Things we call ‘technologies’ are ways of building order in our world.
Langdon Winner, Reference Winner1980: 127
It is tempting to think of the Internet as unprecedented in its potential for political progress. History, however, cautions against such hubris.
Larry Diamond, Reference Diamond2010: 71
Two global phenomena have transformed human society. Over the last two decades, democracy has gained unprecedented support from citizens around the world. Democracy has become the most popular political system among the general public, not only in the wealthy parts of the world, but also in developing countries (Inglehart, Reference Inglehart2003). Because democracy can emerge and consolidate where ordinary people support it, this seemingly universal support for democracy signals a positive condition for further democratization in the developing world.
At the same time, the world has entered a revolutionary new era of information communication technology (ICT). In the early 1990s, only a few million around the world used the Internet and communicated in cyberspace. However, the proportion of the global populace using the Internet has dramatically increased to more than 2.7 billion, constituting 35% of the world population (International Telecommunication Union, 2012). It has been predicted that Internet use via computer or other electronic devices will further transform the social lives of ordinary citizens, especially young generations, as it has changed business (Chadwick, Reference Chadwick2006).
Are these two phenomena associated? Is Internet use conducive to democratic citizenship? If so, why does Internet use influence public attitudes toward democracy? Is its impact on democratic citizenship substantial or minimal? The current literature on political culture and democratization has generally focused on the economic, political, and social forces that shape public orientations toward democracy (Almond and Verba, Reference Almond and Verba1963; Bratton et al., Reference Bratton, Mattes and Gyimah-Boadi2005; Rose et al., Reference Rose, Mishler and Haerpfer1998). With few exceptions (Nisbet et al., Reference Nisbet, Stoycheff and Pearce2012), little effort has been made to unravel the role of this new communication medium in democratizing ordinary citizens in a systematic manner.
Moreover, existing studies on the political consequences of the Internet have centered on either advanced democracies or a specific country and have examined public orientations toward politics (e.g., political knowledge and civic engagement) rather than attitudes toward democracy (Boulianne, Reference Boulianne2009). Little is known about the relationship between expansion of Internet use and democratic citizenship in both developed and developing countries from the comparative perspective of democratization. Consequently, the current literature on Internet use and democratic citizenship is in an embryonic stage and requires a great deal of systemic research to explore the effect of the former on various dimensions of the latter.
This study is an effort to expand the scope of the research by improving our understanding of Internet use and political support for democracy. It has a twofold purpose. Theoretically, applying social learning theories to Internet use and democratic citizenship, it contends that individual use of the Internet has different effects on democratic attitudes in democratic and authoritarian societies (McClosky and Zaller, Reference McClosky and Zaller1984). Specifically, the utility of the Internet has both bright and dark sides, depending on different political contexts (i.e., regime type). Internet use reinforces democratic citizenship in democratic societies but not in authoritarian societies.
Empirically, the present research subjected this theoretical idea to a rigorous test, using public opinion data culled from 45 societies surveyed in the latest World Values Survey and applying multilevel modeling. The results provide systematic evidence for these theoretical expectations, which are consistent with recent studies arguing a conditional effect of the Internet on democratization (Groshek, Reference Groshek2009; Morozov, Reference Morozov2011; Nisbet et al., Reference Nisbet, Stoycheff and Pearce2012). In particular, Internet use appears to have a global reinforcement effect on democratic citizenship in that it has promoted democratic support in democratic societies but not in authoritarian societies. These results indicate the need for caution when considering the potential of the Internet to democratize ordinary citizens in authoritarian countries specifically and to promote democratic development generally.
This article proceeds in five sections. The next section reviews prior studies on Internet use and democratic citizenship and identifies two limitations that this study attempts to overcome. The second section discusses theories of social learning to explain the relationship between Internet use and citizens’ support for democracy. In this section, I advance the core argument of this study: Internet use bolsters political support for democracy in democratic societies but not in non-democratic settings. The third section specifies the research design including the World Values Survey, measurements, and an empirical model, and the fourth discusses empirical results. The final section highlights major findings of this study and discusses its contributions.
Internet and democratic citizenship: three views
The Internet is a global system of interconnected computer networks that enables billions of users worldwide to communicate and exchange information with one another. During the early 1990s, it provided college-educated, young, and/or affluent individuals in advanced countries with opportunities to freely express and exchange information, concerns, and interests. Since the mid- to late-1990s, when private companies began utilizing the Internet for their businesses, the Internet has tremendously changed the ways in which business is conducted and people work. Subsequently, watching the substantial impact of the Internet on business, an increasing number of social scientists and research institutes have started to examine the political and social consequences of the Internet.
So far, scholars studying this subject are loosely divided into three camps and have produced competing findings: optimistic, pessimistic, and synthetic. The first camp argues that the Internet has a positive effect on the political lives of ordinary citizens. The second one contends that its effect is at best neutral or negative. The third one posits that the effect of Internet use is neither negative nor positive in and of itself, but it is conditional, depending on certain characteristics of Internet users or external factors. Because these three groups provide general insights with which to study the Internet and democratic citizenship in the context of democratization, this section reviews them in order and identifies their limitations.
