Can regimes really discourage social networking? Urbanization, mobile phone use, and the dictator's plight
First Monday

Can regimes really discourage social networking? Urbanization, mobile phone use, and the dictator's plight by Shin Haeng Lee



Abstract
Are dictators ever really successful at hindering the diffusion of social networking services (SNSs)? This study reports a panel data analysis on 181 countries observed from 2010 to 2013, to assess country-level predictors of Facebook adoption. The findings show that the spread of such a global SNS decelerates as perceptions of democratic governance deteriorate above and beyond socioeconomic and infrastructural development. Nevertheless, once dictatorship fails to maintain socio-political stability, the diffusion of Facebook accelerates to a greater extent than in democracy. This trend could transcend institutional constraints, as well as socioeconomic developments, because of widespread proliferation of mobile phone use in addition to increasing wireless or shared connectivity in urban areas.

Contents

Introduction
Literature review
Data
Empirical analysis
Results
Discussion and conclusion

 


 

Introduction

Has the diffusion of the Internet and mobile devices caused an equivalent proliferation of social networking services (SNSs) around the world? Or has the inequity in digital access among countries also created a gap in the use of such applications? Considering how crucial the role of information and communication technologies (ICTs) is for developed and developing economies, comparative scholarship has extensively examined the causes of the digital divide between groups of countries as well as the consequences of this divide. What remains unclear is how digital media benefit their users, socially and politically, in different societies. Thus, certain questions come up, which are addressed in this article: Would countries uniformly adopt global SNSs if they had the same opportunities for ICT access; and, if not, in what ways would they differ in their usage patterns?

Global social networking platforms such as Facebook and Twitter diffuse across countries differently from the viewpoint of the access gap. For example, despite the disparity in Internet use between Europe (61 percent) and Asia (26 percent) as of 2011, the gap narrows when it comes to Facebook use among Internet users: 27 percent and 18 percent, respectively (Miniwatts Marketing Group, 2012). The Facebook adoption of the online population of Mexico (98 percent) is vastly different from that of Germany (33 percent), even though both countries had a similar Internet penetration rate as of December 2011. Such a difference in Facebook penetration is attributable to technology affordability and accessibility on the basis of socioeconomic development, as well as the market situation. Yet social media adoption is also influenced by particular national demands for the digital service in line with socio-political circumstances beyond technological availability.

Cross-country comparison of Facebook adoption is important because it provides citizens with unconventional ways to connect and coordinate. Specifically, such an SNS can be used as a means for bridging and boding social capital by transcending individuals’ structural constraints in relation to resources and abilities (Ellison, et al., 2007). Political dissidents also take advantage of online social networks to circulate digital content that could not have been distributed through traditional media. At the meso- and macro-levels, social media enable local activist groups and civil societies to have transnational connectivity as well as large neighborhood networks (Shirky, 2011). Thus, it is not unusual that protesters use such technologies for mobilization of networked collective action (Bennett and Segerberg, 2011). Howard and Hussain (2013) further argued that the current mobilization for political movements such as the Arab Spring in the early 2010s might not be conceivable without their exploitation of social media, although the technology was insufficient for democratization. Therefore, it is presumable that the diffusion of SNSs is conditional on institutional environments where authoritarian regimes may want to keep civil society under the control of the state. That is, dictatorship is unlikely to let people benefit from such a decentralized means of connection and organization beyond control. Indeed, the political willpower of governments has the ability to deter the diffusion of technology by introducing censorship and regulations to evoke a chilling effect on people’s willingness to take part in the global connectivity (Crenshaw and Robison, 2006; Milner, 2006).

Figure 1 demonstrates the adoption pattern of Facebook at a slower rate in authoritarian countries than in democracies in 2012, even with a similar Internet penetration rate [1]. This indicates that Internet adoption is not a sufficient cause of online social networking in all countries; rather, the diffusion of the global SNS is discouraged under less democratic regimes that create an unfavorable atmosphere for political rights and civil liberties. Nevertheless, this may not be the case where more and more countries gain a concomitant desire to enjoy the resulting economic benefits of the supply of ICTs (Howard, et al., 2011; Zittrain and Edelman, 2003). Figure 1 shows that as Internet penetration increases, institutional perspective becomes less convincing in explaining a cross-country variation in Facebook adoption. Therefore, with the expansion of access to ICTs, repressive governance becomes ineffective in hindering social network connectivity.

