actual income compared to that figure. Second, it was counted how many material goods out of a list of seven items (television, dvd player, music cd player, computer, car, apartment/house payed, apartment/house paying) the respondent owns. The third element of the index concerned the self-assessed ease at which the household of the respondent can keep up with bills and other commitments (from ‘having real financial problems’ to ‘keeping up without any difficulties’. All three variables were rescaled between 0 and 10, after which the mean score was calculated.

In order to show the impact of internet usage, it is important to account for alternative explanations of openness to foreign cultures. Two factors which should be considered are the international degree of a person’s social network and the international degree of a person’s travel and communication behavior (both private or for work). For the international social network, information in the Eurobarometer was used about having (a) family member living in another European country, (b) family member living in another non-European country, (c) friends from other European countries, and (d) friends from other non-European countries. For international travel and communication, the data set contains information on whether someone (a) has traveled abroad at least three times in last three years, (b) often communicates with people in other countries via internet or email, and (c) has a job which involves contact with people or organizations in other countries.

Furthermore, it is important to control for how interested respondents are in (popular) culture, to able to disentangle the impact of downloading cultural content from the interest in such content. To approximate cultural interest I use the questions on (a) cinema attendance, (b) visiting a concert, (c) watching a cultural program on TV or listening to such program on the radio, and (d) reading a book. These variables were added up and form the variable cultural consumption.

Independent variables at the country level

For each country, I retrieved the GDP per capita and the percentage of the population who uses internet for the year 2007 on the website of the World Bank. Language proficiency is measured by the percentage of people who speak English in the year 2005. The source of this variable is Eurobarometer data as reported in Gerhards (2014). The individualism score per country is retrieved from the website of Geert Hofstede. Finally, I created dummy variables for European regions, to control for additional locational effects: (a) Western-Central Europe, (b) Northern Europe, (c) British Isles, (d) Southern Europe, (e) Eastern Europe, and (f) Baltic countries.

Statistical analyses

For the multivariate analyses I apply hierarchical linear multilevel analysis to take the nested structure of the data into account. This is necessary to correctly estimate the standard errors at both the individual and the contextual level; using a regular OLS regression would lead to underestimation of standard errors since results for contextual variables are then multiplied by the number of observations at the individual level (Hox, 2010). I first present random intercept models in which the dependent variables are allowed to differ across countries, but the effects of independent variables are assumed to be stables (fixed effects): in models 1 and 2 individual characteristics are entered; in model 3 country characteristics are added. To assess whether the effects of internet usage differ across countries, I estimated cross-level interactions. In these latter models, the tested internet variable in case was included as random coefficient rather than as a fixed coefficient. For all analyses, I used the mixed model function in SPSS.

Results

Descriptive results

 Figures 1–3 show the mean scores for the dependent variables as well as the internet variables per country (Table A1 in the appendix gives the full numbers).³ On average, the Nordic and Eastern European countries display the highest levels of cosmopolitanism, followed by the Southern and Western-Central European countries. The Baltic states and the British Isles stay somewhat behind. In terms of interest in foreign culture, the Nordic and Western-Central European countries and the Baltic states have the highest scores. Here, the lowest scores are found in Eastern European and Southern European countries and, again, the British Isles.

Figure 1. Interest in foreign culture.

Figure 2. Cosmopolitan orientation.

Figure 3. Internet usage.

Internet usage shows the highest means in the Nordic and Western-Central European countries, and the British Isles. Engagement in web2.0 activities and downloading follows to a large extent these general trends, although the latter activity is done slightly less in the British Isles. The Baltic states show relatively high participation in both these activities. Then again, whereas buying culture online is also custom in the Nordic and Western-Central countries, it is not done at lot in the Baltic states. The British Isles come third in buying behavior.

Individual effects

Model 1 in Table 3 shows the results of a multilevel analysis predicting interest in foreign culture using only individual background characteristics. The results for the demographic variables are very much in line with previous empirical studies: people who are higher educated, younger, and are living in larger municipalities have more interest in foreign culture than their counterparts (see Meuleman & Lubbers, 2014; Pichler, 2008; Rössel & Schroedter, 2015). There is no effect of income. Women show a larger interest in foreign culture. These results are net of the impact of having an international network, international mobility, and cultural consumption, which all have large positive effects on interest in foreign culture.