The first group of scholars (optimistic) regards the Internet as a new means to reinvigorate democratic citizenship among mass citizens and to improve the limited accountability and responsiveness of modern representative democracies (Johnson and Kaye, Reference Johnson and Kaye2003; Norris, Reference Norris2011). The group projects that the penetration of the Internet into non-democratic societies will erode authoritarian legitimacy and promote democratization (Howard and Hussain, Reference Howard and Hussain2011; Kedzie, Reference Kedzie and Allison2002). These scholars’ theoretical rationale is twofold. First, the Internet dramatically reduces the transaction cost of communication, association, and participation necessary in a democratic polity (Norris, Reference Norris2002; Xenos and Moy, Reference Xenos and Moy2007). Obtaining and sharing information is a costly human activity, but the Internet changes the cost of learning and communicating and, therefore, has the potential to improve public engagement in democratic politics. In addition, political activists take advantage of this unique feature of the Internet because it not only reduces the mobilization cost but also guarantees two-way communication with those citizens who are otherwise disengaged, apathetic, and uninterested. As a result, the Internet boosts vibrant interactions among political activists and ordinary citizens and shifts a modern representative form of democracy closer to a direct form (Drew and Weaver, Reference Drew and Weaver2006; Tolbert and McNeal, Reference Tolbert and McNeal2003).
Second, the Internet promotes the pluralism of a political communication market in which political elites and the mass media determine what becomes news and frame the information in ways that may be seen as unimportant to the mass citizens (Diamond, Reference Diamond2010; Tewksbury and Althaus, Reference Tewksbury and Althaus2000). This effect is discernible in both democratic and non-democratic societies, but it is more noticeable in the latter because authoritarian governments and pro-governmental media dominate political discourse and control deviant opinions. Cyberspace is relatively free, and the two-way communication of the Internet makes it easy for ordinary citizens and dissident groups to produce and share their own political messages. Thanks to the diffusion of the Internet, they have new opportunities to form their own voices in cyberspace and make them heard in the public arena (Diamond, Reference Diamond2010). Consequently, the Internet empowers ordinary citizens to form their own political discourse and weakens the unilateral influence of political elites and the traditional mass media in a political market (Xenos and Moy, Reference Xenos and Moy2007; Yang, Reference Yang2003).
Empirically, scholars optimistic about the Internet's potential have accumulated supportive evidence from advanced democracies and authoritarian countries. For example, Johnson and Kaye (Reference Johnson and Kaye2003) found that as people become online users, they are likely to engage in political activities such as voting. Using 1996–2000 National Election Studies data, Tolbert and McNeal (Reference Tolbert and McNeal2003) consistently found that Internet access enhances voter information about elections and increases voting participation. Moreover, Tewksbury and Althaus (Reference Tewksbury and Althaus2000) suggested that online news balances the prominent influences of traditional news producers and allows Internet users to be less influenced by cues from the political elites and the mass media. Their studies implied that the Internet reduces the leverage of the traditional mass media and promotes plural voices within a political market. The democratic utility of the Internet has even been reported in authoritarian countries. Analyzing the 2007 China World Values Survey, Lei (Reference Lei2011) showed that Chinese Internet users are not only supportive of democratic norms but also critical of Chinese one-party rule.
The second group of scholars (pessimistic) posits that the Internet disconnects people from a political community and hinders their political engagements (Margolis and Resnick, Reference Margolis and Resnick2000; Sunstein, Reference Sunstein2001). In one of the earliest studies, Kraut et al. (Reference Kraut, Patterson, Lundmark, Kiesler, Mukopadhyay and Scherlis1998) found that Internet use atomizes ordinary people by disconnecting them from their family and society, increasing personal depression. In a more systematic and nationwide study, Nie and Erbring (Reference Nie and Erbring2002) provided evidence that Internet users are more socially isolated from their family and friends. Bimber (Reference Bimber2001) examined five US national surveys on this subject and reported consistent evidence that there is little positive relationship between Internet use and various forms of political participation. In the most recent of these studies, Schlozman et al. (Reference Schlozman, Verba and Brady2010) have found that the Internet is of little help in improving the long-standing patterns of participatory inequality both online and offline among US citizens.
In addition, scholars in this group contend that Internet optimists underestimate authoritarian governments’ ability to control cyberspace and private companies in order to push Internet users away from civic issues and drive them to entertainment and business. Morozov (Reference Morozov2011) reported that authoritarian governments have developed detailed censorship and fine-tuned surveillance systems for cyberspace. For example, the Chinese government combines technical policy measures to control both public discourse on the Internet and access to foreign websites, thereby making the Internet ineffective or negative for democratization (MacKinnon, Reference MacKinnon2011). Not only have these nondemocratic cyber measures spread to other authoritarian governments such as Iran, they are not rare even in new democracies (Diamond, Reference Diamond2010). Moreover, private companies are another important actor to distract public attention from politics. Putnam (Reference Putnam2000) pointed out that ‘most of the time, energy, and creativity of the electronic media is devoted not to news, but to entertainment’ (p. 221). As a result, Internet pessimists cool down the optimists’ expectations of the Internet and provide a realistic perspective that empowered democratic citizenship stems not from the Internet but from the social and political changes behind it.