 

Scatter plot of cross-country Facebook adoption against Internet penetration
 
Figure 1: Scatter plot of cross-country Facebook adoption against Internet penetration.
Note. “Free” refers to a category for countries with their freedom index values from 1 to 2.5 on a scale from 1 (most free) to 7 (least free). “Partly Free” and “Not Free” indicate countries with values from 3 to 5 and from 5.5 to 7, respectively.
Source: Author’s calculations based on data from Facebook, Freedom House, and the International Telecommunication Union.

 

Concerned about the digital divide, many scholars have investigated the socioeconomic, political, and cultural factors involved in cross-country differences in Internet access and usage (Guillén and Suárez, 2005; Hargittai and Hsieh, 2013; Robison and Crenshaw, 2010; Dijk, 2005). But much less is known about the country characteristics that lead to differences in the use of global social networking platforms such as Facebook. This study argues that the global digital divide along the lines of socioeconomic and political developments becomes less prominent in the diffusion patterns of SNSs because of the increasing adoption of mobile phones. Moreover, Facebook adoption accelerates when dictators fail to maintain socio-political stability, because people seek to become more informed and connected via a digital tool in restrictive media environments.

Facebook was chosen as an SNS of study because of its current global reach as the world’s largest social medium, with one billion monthly active users out of 2.5 billion online and seven billion total population in the world (Fowler, 2012). Also, Cosenza (2012) has found that Facebook has SNS market dominance in 126 out of 137 countries while monopolizing the global SNS supply. I conducted a large-N analysis to predict the extent of social network connectivity through Facebook adoption given Internet penetration. The following section reviews country-level factors that include socioeconomic, infrastructural, and institutional conditions, which function as a constraint on or a catalyst for the spread of the digital network.

 

++++++++++

Literature review

Global digital divide

The scholarship on the global digital divide has shown that cross-national diffusion patterns of ICTs mirror and even reinforce the pre-existing contours of multidimensional inequality between countries as well as within them (Chinn and Fairlie, 2006; Norris, 2001). First, this view tends to a Matthew-type effect based on the notion that “the rich get richer” (Dijk, 2005). In relation to supply, material affluence benefits the rich disproportionately to adopt a new means of social networking and reduce the cost of access over the poor (Beilock and Dimitrova, 2003). Also, the pre-existing development of telecommunications infrastructure has an overwhelming influence on the extent to which people are supplied with “network readiness,” reinforcing inequality between developed countries and the rest of the world (Andrés, et al., 2010). In less technologically equipped countries, however, structural constraints require relatively high costs of digital access and hinder the spread of SNS use. The more such economic and infrastructural resources are supplied, the more opportunities are available for people to benefit from new technologies for social networking.

Given the trend to globalization, scholars have also shown that digital technology diffuses at faster rates in cosmopolitan societies: market openness is conducive to the diffusion of new ICTs (Crenshaw and Robison, 2006; Drori and Jang, 2003; Guillén and Suárez, 2005). For example, the European Community’s unification has resulted in a faster diffusion of new ideas, technologies, and products among the member nations than among non-member nations (Mahajan and Muller, 1994). Drori and Jang (2003) lend additional support to the outweighing influence of global economic system on the connectivity of information technology. To predict a change in a country’s number of Internet users, Andrés, et al. (2010) further identified its network externalities as a significant determinant of supply and demand conditions for technology adoption. Vibrant international trade and foreign direct investment (FDI) spur technology adoption because of not only their role in economic performance but also their network capacity. Accordingly, despite the international divergence in technology adoption between rich and poor countries, a country’s embeddedness in the global economic system can contribute to the proliferation of SNSs.

The adoption of Facebook is driven not just by economic and infrastructural imperatives but also by sociological ones. The previous literature pointed out a disparity of technology adoption along the lines of socio-economic status (Chinn and Fairlie, 2006; Hargittai, 1999). Especially, socioeconomic status matters because a “knowledge gap” exists in the usage pattern of digital media beyond the access gap. Given how people actively use digital ICTs beyond mere access, for instance, literacy and education are considered important contributors to the diffusion of the technology (Hargittai, 1999). Hargittai and Hsieh (2013) put emphasis on “digital inequality” in individuals’ autonomy and skills to use the technology for diverse “capital-enhancing” activities, corresponding to existing social stratification lines. Regarding a cross-country difference in the adoption of the global SNS, therefore, it is not independent of the national average of human development based on literacy, education, and standard of living, as well as income and trade.