Table 3. Multilevel analyses of interest in foreign culture and cosmopolitan orientation.

	Interest in foreign culture			Cosmopolitan orientation
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Individual level
Years education (0–20)	.060 (.003)***	.059 (.003)***	.059 (.003)***	.044 (.003)***	.045 (.003)***	.044 (.003)***
Female	.096 (.018)***	.105 (.018)***	.105 (.018)***	−.021 (.020)	−.025 (.020)	−.024 (.020)
Age/10	−.040 (.005)***	−.028 (.006)***	−.029 (.006)***	.025 (.006)***	.020 (.006)***	.020 (.006)***
Living in village	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
Living in small town	.057 (.022)**	.051 (.022)*	.049 (.022)*	−.080 (.023)**	−.078 (.023)**	−.081 (.023)**
Living in large town/city	.175 (.025)***	.168 (.025)***	.167 (.025)***	−.047 (.027)∼	−.046 (.027)^∼	−.045 (.027)∼
Income level (0–1)	.031 (.066)	−.022 (.067)	−.031 (.068)	.828 (.071)***	.843 (.073)***	.856 (.073)***
Intern. social network (0–1)	1.604 (.041)***	1.587 (.041)***	1.588 (.041)***	.649 (.044)***	.656 (.044)***	.660 (.044)***
Intern. travel + comm. (0–1)	1.410 (.043)***	1.339 (.046)***	1.340 (.044)***	.129 (.046)***	.169 (.047)***	.172 (.047)***
Cultural consumption (0–1)	1.969 (.042)***	1.925 (.043)***	1.927 (.043)***	1.708 (.045)***	1.727 (.046)***	1.725 (.046)***
Overall internet usage (0–1)		.058 (.031)	.058 (.031)		.013 (.033)	.013 (.033)
Internet downloading (0–1)		.028 (.047)	.030 (.047)		−.067 (.051)	−.067 (.051)
Internet buying (0/1)		.093 (.030)**	.090 (.030)**		−.090 (.032)**	−.088 (.032)**
Internet web2.0 (0–1)		.215 (.070)**	.216 (.070)**		−.067 (.075)	−.068 (.075)
Country level
Region: British Isles			−1.305 (.202)***			−.651 (.225)*
Region: Eastern Europe			−1.139 (.229)**			−.786 (.255)**
GDPpp			−.089 (.022)**			−.091 (.024)**
% people using Internet			.006 (.005)			−.001 (.005)
% people speaking English			.009 (.003)**			−.002 (.003)
Individualism score			.015 (.004)**			.014 (.004)**
Intercept	1.682 (.097)***	1.643 (.096)***	2.394 (.306)***	4.439 (.101)***	4.457 (.101)***	6.087 (.341)***
Variance country level	.127 (.044)**	.121 (.042)***	.018 (.007)**	.135 (.046)***	.131 (.045)***	.022 (.009)*
Variance individual level	1.940 (.018)***	1.937 (.018)***	1.936 (.018)***	2.263 (.021)***	2.262 (.020)***	2.262 (.020)***
Observation country level	29	29	29	29	29	29
Observations individual level	26,246	26,246	26,246	26,246	26,246	26,246
−2 restricted log-likelihood	85,418.037	85,371.693	85,336.762	89,172.961	89,157.254	89,125.604

Note: Weighted by Eurobarometer variable ‘W22 Weight Euro 27’. Maximum likelihood estimates.

***Significance: p < .001.

**Significance: p < .01.

*Significance: p < .05.

^∼Significance: p < .10.

In model 2 I add the internet variables. There is no effect of the frequency at which someone uses the internet, but I do find positive effects of two specific forms of internet usage. Buying cultural products (b = 0.093, p = .002) and using interactive online features (b = 0.215, p = .002) have an independent positive impact on interest in foreign culture. Downloading does not alter one’s interest. Again, these effects are regardless one’s international social network, international travel and communication behaviors, and cultural consumption. In other words, taking into account that people differ in their levels of culture behavior as well as their practical everyday international orientation, both buying culture via the internet and being highly interactive on the web increases one’s interest in foreign culture.