The third group of scholars (synthetic) asserts that the political effects of the Internet may be either negative or positive depending on other individual and national conditions (Groshek, Reference Groshek2009; Xenos and Moy, Reference Xenos and Moy2007). Scholars with this approach concur on the conditional effects of Internet use on democratic citizenship, whereby attentive citizens using the Internet become more engaged, but inattentive people without Internet access become less involved in democratic politics (Norris, Reference Norris2002). The Internet is known to promote this reinforcement effect because it increases citizens’ ability to customize their political and non-political information. Social psychologist Robert Kraut called this phenomenon a ‘rich get richer’ model (Kraut et al., Reference Kraut, Kiesler, Boneva, Cummings, Helgeson and Crawford2002).
Empirically, scholars in this group have examined what factors foster or hinder the effect of Internet use on civic engagement. For example, Norris (Reference Norris2002) argued the effect of the Internet is contingent on the user's purpose and reinforces the existing inequality of civic engagement among citizens. In particular, she found that informational and communicative uses of the Internet are conducive to public engagement with politics, whereas recreational and entertainment uses erode it. Political interest, discussion, and even personality are known to amplify the reinforcement effect of Internet use on the civic life of ordinary people (Kraut et al., Reference Kraut, Kiesler, Boneva, Cummings, Helgeson and Crawford2002; Nisbet and Scheufele, Reference Nisbet and Scheufele2004; Xenos and Moy, Reference Xenos and Moy2007). These studies all point to the polarizing and reinforcement effect of Internet use whereby attentive citizens become more politically oriented and inattentive ones become more disconnected from politics. At the national level, scholars in this group contend that the political effect of the Internet is dormant in day-to-day situations, but it is activated during periods of intensity, such as elections and democratic uprisings, and results in political consequences (Howard and Hussain, Reference Howard and Hussain2011; Putnam, Reference Putnam2000; Shirky, Reference Shirky2011). This synthetic group has been gaining more scholarly currency and has opened new research fields that the other two groups have not identified.
These three perspectives offer relevant insights into the relationship between the Internet and democratic citizenship in the context of democratization. Although these insights are valuable, few scholars have attempted to examine the relationship in a systematic manner within the literature of political culture and democratization. Recently, Nisbet and his colleagues (Reference Nisbet, Stoycheff and Pearce2012) integrated these insights and provided the most systematic research regarding the effect of Internet use on public attitudes toward democracy at the individual level. Given that the Internet has penetrated into developing societies and the number of Internet users has been expanding dramatically, the researchers combined two datasets from the Afrobarometer and Asia Barometer Surveys and investigated the democratic utility of Internet use. Specifically, they checked whether Internet use is conducive to democratic citizenship by increasing committed support for democracy. Consonant with the optimistic view, the results indicated that citizens who frequently use the Internet are more attached to democracy and more detached from various forms of authoritarianism than are Internet non-users (Norris, Reference Norris2011). Consistent with the synthetic perspective, Nisbet and his colleagues suggested that the democratic effect of Internet use is stronger in those democratic societies with a high level of Internet infrastructure, but it is minimal in authoritarian societies with middle-to-low levels of Internet infrastructure (Groshek, Reference Groshek2009).
In sum, prior studies have provided a great deal of information about the impact Internet use has on public norms and values toward politics and democracy. Generally, they agree that the Internet empowers citizens’ role in the politics of advanced democracies, but it has an ambiguous role in democratizing citizens of authoritarian regimes. Despite these contributions, there are two limitations of the current literature. First, theoretically, many studies have shown there is a conditional effect of Internet use on democratic citizenship in different regimes, but it is not clear why such an effect exists. Second, prior studies have not provided systematic and comparative research. They have studied either developing countries or developed democracies but not both. As Nisbet and his colleagues (Reference Nisbet, Stoycheff and Pearce2012) pointed out, ‘cross-national scholarship examining the relationship between Internet use and demand [support] for democracy more globally, or in other regions specifically, is clearly necessary’ (pp. 260–1). This study is designed to overcome these two problems.
Theoretical nexus between Internet use and democratic support
Does Internet use play a role in democratizing political attitudes among mass citizens? If it does, how is the Internet conducive to democratic citizenship? Is the democratic utility of Internet use universal or particular? This study argues that the Internet is neither democratic nor non-democratic in and of itself because it is one of the new mediums whose effects are highly dependent on users and certain conditions (Bimber, Reference Bimber2003; Diamond, Reference Diamond2010; Xenos and Moy, Reference Xenos and Moy2007). It contends that the democratic utility of Internet use is contingent on regime type. In particular, Internet use reinforces democratic citizenship in democratic societies but not in authoritarian societies. This is because the information content produced and exchanged within cyberspace differs qualitatively between the two societies, and Internet users tend to consume rather than challenge the information content of cyberspace in their country.