H1: Facebook adoption will take place at a faster rate in countries with higher levels of socioeconomic and infrastructural development.

Leapfrogging conditions

Despite the continuity of the global digital divide, it seems that fewer and fewer countries are isolated from the diffusion of ICTs as a result of the expansion of transnational market forces. With regard to the rapid spread of inexpensive mobile devices, furthermore, recent studies have examined whether the digital inequality narrows between rich and poor countries. Some comparative studies found the effectiveness of policy reform, which helped bridge the digital divide (Billon, et al., 2009; Howard and Mazaheri, 2009; Robison and Crenshaw, 2010). Privatization, de-monopolization, and deregulation of telecommunications are ways for developing countries to leapfrog the barriers of socioeconomic and infrastructural inequalities across countries. Howard and Mazaheri (2009) found that market competition was not necessarily superior to public oversight in improving network readiness. Nevertheless, it is certain that skillful enactment of policy reform accelerates the provision of access to affordable ICTs while bypassing older technological stages that developed countries had to undergo. The so-called “last-mover advantage” is manifested by the feasibility of wireless connectivity systems instead of the costly establishment of wired infrastructure (Chadwick, 2006; Chinn and Fairlie, 2006). Moreover, Internet-accessible mobile phones (dubbed “smartphones”) make it easier and cheaper to access Facebook than personal computers do for those who lack digital literacy and skills.

Previous studies also suggest that urbanization can help reduce the gap between the haves and the have-nots (Caselli and Coleman, 2001; Chinn and Fairlie, 2006). The high density of the urban population facilitates the expansion of digital networks via public locations such as cybercafés or libraries, where people can share Internet access (Billon, et al., 2009). The concentration of resources in urban areas serves as another leapfrogging condition via the market development of second-hand mobile devices (Pearce and Rice, 2013), as well as affordable deployment of wireless connectivity. In the same vein, Robison and Crenshaw (2010) captured that concentrations by large cities on their capital and infrastructure enabled developing countries to lower the cost of the adoption of ICTs (see also Crenshaw and Robison, 2006). Accordingly, the more people populate a country’s urban areas, the more likely they will have opportunities to access social media using common ICTs despite a low level of personal-technology adoption.

Of course, such leapfrogging effects are conditional on regulatory institutional environments, because economic liberties and market competition act for the adoption of new technologies in developing societies (Robison and Crenshaw, 2010). Some studies have focused further on the power of regime type in light of the notion that authoritarian rules have the capacity as well as the desire to keep a grip on media and communication network (Guillén and Suárez, 2005; Milner, 2006). Indeed SNSs, marked by decentralized flows of information, are more likely to trouble dictators than are the mass media who implement coercive policies or deploy tools for filtering unwanted content, distorting opinions, or even blocking access to specific Web sites. In view of this, political openness is likely to fuel the lowering of entry barriers to using the global communication platform.

Nevertheless, the dichotomies between democratic and authoritarian regimes and between closed and open markets may not determine international patterns for the adoption of the global SNS. As more and more countries seek the perceived economic benefits of the technology adoption and the resulting policy reform, the influence of regime type and political liberties on technology diffusion becomes subtle and complicated. Kedzie (2002) put forth the concept of the “dictator’s dilemma” to emphasize the need of authoritarian regimes to co-opt digital technologies for their economic development. Moreover, Howard, et al. (2011) have shown that even democracies often engage in “blocking Internet access and disabling digital networks,” beyond mere censorship of digital content, in response to social and political unrest. In this respect, Corrales and Westhoff (2006) found that democratic development has become less important to the diffusion of ICTs in market-oriented, high-income societies (see also Robison and Crenshaw, 2010).