Country-level effects

Model 3 adds the country-level characteristics. Given the results of the descriptive analyses, I include dummies for the regions of the British Isles and Eastern Europe as control variables. The results indicate that there is no effect of the degree of internet penetration in a country. Apparently, internet usage only affects interest in foreign culture at the individual level. However, two macro-level factors do increase the interest in foreign culture: a higher percentage of people speaking English (b = 0.009, p = .004) and a more individualist value orientation (b = 0.015, p = .001). Once these factors are taken into account, GDP has a negative effect. Thus, being richer or being more online does not increase a country’s interest in foreign culture; it is English-language proficiency and individualism.

Table 3 also presents the results of the multilevel regression for cultural cosmopolitan orientation. Model 4 gives the estimates of the individual predictors, minus the internet activities. Europeans have stronger cultural cosmopolitan orientations to the degree that they are higher educated, older, and have higher income levels. Thus, extending the conception of being interest in foreign cultures to identity issues and more general perceptions implies different results: cosmopolitan citizens are still highly educated, but are older and richer. There is a very limited influence of internet usage, as can be seen in model 5. Only buying culture online is related to displaying a cosmopolitan orientation, but it is negatively associated (b = −0.090, p = .005). These findings suggest that for more general – and arguably more deeply embedded – value orientations, internet usage does not have a profound impact. Model 6 adds the country characteristics. Again, I control for the regions of the British Isles and Eastern Europe. Whereas these regions are clearly less cosmopolitan than the other regions, it proves difficult to explain the regional differences through the more theoretically substantive variables. Internet usage and English proficiency are not associated with cosmopolitanism. More individualist countries are more cosmopolitan, however (b = 0.014, p = .003). Again, a negative effect is found for GDP: cosmopolitanism decreases as countries get richer (b = −0.091, p = .002).

Moderating effects of country context

Finally, I analyze whether the relation between individual internet usage and openness to foreign culture is moderated by country characteristics via estimating cross-level interactions. I conduct these interactions in separate models. The baseline is model 3 (or 6) of Table 3, but now the variances are allowed to vary across countries, and I add a cross-level interaction. All significant cross-level interactions are reported in Tables 4 and 5.⁴

Table 4. Overview significant cross-level interactions for interest in foreign culture.

Individual variable	Fixed effect	Country variable	Fixed effect	Cross-level interaction
Downloading	.700 (.291)*	GDPpp	−.063 (.020)**	−.026 (.012)*
Web2.0	1.032 (.392)**	Individualism	.014 (.004)**	−.012 (.006)*

Note: Full models are identical to models 3 and 6 of Table 3, except for inclusion cross-level interaction (entered one at a time) and random slope (unstructured variance components).

***p < .001.

**p < .01.

*p < .05.

^∼p < .10.

Table 5. Overview significant cross-level interactions for cosmopolitan orientation.

Individual variable	Fixed effect	Country variable	Fixed effect	Cross-level interaction
Freq internet usage	.677 (.326)^∼	Internet penetration	.005 (.005)	−.010 (.005)^∼
Freq internet usage	.606 (.309)^∼	GDPpp	−.076 (.022)**	−.024 (.013)^∼
Freq internet usage	.382 (.187)^∼	Eng proficiency	.003 (.003)	−.007 (.004)^∼
Freq internet usage	.816 (.297)**	Individualism	.018 (.004)***	−.012 (.005)*
Downloading	.727 (.359)^∼	Individualism	.013 (.004)**	−.012 (.005)*
Web2.0	.944 (.479)^∼	Internet penetration	.002 (.003)	−.016 (.005)^∼
Web2.0	1.129 (.423)*	GDPpp	−.076 (.024)**	−.051 (.018)**
Web2.0	.532 (.254)^∼	Eng proficiency	−.0004 (.003)	−.012 (.005)*
Web2.0	1.027 (.471)*	Individualism	.015 (.004)**	−.019 (.007)**

Note: Full models are identical to models 3 and 6 of Table 3, except for inclusion cross-level interaction (entered one at a time) and random slope (unstructured variance components).