Social learning theories provide theoretical accounts for the different effects of Internet use in democratic and authoritarian societies (McClosky and Zaller, Reference McClosky and Zaller1984; Sniderman, Reference Sniderman1975). Social learning theories are predicated on two premises. First, the political environment in which citizens live plays a significant role in determining the particular type of political norms and values they learn. Specifically, democratic and authoritarian societies motivate citizens to learn and accept a democratic or authoritarian regime, respectively. Democratic societies encourage ordinary people to endorse democracy and reject authoritarianism, but authoritarian societies make an effort to sustain authoritarian support and use various oppressive measures to prevent their citizens from developing support for democracy.
Second, citizens are not equally capable of learning political norms. Those who can access new information technologies are more capable of learning socially dominant norms and becoming sophisticated about politics than those who cannot (Bimber, Reference Bimber2003; Norris, Reference Norris2002). The Internet enables users to customize a large amount of online content and process the information within a short time. Internet users are more quickly exposed to and made aware of current issues and affairs than Internet non-users are, which means that existing gaps in civic engagement widen between those with Internet use and those without.
Nonetheless, quick access to current issues among Internet users does not necessarily mean that they are more conscientious and responsible citizens than non-users (Sunstein, Reference Sunstein2001). Although young and educated adopters of the Internet in the early 1990s may have developed critical and conscientious voices, as Internet use became popular, socially dominant groups such as government, private companies, and opinion leaders quickly occupied cyberspace (Margolis and Resnick, Reference Margolis and Resnick2000). As Downs (Reference Downs1957) aptly pointed out, politics is a remote issue for most citizens and, thus, ordinary people are likely to accept dominant discourse in an uncritical manner rather than produce and disseminate revolutionary ideas.
In short, social learning theories hold that development of pro-democratic attitudes varies across individuals and political contexts because it is determined by contextual interactions between the political environment and Internet use. When these theoretical premises are specifically applied to democratic citizenship, they predict that in democratic societies Internet users are more active than non-users in endorsing democracy as the preferred political system, because democratic societies produce positive messages about democracy and continue to justify its superiority over authoritarianism. However, the democratic utility of Internet use is likely to disappear in authoritarian countries because the cyberspace in non-democratic settings is highly controlled by authoritarian governments. Although Internet users in these countries are able to view more information and news than non-users are, those who use the Internet are allowed to access only controlled online information content, such as pro-government propaganda, soft local issues, and entertainment.
Why does Internet use have different effects on support for democracy in democratic and authoritarian societies? Social learning theories focus specifically on two agents: government and opinion leaders. In democratic societies, government allows citizens to produce and exchange any information in cyberspace. In this open and free market of political information and thoughts, opinion leaders, including political elites and civic organization leaders, adopt the Internet to reinforce democratic citizenship (Bieber, Reference Bieber2000). In particular, they make a serious effort to mobilize Internet users by providing a large number of political messages delivering democratic content. Because political elites and opinion leaders of advanced democracies often disagree on specific policy issues but agree on regime principles, democratic content tends to be dominant in cyberspace. A series of social communications in the cyberspace of stable democratic societies eventually leads Internet users to become aware of democratic politics and supportive of democratic norms and values (Druckman et al., Reference Druckman, Perterson and Slothuus2013; Johnson and Kaye, Reference Johnson and Kaye2003; Scheufele and Nisbet, Reference Scheufele and Nisbet2002). As a result, Internet users in these stable democracies are more likely than non-users to be democratic citizens endorsing democracy and rejecting authoritarianism (Norris, Reference Norris2011).
Internet users in authoritarian societies cannot avoid governmental attempts to control cyberspace, and opinion leaders are not allowed to disseminate anti-government and democratic messages through the Internet (Mozorov, Reference Morozov2011; Stockmann and Gallagher, Reference Stockmann and Gallagher2011). Instead, they are more likely to be exposed to pro-government propaganda and learn these messages more quickly than those without Internet access. These early adopters of the Internet in authoritarian societies are likely to be close to those who have benefits from their political system because they are located at high levels of social stratification. Thus, Internet users in these societies not only would find it hard to internalize democratic norms but also would have difficulty expressing support for democracy and rejection of authoritarianism.
This non-democratic effect of Internet use would be most discernible when authoritarian governments and pro-government mass media successfully control flow of information content in cyberspace (Diamond, Reference Diamond2010; MacKinnon, Reference MacKinnon2011). Due to detailed censorship and surveillance systems by these authoritarian governments, Internet use is likely to remain restricted to commercial and entertainment purposes and government agencies’ use of the Internet to amplify their own propaganda and manipulate public opinion among ordinary peop+le. As a result, the empowered ability of Internet users is not likely to result in democratic demand or support.
In sum, the hypothesis I test in this study is fairly simple: Internet use promotes democratic support only in democratic societies, and the effect of Internet use becomes negative or nullified in authoritarian societies.
Research design
Data
To test the effect of Internet use on democratic citizenship, this study employs survey data from 45 societies in the latest World Values Survey (hereafter, WVS). The WVS investigated social and political values among more than 80,000 people across 57 societies during the period 2005–08. The WVS is the most recent survey examining democratic support and Internet use among global citizens. Selecting countries where all the necessary questions for this study were available resulted in a sample of 45 societies and 61,873 respondents across seven major regions (see Appendix A for the sample country information).