Furthermore, the proliferation of digital media use can sometimes tally with dictators who seek to exploit the technology as a tool of political control (Deibert, et al., 2010; Morozov, 2011). Some scholars noted that the power of the state negated the effects of Internet diffusion on democratization by manipulating ICTs as panoptic power, such as subjecting people to propaganda and surveillance, and conducting espionage (Corrales and Westhoff, 2006; Kalathil and Boas, 2003; Pearce and Kendzior, 2012). Digitally networked authoritarianism is no longer idiosyncratic, given the prevalent implementation of a deliberate policy toward market liberalization reforms in the telecommunications industry while keeping an intranet system from the transnational network structure (Howard, et al., 2011). Thus, leapfrogging into networked societies does not necessarily result from political openness or liberalization but rather from the increasing availability of affordable and accessible ICTs.

H2: If mobile phone adoption and urbanization increase in a country, then its Facebook adoption will accelerate above and beyond socioeconomic and infrastructural conditions.

Socio-political circumstances

Although the digital-divide approach has extensively proven its analytical strength, it has been relatively preoccupied by the adopting units’ socioeconomic, technological, and institutional capabilities to have more supplies. However, the leapfrogging opportunities become available worldwide more and more through smartphones and wireless connectivity systems. This recent trend puts an emphasis on the demand side for, as well as the supply side of, social networking opportunity online. As a result, the global-digital-divide model has room for the impact that societal demands have on international technology diffusion.

Rogers (2003), credited with the “diffusion of innovations” model, offers a distinct approach to the understanding of Facebook adoption. Specifically, he highlighted the mechanism by which technology adoption spread through a communication system over time. Also, the model characterizes the S-shaped pattern of the diffusion that takes place at slower rates among early adopters and subsequently accelerates among mass recipients until the market matures. This pattern of diffusion sheds light on the network effect insofar as the incentives for the technology adoption become prevalent at an exponential rate. That is, the diffusion rate of the SNS can be further hastened by user increase in size and the growing circulation of its perceived utility and resulting benefits. Thus, previous studies found that the number of technology users per capita in the previous year predicted the adoption rate in the following year (Andrés, et al., 2010; Robison and Crenshaw, 2010).

At the international level, the diffusion of Facebook can be also considered a matter of how countries are positioned to learn the advantage of and judge the needs for the service “from overseas rather than generating such knowledge domestically” [2]. This suggests that a recipient country’s demand for the adoption of Facebook can increase because of its stronger relationship with the U.S. as the lead country — eventually leading to the emergence of the global SNS. In other words, a tighter economic, political, and/or cultural relationship with the lead country facilitates the spillover of technology into the adopters. This view explains why a certain technology diffuses at different rates across countries from what is expected given their socioeconomic, infrastructural, and institutional merits.

In addition, societal demands for a new communication and organizing channel are likely to be augmented by the growth of civil society where activists and civic groups pursue their interests independently from the state (Diamond, 1994). Putnam’s (2000) concept of social capital also pointed to the development of a civil society in association with the formation of trust and social ties. In view of this, institutional arrangements matter to Facebook adoption not because they have an influence on the technology supply side but because the political system has the ability to shape such structural and cultural aspects of social capital that are in turn related to the demand side for the SNS. Indeed, Norris (2002) found that a country’s aggregate level of social cohesion was strongly correlated with civil liberties and democratic involvement. Furthermore, the presence of adverse institutional conditions to the development of civil society development is thus likely to hinder the adoption of digital networking tools given their specific ability to enhance the frequency of peer-to-peer interactions and improve the nature of social connectivity (Ellison, et al., 2007). Thus, the development of participatory democracy will be related to growing demands for SNSs that provide the opportunity for social relationships and civic actions.

When a certain technology’s capacity meets a specific need of a society, it diffuses at faster rates. Conversely, where the same technology is considered detrimental to its adopters and promoters in other contexts, its distribution is retarded. Taking this view, Corrales and Westhoff (2006) found that a country’s restriction on political liberties gave primacy to the adoption of television over the Internet. Hence, it is not uncommon that dictators seek to interfere with the diffusion of SNSs to squelch internal unrest by cutting off a national network from global digital connections (Howard, et al., 2011) [3].