***p < .001.

**p < .01.

*p < .05.

^∼p < .10.

The most striking general result is that several variables probing individual internet usage become significant when allowing the effect to vary across countries. This particularly pertains to cosmopolitanism orientation. Thus, although the random intercept models showed little association between internet usage and openness toward foreign culture, it appears that for some individual variables the effect was suppressed. Since these variables have a different impact in different contexts, the overall impact was not visible.

More concretely, for interest in foreign cultural products I find two significant cross-level interactions. Downloading now has significant positive effect on interest in foreign culture (b = 0.700, p = .027), but in countries with a higher GDP this impact is significantly smaller than in poorer countries (b = −0.026, p = .047). In a similar vein, there is evidence that the impact of web2.0 activities – positive main effect (b = 1.033, p = .015) – is smaller in countries which are more individualized (interaction: b = −0.012, p = .049).

For cosmopolitan orientations, again, I find significant cross-level interactions for downloading and web2.0 activities (see Table 5). Downloading is positively associated with cosmopolitan orientations (b = 0.727; p = .058), but this relationship is weaker in more individualized countries. Engaging in web2.0 activities now has significant positive effects on cosmopolitan orientations – in contrast to models 2 and 3 – and for all four country-level factors this effect differs across contexts. The positive association of engaging in web2.0 activities is smaller to the extent that countries (a) are richer, (b) have higher levels of internet penetration, (c) have higher levels of English proficiency, and (d) are more individualistic.

Note that I did not find significant cross-level interactions for online buying. Remembering the relative strong effects in Table 4, this implies that these effects – positive for interest in foreign culture, negative for cosmopolitanism – are consistent across countries.

Discussion hypotheses

Based on these results, I can now discuss the hypotheses. For the explanation of showing an interest in foreign culture, I find support for H1a (positive effect web2.0 usage), and mixed results for H2a (positive effect of buying, and only a positive effect of downloading if this is allowed to vary across countries). At the country level, I find no support for H3a and H4a (no effect internet penetration; negative effect GDP), but support for H5a and H6a (positive effect English-language proficiency and individualism). It should be noted that individual effects are rarely moderated by country-level variables.

For the explanation of a broader cosmopolitan orientation, I initially did not find support for the hypotheses at the individual level (online buying has a negative effect). However, if I take into consideration that the strength of the relationship between internet usage and cosmopolitan orientation can change according to the country in which the relationship is situated, more significant results emerge. In the cross-level interactions models, I find support for H1b (positive effect web2.0 usage), and support for H2b in one instance (downloading has a positive effect if it is taken into account that its effect differs in more individualist countries). At the country level, I find no support for H3b, H4b, and H5b (no effects internet penetration and English-language skills, negative effect GDP), but support for H6b (positive effect individualism).

Conclusion and discussion

Internet is often conceived as a transnational medium due to its interconnectivity which in the past decades has spread exponentially across the globe. This global diffusion raises the question how connected individual users truly are, not just literally, but also symbolically: how open users are to foreign cultures that they can encounter via the web. The current paper examines how individual internet usages and conditions for using the internet at the country level are associated with openness to foreign culture in Europe. Drawing on literature in the field of cosmopolitanism, I distinguished between two indicators of such openness: interest in concrete foreign cultural products or expressions (as a form of ‘ordinary cosmopolitanism’) and cosmopolitan orientations (expressing broader beliefs in multiculturalism and tolerance, as well as identifications beyond the nation state).

Analyzing data on 29 European countries, I find significant associations between internet usage and openness to foreign culture. My results corroborate and extend the findings of Norris and Inglehart (2009) that using the internet is related to the sociopolitical value orientations of individuals. More concretely, Europeans who are more engaged in interactive internet practices show more interest in foreign cultural expressions (such as films, books, television), and have a larger cosmopolitan orientation (e.g., show attachment to other parts of the world, are interested in meeting other people). For concrete cultural expressions, these effects are the same across Europe; for the broader cosmopolitan orientation, these positive effects only emerge, when I take into consideration that the individual effects differ across European countries. That is, they are weaker in countries that (a) are richer, (b) have larger English proficiency, (c) have larger internet penetration, and (d) are more individualist. Thus, the impact of internet usage on cosmopolitan orientations is moderated by the context in which it is being used.