The sample reflects substantial diversity in both democratic and economic development. Specifically, following the Freedom House regime category, the sample consists of 34 ‘Free,’ six ‘Partially Free,’ and five ‘Not Free’ countries. Within the 34 free societies, 15 are old democracies and 19 are new democracies having experienced democratic transition after 1974 when the third wave of democratization began in Portugal. In addition, the sample contains 18 members of the Organisation for Economic Co-operation and Development (OECD), which is a forum of rich and advanced countries. On the other hand, seven countries are poor countries, having a gross domestic product (GDP) per capita of less than $3,000. Given that such diversity of the country sample provides fertile ground to test the core hypothesis identified in the previous section, the cross-national analysis at the global level will make a significant contribution to the current literature of Internet use and democratic political culture.
Dependent variables: democratic support
The dependent variable for this study is political support for democracy. In the literature of democratic political culture, there is general consensus that political support for democracy takes the form of a choice between democracy and its alternatives and that the choice is naturally dynamic. In the developing world, where a number of countries have experienced democratic transition, the cultural norms of the previous authoritarian regimes tend to cohabit with the newly or partially installed institutions and procedures of democracy. Citizens with little experience of and limited sophistication concerning democratic politics may be uncertain whether democracy or dictatorship offers more satisfying solutions to the problems facing their societies. Due to such uncertainty, citizens who are democratic novices often concurrently embrace democratic and authoritarian political propensities. Considering this ambivalence, scholars on this subject suggest that democratic support should be measured in two dimensions: one involving favorable orientations to democracy and another involving the opposite trend in orientations toward authoritarianism. Consequently, scholars in this research have measured political support for democracy by gauging the extent to which individual citizens simultaneously accept democracy and reject authoritarianism (Inglehart, Reference Inglehart2003; Mattes and Bratton, Reference Mattes and Bratton2007; Rose et al., Reference Rose, Mishler and Haerpfer1998; Shin, Reference Shin, Dalton and Klingemann2007).
Following this analytical convention, this study uses four questions from the fifth wave of the WVS to construct political support for democracy in three sequential steps. The four questions investigate global citizens’ preferences for three different regime types (democratic system, civilian dictatorship, and military rule) and importance of living in a democratic society (see survey questions in Appendix B). The first three questions allow citizens to express one of four responses from ‘Very good’ to ‘Very bad,’ and the final question offers them a choice of ten responses from ‘Not at all important’ to ‘Absolutely important.’ Because the first three questions were operationalized as a four-point scale index, the ten options of last one were transformed into an index with the same scale (1,2 = 1; 3,4,5 = 2; 6,7,8 = 3; 9,10 = 4). First, I measured acceptance of democracy by summing the respondent's preference for a democratic system and personal importance of living in a democracy, which ranged from 2 (weak acceptance) to 8 (strong acceptance). Second, preferences against civilian dictatorship and military rule were combined to produce rejection of authoritarianism ranging from 2 (weak rejection) to 8 (strong rejection). These two measures equally represent pro-democratic and authoritarian orientations, respectively. Finally, this study summed both acceptance of democracy and rejection of authoritarianism, which yielded a composite index of political support for democracy ranging from 4 (support for authoritarianism) to 16 (support for democracy). These measures allow this study to examine the effect of Internet use on both democratic support and its subcomponents in a balanced and detailed manner.
Independent and control variables
The core independent variables for this study are Internet use as an information source at the individual level and the level of democracy at the country level. People use various sources to learn about their environment. The sources include newspapers, news broadcasts, books, talks with friends, and the Internet. The WVS asked respondents whether they used the Internet to obtain information last week or not. The response is dichotomous. As a country-level variable, level of democracy is measured by the reversed Freedom House regime ratings for the survey year, which ranges from 0 (stable authoritarianism) to 12 (advanced democracy).
This study includes ten control variables from five theoretical models of democratic political culture: modernization, social capital, political learning, performance evaluation, and demography. First of all, GDP per capita for economic development at the country level and educational attainment and income at the individual level are controlled for theories of modernization, contending that democratic political culture emerges as a society experiences socioeconomic development (Lipset, Reference Lipset1960). Second, theories of social capital argue that vibrant civic activities and high levels of interpersonal trust provide a social foundation for democratization and democratic governance (Almond and Verba, Reference Almond and Verba1963). This study controls for effects of social capital on democratic citizenship by including membership in civic organizations and interpersonal trust. Third, recent theories of political learning and democratization assert that democratic citizenship rises as citizens becomes interested in politics and follow political issues (Bratton et al., Reference Bratton, Mattes and Gyimah-Boadi2005). I include political interest and media use to represent this model. Fourth, theories of performance evaluation contend that citizens orient themselves toward or away from democracy and its alternatives, depending on how they judge the performance of their democratic regime. To control public evaluation of democracy, this study includes the perceived extent of democracy and financial satisfaction at the individual level (Rose, Mishler, and Haerpfer, Reference Rose, Mishler and Haerpfer1998). Finally, as common in other studies, two demographic characteristics are controlled in the following analyses: age and gender.