When governance fails, however, things change. In times of crisis and upheaval, the incentives for adopting the tool surge because its users can get not only informed but also connected and organized, especially when they lack alternative options for civic conversations (Etling, et al., 2010). By the same token, Howard, et al. (2011) argue that digital networks “become [a] crucial component of political communications during other kinds of regime transition, foreign military intervention, natural disasters, and social protests that challenge a regime’s legitimacy” [4]. Accordingly, the adoption of the currently most dominant global SNS, Facebook, is thus likely to be influenced by citizen perceptions of socio-political stability. Furthermore, this phenomenon will be amplified when the public is better able to pursue a higher sense of participatory citizenship in making civic voices heard and to learn the utility of the digital tool from the lead country. In addition, a country’s development in leapfrogging conditions will further boost the societal demands for the SNS in response to political instability.

H3: If a country has higher levels of imports from the U.S., democratic governance, and socio-political instability, then its Facebook adoption will increase.

H4: The effects of socio-political instability on Facebook adoption will be amplified by leapfrogging and democratic conditions.

 

++++++++++

Data

Dependent variable

This study used the numbers of Facebook users for 181 countries observed four times between 2010 and 2013. The present coverage of the research design was a product of data availability on country-level Facebook penetration. Facebook provided daily updated user statistics from over 200 different countries via the advertisement targeting function based on its users’ Internet Protocol addresses as well as their updated profiles (Facebook, 2012). Using this function, I gathered the cross-country data on the estimated numbers of active Facebook users over 13 years of age in December 2010, May 2011, September 2012, and October 2013.

The diffusion process of Facebook is not linear but rather follows a sigmoid pattern, so that the adoption rate should vary across countries according to their stage of market maturation. Also, the development of the global SNS should be dependent on the availability of alternative services and their relative advantages. Thus, for a standardized cross-country comparison, it is important to weight the adoption rate of Facebook against that of the Internet. Doing so allowed for measuring the degree to which a country had Facebook users in relation to the development stage of network readiness.

Accordingly, I constructed a Facebook Adoption Index (FAI) parallel to the Technology Distribution Index (TDI) [5]. Different from TDI’s economic-based scale, FAI weighed a country’s relative share of global Facebook users against its relative share of the world population using the Internet. That is, the FAI for each country, serving as the unit of analysis, was derived from the ratio of the following two ratios: the ratio of Facebook users in a given country to the sum of the users over all countries and the ratio of the country’s number of Internet users to the sum of the Internet population of all the countries in the analysis. Since FAI accounted for each country’s relative share of Internet users over the world, it made it possible to conduct a normalized comparison that controlled for the rapid increase in the worldwide penetration of Facebook use. Taking the natural log of the ratio of the two ratios also normalized its distribution to permit using least-squares linear regression models with a continuous, unbounded dependent variable.

 

equation 1

 

Independent variables

This study put forth disparate explanatory models corresponding to multi-dimensional predictors of FAI: the digital divide, leapfrogging, and socio-political variables. The independent variables were gathered with a one-year lag in assuming their effects on the dependent variable, FAI. It was also presumed that the adoption rate of Facebook in the current year is influenced by the user size in the previous year. Thus, a one-year lagged dependent variable was included in every model. This model specification held constant the initial effect of Facebook adoption until the years of interest while focusing on the recent surge of the service. Table 1 presents the descriptive statistics for the variables.

 

Table 1: Descriptive statistics for all variables.Note: a indicates log-transformed data.
Source: See Data section for data sources.
NameDescriptionMinimumMaximumMeanStandard deviation
FAIaFacebook Adoption Index-3.961.59.06.82
IDIICT Development Index.718.864.272.17
Mobile phonesaTechnology Distribution Index for mobile phone subscriptions-2.082.48.35.93
GDP per capitaGDP per capita based on purchasing power parity (in U.S. dollars)332.8891,387.8916,665.5617,741.74
HDIHuman Development Index.31.94.69.16
UrbanizationaShare of global population living in urban areas.00.14.01.01
Youth populationShare of global population living in urban areas.00.14.01.01
TradeSum of exports and imports of goods and services (percent of GDP)12.07447.2487.5355.58
U.S. importsImports of goods and services from the U.S. (in millions of U.S. dollars).15292,650.536,916.2524,991.89
VoiceVoice and accountability-2.211.75.03.96
StabilityPolitical stability and absence of violence/terrorism-2.682.28.05.98

 