Online buying of cultural products positively affects interest in foreign cultural expressions, but negatively affects cosmopolitan orientation – irrespective of the country one comes from. This result is in line with Skrbis and Woodward (2007), who showed how the everyday expression of cosmopolitanism may differ from more deeply engrained ideas on multiculturalism and other cultures in the broad sense. Of course, the Eurobarometer data do not allow us to speculate on which foreign cultural expressions individuals like (e.g., American Hollywood films or French art house cinema). Downloading does not seem to have much effect on Europeans’ degree of openness toward foreign culture (except in two cross-level interaction models where I find a positive effect). Thus, my results do not provide much evidence that downloading is being used to explore the available cultural content and consequently contributes to a more cosmopolitan outlook.

What do the moderating effects of cultural contexts imply? On one level, they remind us that the tendency to study internet usage and its effects in single countries (often the United States or Western European countries) only generates a partial understanding of how internet is used. In that sense, it underscores – from a different, more macro-level angle than often taken – how technological and social processes are interdependent. At another level, the cross-level interactions suggest that the relationship between internet usage and social effects – theorized by many (cf. Neves, 2013) – has a point of diminishing return. As societies get richer, better connected to the web, etc., internet becomes a regular medium which is less used for exploring new experiences and more integrated in existing practices and beliefs.

Some limitations need to be mentioned. Obviously, the data come from 2007; so the internet usage measurements in all likelihood do no longer reflect the current situation. This does not devalue the results of this study, as the main purpose of this article was theoretically based: to study the relationship between online activities and openness to culture, rather than a description of internet usage in Europe. Unfortunately, more recent data on the studied concepts are – as far as I know – not available: that is, data which not only contain information on internet usage and cosmopolitanism, but also contain the necessary control variables to exclude spurious effects. Finally, one can ask whether societal and technological developments have changed the relationship under study. Arguably, the most obvious change compared to 2007 is the rise of social media. Although these are highly interactive forms of media, they also tend to be associated with ‘bonding’ properties: reaffirming existing ties (e.g., Reich, Subrahmanyam, & Espinoza, 2012; Mesch, 2015), which may have a different impact on openness to foreign culture than interactive activities outside of specific Social Networking Service platforms. Another point for future research would be to extend this study beyond Europe, since the used measures have admittedly a strong European perspective. Regardless of the specific cases to be analyzed, this article shows the relevance of a cross-national approach beyond the digital divide. Everyday internet activities affect individuals’ affinity with the global, but the extent to which this occurs differs across social contexts.

Acknowledgements

The author would like to thank Amanda Brandellero for creating the maps included in the manuscript.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes on contributor

Marc Verboord is Associate Professor in the department of Media and Communication at Erasmus University Rotterdam. His research focuses on cultural consumption patterns, cultural socialization, classification of cultural products, media use, and Internets impact on the social valuation of cultural products [email: [email protected]].

Notes

1. A search for more recent data sets that contain the right combination of variables (asking for cultural practices, Internet usage, and cosmopolitan orientations) yielded not results. While more recent Eurobarometer data do ask about Internet usage in general and, to some extent, about cosmopolitanism, the cultural angle is absent. Also other international survey programs (e.g., ISSP or WVS) do not combine the information needed to test the hypotheses.

2. Results of factor analysis are available at request.

3. The maps were made with ArcGIS, and used the Jenks natural breaks classification method (5 categories).

4. I do not report the full models because of space restrictions. They are, of course, available at request.

Appendix Table A1. Mean scores for dependent and selected independent variables per EU country and country region.