Empirical model
One major problem of cross-national survey research is the high rate of missing data as a result of non-responses. In the data used, the percentages of missing values ranged from 0.1% for female dummy to 9.9% for individual preference for civilian dictatorship. When all variables of interest are included in the empirical model, the model excludes 17,473 (28.7%) as a result of list-wise deletion. Because this practice would lead to ‘a loss of valuable information at best and severe selection bias at worst’ (King et al., Reference King, Honaker, Joseph and Scheve2001: 49), I employed multiple imputation, which involves imputing m values for each missing cell based on the existing information. I used Amelia II, developed by King and his colleagues, to generate ten imputed data sets. This resulted in no missing values for any predictors in the multiple imputed data sets. To conduct the separate analyses of these data sets, I ran HLM 6.8, which calculated the average of the parameter estimates from the separate analyses and computed the standard errors (Raudenbush and Bryk, Reference Raudenbush and Bryk2002).
Given that the WVS has a hierarchical data structure comprising both individual-level and country-level variables, multilevel modeling is the best method for this study (Raudenbush and Bryk, Reference Raudenbush and Bryk2002). However, the preliminary step before conducting major empirical models for this study was to check whether multilevel modeling is statistically necessary. To test the necessity of multilevel modeling, this study conducted an analysis of variance (ANOVA) of democratic support by excluding all the independent and control variables. The ANOVA shows that 81% of the variance in democratic support comes from the individual level and 19% from the country level. The variance components are statistically significant, which suggests that democratic support simultaneously varies at the country and individual levels and that ignoring the hierarchical structure of democratic support would produce inaccurate inferences. Thus, the use of multilevel modeling is justified for this research.
Empirical results and discussion
The core argument for this study is that the effect of individual Internet use on democratic support is conditioned by the national development of democracy. Internet users in advanced democracies are likely to be exposed to enriched democratic content in cyberspace, whereas Internet users in stable authoritarian countries are not allowed to view such democratic content due to authoritarian control of online information flow. It logically suggests that individual Internet use generally exerts a positive effect on democratic support, but its effect is more robust in democratic countries than non-democratic countries.
To test these expectations, this article estimated two empirical models. As a base model, Model 1 included all the independent and control variables at both the individual and country levels. As an interaction model, Model 2 added cross-level interaction between individual Internet use and level of democracy. The first model tests whether individual Internet use is positively associated with support for democracy and the second one checks whether the relationship between the former and the latter varies by level of democratic development.
The results are presented in Table 1. As a basic analysis, the first model in Table 1 shows the coefficients for country-level and individual-level predictors. Internet use appears to have a positive effect on democratic support and the effect reaches statistical significance. It means that at the global level, Internet users are generally more supportive of democracy over authoritarianism than non-users. The empowered ability of Internet users to view and learn online political content appears to increase the gap in democratic citizenship between them and non-users. However, national development of democracy is positively associated with democratic support, although the effect is not statistically significant. This suggests that democratic support does not vary between democratic and authoritarian societies and reflects the recent rise of democratic support in the developing world (Inglehart, Reference Inglehart2003).
Table 1. Multilevel analyses of democratic support
Notes: Significance level: *p ≤ 0.10; **p ≤ 0.05; ***p ≤ 0.01. Standard errors are not reported due to limited space. Germany is divided into the West and the East due to different histories of democracy. So the sample is 44 countries and 45 societies.
Regarding the control variables, most of them are statistically significant with the exception of income and female dummy. As expected, political support for democracy substantially differs across different levels of educational attainment, lending strong support for the modernization theory (Lipset, Reference Lipset1960). Within the social capital theories, interpersonal trust has a significant and positive effect on democratic support, but membership in civic organizations has the opposite effect. This result suggests that people who psychologically trust others are more supportive of democracy than those who distrust others. The results together indicate that civil society plays an ambiguous role in generating political support for democracy, although it has a large role in producing other civic virtues (Tarrow, Reference Tarrow1996).
In the model of political learning, both variables are statistically significant and their effects are substantial. Politically attentive citizens characterized by strong political interest and media use are more committed to democracy over authoritarianism than the politically alienated (Bratton et al., Reference Bratton, Mattes and Gyimah-Boadi2005). Within the theories of performance evaluation, financial satisfaction and perceived extent of democracy are also statistically significant but the effects are opposite: negative for the former and positive for the latter. This suggests that citizens direct their dissatisfaction with the financial situation and positive evaluation of democratic performance toward support for democracy (Rose et al., Reference Rose, Mishler and Haerpfer1998). Finally, old people are more likely to support democracy than young people, but there is no difference between genders.
How strongly does Internet use influence democratic support? Figure 1 provides a summary of the substantive effects on democratic support of both individual- and national-level predictors. Because these variables are scaled in different ways, I compare changes in standard deviations to estimate the substantive effect of a given determinant. More specifically, I estimate how a one-unit standard deviation increase in Internet use and control variables will result in the predicted change in democratic support.