First, the digital divide model took account of socioeconomic, infrastructural, and market openness variables: ICT Development Index (IDI), gross domestic product per capita at purchasing-power parity (GDP per capita, PPP), Human Development Index (HDI), and trade volume. IDI, published annually by the International Telecommunication Union (ITU), is a composite measure of accessibility to, usage intensity of, and skill readiness for ICTs across countries. HDI, published by the United Nations Development Programme, gauges human development based on the average levels of life expectancy, education, and income. GDP per capita (PPP) is a measure of cross-country economic comparisons given the relative standard of living. To assess a country’s market integration into the global network, trade was also included as the ratio of the sum of exports and imports of goods and services to GDP. Information on GDP per capita and trade was available from World Development Indicators (WDI), published annually by the World Bank.

The leapfrogging model took into account the index for the distribution of mobile phones, given a country’s relative share of the global supply of mobile phones weighed against its relative economic power [6]. This TDI was used to account for the recent upsurge in wireless access via mobile devices to digital connectivity. The numbers of mobile-cellular subscriptions were gathered from the WDI. In addition, this model included a global urbanization variable that measured a country’s annual share of the total world population living in urban areas (Crenshaw and Robison, 2006). In evaluating the leapfrogging effects, each country’s youth population (aged 15–24 years) was controlled for. Indeed, the younger are often early adopters of mobile phones, as well as city dwellers, who are more likely to use online social networking tools than the elderly. Information on urban and youth population came from WDI and the UNESCO Institute for Statistics, respectively.

Finally, the socio-political model included country-level indicators of imports from the United States. The estimates of a country’s volume of imports from the U.S. came from the U.S. Census Bureau. Also, this model took into consideration each country’s political and institutional environments with respect to how conducive its governance is to the development of participatory democracy and civil society, as well as how stable it is. Data came from two dimensions of Worldwide Governance Indicators (WGI) in the World Bank database: 1) voice and accountability and 2) political stability. First, voice and accountability captured perceptions of citizens’ ability to participate in civic activities without government interference and to produce an intended result in political processes [7]. Political stability addressed “perceptions of the likelihood that the government will be destabilized or overthrown by unconstitutional or violent means, including politically-motivated violence and terrorism” [8]. Based on a number of proxies, these composite indicators have been measured annually in the standard normal units for not only cross-country but also over-time comparisons: the higher the value, the better the quality of governance the public perceives.

Scaling and missing values

In addition to FAI and the distribution index of mobile phones, some other variables were logged to transform their skewed distribution into the normal one: GDP per capita, trade, urbanization, and imports from the U.S. The IDI, HDI and WGI variables were not log-transformed, because they were already comprised of standardized scores having the normal distribution across countries. Because the model included interaction terms comprised of leapfrogging and socio-political factors, the independent variables were standardized by centering them on their means.

For incomplete data on the number of Internet users and GDP per capita, I filled in missing values using the CIA World Factbook. Even so, some data were still missing, such as trade (81 percent complete) and IDI (83 percent complete). For the remainder of such incomplete data, I conducted multiple imputations using the Amelia II package developed by Honaker, et al. (2012) [9]. Missing values were then averaged out over the datasets resulting from the imputting procedure.

 

++++++++++

Empirical analysis

In examining the hypotheses proposed above, this study used a panel data set of annual observations on 181 countries for the years 2010 to 2013. Using the plm package (Croissant and Millo, 2008), the ordinary least squares (OLS) estimates in linear regression models derived from the panel data set with complete data [10]. One merit of this panel-data analysis is learning about the dynamics of each country’s change in FAI, which can hardly be done from a single cross section.

However, the model needs to address certain concerns. Above all, cross-country variations in FAI may be attributable to country-specific but unobserved effects. Particularly, political culture is likely to shape the extent to which people co-opt the digital technology for social networking although that is hardly captured by the aggregate indicators. More importantly, countries with pre-existing SNSs should differ from have-nots in the rate of Facebook adoption. In some countries, their intranet systems frequently restricted their users from using the global service. Meanwhile, such countries are equipped with strong local SNSs instead, even though Facebook officially had active users in those countries. Especially, domestic SNSs in China have established relative supremacy over U.S.-based SNSs in the market: Qzone and Sina Weibo have become the most active Chinese Web sites, whereas Facebook lags far behind in its national penetration. Therefore, China was excluded from this analysis to determine country-level predictors of FAI [11].