	DVs		IVs: Individual level				IVs: Country level			Individualism
	cosmo	For_cult	Internet freq	downl	Buy	Web2.0	% Internet	GDPpp	% English
Western-Central	6.24 (1.61)	3.88 (1.98)	.50 (.44)	.13 (.26)	.23 (.41)	.09 (.18)	72.3 (6.2)	27.4 (2.5)	49.5 (14.7)	69.4 (5.2)
France	6.19	3.85	.47	.14	.23	.12	66.1	26.5	34.3	71
Belgium	6.55	4.14	.53	.16	.13	.10	64.4	28.7	51.8	75
Netherlands	6.14	4.86	.76	.24	.32	.11	85.8	31.0	87.6	80
Luxembourg	6.78	6.59	.56	.16	.26	.10	78.9	65.3	66.7	60
Germany (west)	6.32	3.77	.49	.11	.23	.06	75.2	26.5	51.1	67
Germany (east)	6.22	3.53	.43	.09	.19	.05	75.2	26.5	51.1	67
Austria	5.85	3.11	.42	.14	.18	.07	69.4	30.2	54.9	55
Nordic	6.55 (1.39)	4.82 (2.19)	.70 (.39)	.17 (.30)	.35 (.48)	.12 (.21)	82.5 (1.6)	28.4 (0.9)	78.2 (10.8)	69.7 (4.7)
Denmark	6.74	5.03	.73	.20	.38	.12	85.0	29.7	84.2	74
Finland	6.43	4.26	.64	.14	.19	.09	80.8	27.3	60.4	63
Sweden	6.50	5.02	.72	.17	.42	.15	82.0	28.2	85.0	71
British Isles	5.81 (1.76)	3.50 (1.82)	.54 (.44)	.11 (.23)	.21 (.41)	.06 (.15)	74.3 (3.3)	28.3 (1.4)	99.5 (0)	87.8 (4.5)
Great Britain	5.82	3.56	.55	.11	.22	.06	75.1	27.9	99.5	89
Ireland	5.82	2.79	.44	.10	.11	.05	61.2	33.7	99.5	70
Northern Ireland	5.68	3.10	.42	.12	.19	.08	75.1	27.9	99.5	89
Southern	6.16 (1.63)	3.18 (1.55)	.33 (.41)	.09 (.22)	.04 (.20)	.05 (.14)	45.8 (7.3)	23.3 (2.0)	28.2 (8.8)	59.0 (17.4)
Italy	6.14	3.08	.34	.08	.03	.03	40.8	24.3	29.3	76
Spain	6.15	3.28	.36	.12	.06	.07	55.1	23.9	20.7	51
Portugal	6.17	3.04	.28	.06	.04	.06	42.1	17.5	27.2	27
Greece	6.23	3.16	.20	.07	.03	.03	35.9	20.8	43.9	35
Cyprus	6.64	3.68	.27	.11	.04	.06	40.8	21.9	72.2	35
Malta	6.73	5.59	.43	.11	.11	.08	46.9	18.0	95.2	59
Slovenia	6.46	4.08	.43	.17	.07	.08	56.7	20.4	56.0	27
Eastern	6.45 (1.70)	3.01 (1.66)	.33 (.42)	.10 (.24)	.08 (.27)	.06 (.15)	44.2 (10.4)	12.2 (2.9)	23.8 (4.0)	52.0 (16.4)
Poland	6.63	2.99	.36	.10	.12	.07	48.6	12.4	25.9	60
Hungary	6.81	3.36	.30	.08	.08	.06	53.3	15.3	16.2	80
Czech Republic	6.09	2.95	.42	.13	.11	.08	51.9	17.5	23.8	58
Slovakia	6.28	3.35	.34	.09	.05	.07	61.8	14.7	24.2	52
Bulgaria	6.20	2.59	.21	.07	.01	.04	33.6	8.7	15.4	30
Romania	6.26	2.98	.28	.10	.03	.06	28.3	8.8	26.6	30
Baltic states	6.01 (1.71)	3.54 (2.00)	.44 (.45)	.14 (.27)	.06 (.24)	.11 (.20)	55.3 (6.4)	13.8 (0.9)	30.8 (6.1)	62.8 (4.5)
Estonia	6.18	4.28	.55	.19	.14	.20	66.2	15.7	41.4	60
Lithuania	5.92	3.04	.38	.11	.03	.09	49.9	13.6	25.7	60
Latvia	6.07	4.07	.47	.16	.07	.10	59.2	13.1	34.1	70

Between brackets standard deviations (for sake of readability only for regions). Individual scores and regional country scores are weighted results.