Figure 1. Predicted changes in democratic support by one standard deviation increase of the independent and control variables
The results in Figure 1 suggest that the effect of Internet use on democratic support is not so strong. Educational attainment and perceived extent of democracy emerged to have the most substantial effect on political support for democracy. In addition, three predictors in a second group have a moderate effect: political interest, media use, and age. Internet use belongs to a third group, whose predictors have a small effect on public support for democracy: membership in civic organization, interpersonal trust, and financial satisfaction. Considering that a set of relevant variables are controlled for, these results indicate that the estimated effect of Internet use on democratic support has an independent effect on democratic support, but its impact is at best very modest.
Is the effect of Internet use contingent on regime type? Model 2 presents a specific picture that the effect is highly dependent on levels of democracy, lending strong support to the core hypothesis. In Model 2, the uninteracted Internet use variable captures its effect on democratic support when the country's Freedom House index is the lowest (0). For this stable and extreme authoritarian regime, the effect of Internet use is obviously negative but statistically insignificant. However, the interaction term between Internet use and the Freedom House index is positive and statistically significant. This finding suggests that the effect of Internet use is contingent on regime type: As the country's overall level of democracy increases, the effect of Internet use on democratic support tends to rise as well. Moreover, this positive effect of the interaction term is robust even when robust sets of control are employed in the empirical Model 2.
Based on the estimates from Model 2, Figure 2 provides a graphical illustration of the cross-level interaction between individual Internet use and national democratic development. Figure 2 draws a line graph plotting the estimated effect of Internet use on democratic support at different levels of the Freedom House index. I also plot the estimated 95% confidence intervals. Given that the country sample collected from the latest WVS contains more democracies than authoritarian countries, the expected effect of Internet use is more precisely measured in the upper right section of Figure 2 than in the lower left section.
Figure 2. Predicted effect of internet use on democratic support at different levels of Freedom House Index
Figure 2 offers a clear message: in democratic societies, Internet use breeds support for democracy, but such democratic utility is nullified in authoritarian societies. In societies with the middle-to-low range of the Freedom House index, the effect of the Internet is negative but not statistically significant. In fact, the effect is not different from 0. Specifically, consider that Freedom House categorizes levels of democracy 0 to 3 as ‘Not Free,’ 4 to 8 as ‘Partially Free,’ and 9 to 12 as ‘Free’. Individual use of the Internet exerts a significant effect on democratic support only in countries guaranteeing certain levels of civil liberties and political rights.
Previous analyses demonstrated that the democratic effect of Internet use is highly dependent on regime type. This study takes this general finding one step further and inquires whether the effects upon acceptance of democracy and rejection of authoritarianism are equal. Does Internet use have similar or different effects on pro-democratic and authoritarian orientations? To answer this question, I regressed both acceptance of democracy and rejection of authoritarianism on the same predictor variables in Model 2. Then, using these results, I drew a line graph (Figure 3) plotting the estimated effect of Internet use on democratic support's two subcomponents at different levels of the Freedom House index.
Figure 3. Predicted effect of internet use on acceptance of democracy and rejection of authoritarianism at different levels of Freedom House Index
Figure 3 indicates that Internet use has somewhat different effects on acceptance of democracy and rejection of authoritarianism. Internet use is found to have a stronger effect on the former than on the latter. With regard to acceptance of democracy, the democratic effect of Internet use exists only in democratic societies but is minimal. Given that the slope of the first graph is very moderate, it seems that regime type does not matter substantially in this case. This finding suggests that democracy has been widely accepted around the world over the last two decades and, thus, strong acceptance of democracy is evident in both Internet users and non-users within a country and between countries (Inglehart, Reference Inglehart2003; Norris, Reference Norris2011). However, it appears that the democratic utility of Internet use centers on rejection of authoritarianism. Because the slope of the second graph is steep, the effect of Internet use is stronger in democratic societies and significantly increases as a country becomes more democratic. Authoritarianism is losing its currency among Internet users of democratic countries.
Substantially, these findings provide systematic and comparative evidence for a conditional effect of Internet use on public support for democracy, which prior studies based on regional and national surveys have not demonstrated. Internet use plays a greater role in democratizing ordinary citizens living in advanced democracies. In contrast, it is hard to breed democratic support in authoritarian societies (Mozorov, Reference Morozov2011; Nisbet et al., Reference Nisbet, Stoycheff and Pearce2012). It appears that the democratic potential of the Internet is effectively limited in authoritarian countries.
The results presented in this study provide a more realistic picture in that Internet use does not automatically produce such democratic utility, but its effect is highly dependent on regime type. In advanced democracies, government, political parties, activists, and opinion leaders produce and disseminate democratic information content into cyberspace; Internet users are more likely to be exposed to this content and learn it, compared with Internet users in non-democratic societies. The Internet facilitates such democratic information production and exchange among ordinary citizens and increases public support for democracy only in those countries already enjoying civil liberties and political rights.
On the other hand, both production and exchange of such democratic content in cyberspace are highly controlled by authoritarian governments in non-democratic settings (Diamond, Reference Diamond2010; Morozov, Reference Morozov2011). In these societies, Internet use is likely to be oriented toward entertainment and business rather than politics. Moreover, Internet users in these societies are oriented to learn pro-government information and propagandas. For this reason, Internet use rarely exerts a democratic utility, and Internet users are not likely to express support for democracy and rejection of authoritarianism. These different environments surrounding national cyberspace result in the different effects of Internet use on democratic support: positive in democratic societies and negative or null in authoritarian ones.