To control for the cross-country heterogeneity, the panel model included country fixed effects using the within-group transformation [12]. This method measured the association between country-specific deviations of the dependent variable and of the explanatory variables from their own time-averaged values. Doing so accounted for the country-specific unobserved effects on FAI, which ought to intervene in the process of technology diffusion. This fixed effects model certainly makes it difficult to identify the effects of any time-invariant conditions such as institutional arrangements and cultures because they are absorbed into country fixed effects. Nevertheless, using the WGI enabled the influence of such slow-moving factors to be estimated because of its small but year-to-year changes [13].

To test any perfect linear combination of the independent variables, the correlation matrix indicated no serious concern about multicollinearity (see Appendix) [14]. Yet, the regression estimates could be still biased, because the error structure violates the Gauss-Markov assumptions in three respects: 1) panel heteroskedasticity, in which countries had different error variances; 2) contemporaneous correlation of the errors, meaning a correlation of the error for one country with the errors for other countries in the same year; and, 3) serially correlated errors for a country (see Beck and Katz, 1995).

In order to address panel heteroskedasticity and spatial autocorrelation of the errors, the OLS estimates were corrected with panel-corrected standard errors for a panel model proposed by Beck and Katz (1995) [15]. The problem of serial autocorrelation was ameliorated by the inclusion of a lagged dependent variable in the model. Furthermore, a first-difference model was run for the sake of robustness testing, because this method works more efficiently to relax the problem of serial autocorrelation in the idiosyncratic errors than the fixed effects estimator does (Wooldridge, 2002) [16].

 

++++++++++

Results

Table 2 presents the findings on hierarchical regressions using the estimated fixed-effects models, which provide support for the robustness of the leapfrogging and socio-political models in predicting FAI [17]. The explanatory leverage of the models is also contingent on political instability that accelerates the adoption of Facebook.

 

Table 2: Fixed effects panel linear models predicting Facebook adoption.Note: N=543 (n=181, T=3); OLS coefficients with panel-corrected standard errors in parentheses.
#p≤0.1; *p≤0.05;**p≤0.01;a indicates a log-transformation.
Variables123456
Lagged FAIa.46**(.05).41** (.05).52** (.05).36** (.05).35** (.05).36** (.05)
Digital divide      
IDI.48**(.12)   .26* (.12).23* (.11).15 (.12)
GDPa-.29 (.22)   .66* (.27).61* (.27).72** (.27)
HDI.35 (.36)   -.42 (.36)-.31 (.35)-.39 (.35)
Tradea-.03 (.02)   .02 (.02).01 (.02).02 (.02)
Leapfrogging      
Mobile phonesa .34** (.10) .50** (.13).42** (.12).46** (.12)
Urbanizationa 6.64** (1.97) 6.34** (1.30)6.60** (1.25)6.50** (1.24)
Youth population .11 (.12) .18 (.12).20# (.12).18 (.12)
Socio-political      
U.S. importsa  .52** (.13) .25* (.12).24* (.12)
Voice  .40* (.16) .49** (.15).52** (.15)
Stability  -.37** (.11) -.34** (.10)-.18 (.11)
Interaction      
Mobile phonesa × stability     -.15* (.07)
Urbanizationa × stability     .14 (.16)
U.S. importsa × stability     -.11 (.11)
Voice × stability     .31** (.12)
Adjusted R2.21.25.22.27.30.31

 

When including the effects of the lagged dependent variable, first, the global digital-divide model (Model 1) accounted for 21 percent of the variation in FAI. Among the component variables, only IDI was significantly associated with FAI. The more development of ICTs a country has, the more it witnessed an increase in the adoption of Facebook with respect to Internet penetration. However, Model 1 showed that none of GDP per capita, HDI, and trade was a significant predictor of FAI. Also, the explanatory power of the digital-divide model explained the variation in FAI to a lesser extent than the leapfrogging (Model 2) and socio-political models (Model 3) do [18]. H1 is partially supported.