Conclusion
Over the last two decades, an increasing number of citizens around the world have used the Internet via either computer or other electronic devices. At the same time, democracy has become the most preferred political system among the global mass publics regardless of their locations. Although an increasing number of scholars and international institutes have sought to examine the role of the Internet on democratic citizenship around the world, few studies have attempted to investigate the relationship from a comparative perspective. Is the effect of Internet use negative or positive on democratic support? Does it vary across countries? If so, why does it vary and what factors condition it? Country- and region-based studies have touched on these questions, but answers have not been sought in a theoretical and empirical manner.
Hence, the contribution of this article is twofold. Theoretically, I used social learning theories and drew predictions that Internet use has different effects on democratic support depending on regime type (McClosky and Zaller, Reference McClosky and Zaller1984). In democratic societies, political activists, and opinion leaders freely produce democratic ideas and political thoughts and disseminate them to ordinary citizens through the Internet. This is the social learning process of democracy in cyberspace whereby Internet users are more likely than non-users to learn this online political content. However, authoritarian governments have developed detailed censorship and surveillance systems to control cyberspace so that Internet users in non-democratic settings are likely to be exposed to online information content filtered by their government (Diamond, Reference Diamond2010; MacKinnon, Reference MacKinnon2011). Moreover, authoritarian governments attempt to restrict Internet use of ordinary citizens to business and entertainment and prevent them from searching democratic websites and exchanging democratic ideas in cyberspace. Due to this authoritarian regime effect, Internet use reinforces democratic citizenship not in authoritarian countries but in democratic nations around the world.
Empirically, I subjected these theoretical expectations to a systematic test, using public opinion data of 45 societies included in the WVS. The results support these theoretical expectations. Internet use generally increases democratic support around the world. However, taking a close look at the relationship between them, it appears that Internet use bolsters public support for democracy in democratic societies but not in authoritarian ones. Furthermore, Internet use exerts a stronger effect on rejection of authoritarianism than on acceptance of democracy.
These findings indicate that Internet use has both bright and dark sides in terms of democratic citizenship. The bright side is that advanced democracies can take advantage of the Internet to improve democratic citizenship. The dark side is that authoritarian states can utilize the potential of the Internet to deter the development of democratic citizenship and sustain authoritarian regimes. When both sides are considered together, it is clear that the Internet has a polarizing effect on democratic citizenship between democratically advanced societies and stable authoritarian countries. All in all, these observations inspire a cautious perspective on the Internet and democratization and bring to mind the conventional wisdom that ‘technology is merely a tool, open to both noble and nefarious purposes’ (Diamond, Reference Diamond2010: 71).
About the author
Youngho Cho (Ph.D.) is a lecturer at Hanyang University, Seoul, South Korea. He has several forthcoming articles on public attitudes and knowledge about democracy in the journals Democratization, Political Studies, and Social Indicators Research.
Appendix A. Country sample from the fifth wave of the World Values Surveys
Appendix B. Survey questions
Political Support for Democracy (A composite index)
I'm going to describe various types of political systems and ask what you think about each as a way of governing this country. For each one, would you say it is a very good, fairly good, fairly bad, or very bad way of governing this country?
V148. Having a strong leader who does not have to bother with parliament and elections/
V150. Having the army rule/V151. Having a democratic political system
V162. How important is it for you to live in a country that is governed democratically? On this scale where 1 means it is ‘not at all important’ and 10 means ‘absolutely important’ what position would you choose?
Internet use
People use different sources to learn what is going on in their country and the world. For each of the following sources, please indicate whether you used it last week or did not use it last week to obtain information: V228 Internet, Email
Education
V238. What is the highest educational level that you have attained?
Income
V253. On this card is a scale of incomes on which 1 indicates the ‘lowest income decile’ and 10 the ‘highest income decile’ in your country. We would like to know in what group your household is.
Membership in Civic Organizations (A composite index)
Now I am going to read off a list of voluntary organizations. For each one, could you tell me whether you are an active member, an inactive member or not a member of that type of organization? V25. Sport or recreational organization/V26. Art, music, or educational organization/V29. Environmental organization/V31. Humanitarian or charitable organization/ V32. Consumer organization.
Interpersonal Trust
V23. Generally speaking, would you say that most people can be trusted or that you need to be very careful in dealing with people?
Political Interest
V95. How interested would you say you are in politics?
Media Use (A composite index)
People use different sources to learn what is going on in their country and the world. For each of the following sources, please indicate whether you used it last week or did not use it last week to obtain information: V223. Daily newspaper/ V224. News broadcasts on radio or TV
Perceived Extent of Democracy
V163. How democratically is this country being governed today? Again using a scale from 1 to 10, where 1 means that it is ‘not at all democratic’ and 10 means that it is ‘completely democratic,’ what position would you choose?
Financial Satisfaction of Household
V68. How satisfied are you with the financial situation of your household?