Controlling for the youth population as well as the lagged dependent variable, Model 2 showed the positive effects of mobile phone distribution and urbanization on FAI. This leapfrogging model explained 25 percent of the variation in FAI, supporting its explanatory superiority over the digital-divide model. Particularly, wireless Internet connectivity via mobile phones was indeed found to promote the adoption of Facebook in relation to the total Internet population in each country. The urban population also had sizable coefficients to explain the variation in FAI within a country, meaning that urbanization accelerated the diffusion of Facebook. It is therefore certain that the development of such leapfrogging conditions facilitates the accessibility, availability, and even affordability of the global SNS. This finding lends support to H2.

Model 3 confirmed the significance of the socio-political predictors of FAI whose 22 percent of variation was explained by this model including the lagged FAI. As a country augmented the import of goods and services from the U.S., the country’s adoption of Facebook relative to the Internet use increased. The diffusion of the global SNS also expanded when perceptions of political rights and civil liberties became heightened. The socio-political instability was associated with FAI, insofar as the relative utility of the SNS increased under perceived high-risk conditions with respect to governance institutions. Therefore, H3 is confirmed.

Model 4 examined whether the explanatory power of the leapfrogging model endured when controlling for the digital-divide model. The findings lend additional support to H2, so much so that mobile phone distribution and urbanization maintained their positive influence on FAI. That is, as far as the global SNS is concerned, developing countries could leapfrog into a digitally networked society because of their rapid growth of mobile phone penetration and urban agglomerations, above and beyond existing disparities in socioeconomic and infrastructural conditions. Yet this model revealed the positive impact that GDP per capita had on digital-network connectivity, yielding favorable evidence for H1. When the leapfrogging factors are held constant, the digital divide is still observable in the diffusion of Facebook, whose diffusion takes place at a faster rate in a more developed country.

Will the effects of the socio-political variables remain significant beyond the explanatory power of the digital divide and leapfrogging models? Model 5 demonstrates this to be true. In this model, the coefficients of the variables — imports from the U.S., voice and accountability, and political stability — were unchanged in direction, as well as significance, after controlling for the influence of IDI, GDP per capita, the distribution index for mobile phones, and urbanization variables. The results offer additional evidence to support H3.

Regarding H4, Model 6 included a series of interaction terms to test whether the effects of a socio-political crisis and/or upheaval on FAI were contingent on the leapfrogging and democratic institutional factors. As seen in the final column of Table 2, the encouraging effects that perceived the instability of political institutions on Facebook adoption were amplified by the increase in mobile-phone adoption and anti-democratic governance. To achieve a more nuanced explanation of these interaction effects, first, predicted values of FAI were plotted against the degree of mobile-phone adoption under different conditions of perceived political stability. As shown in Figure 2, although Facebook adoption accelerated by increasing mobile-phone adoption, the tendency was more intensified under less stable governance conditions.

 

Relationship between mobile phones and Facebook adoption contingent on political stability
 
Figure 2: Relationship between mobile phones and Facebook adoption contingent on political stability.

 

Moreover, institutional instability moderated the positive impact of the democratic consolidation on Facebook adoption. Figure 3 illustrates how the interaction effect of the two WGI variables predicts FAI: the positive effects of voice and accountability were weakened by a decrease in the perceived level of political stability. Particularly, the diffusion of Facebook in response to fragile and unstable governance was observable only in countries that had more adverse circumstances for democratic institutions. H4 is thus supported.

 

Relationship between voice and accountability and Facebook adoption contingent on political stability
 
Figure 3: Relationship between voice and accountability and Facebook adoption contingent on political stability.

 

Because of the concern about the serial autocorrelation of the error terms, a series of first-difference models was run for testing the robustness of the parameter estimates. As shown in Table 3, doing so yielded similar estimates of the coefficients for the covariates of interest. The effects of mobile-phone adoption, urbanization, and WGI variables on FAI remained the same as those in the fixed-effect models with respect to their direction and significance. The explanatory power of both leapfrogging and socio-political models was also superior to that of the digital-divide model. In fact, the effects of IDI and GDP per capita were hardly observed as significant in the first-difference models, controlling for the leapfrogging or socio-political variables. Unlike the fixed-effects estimator, however, the trade variable became significant while the import from the U.S. was trivialized with respect to their effects on Facebook adoption. Nevertheless, the interaction effect between mobile-phone adoption and political instability was still found effective in predicting FAI. Although its coefficient weakened seriously, the interaction term between the WGI variables was significant at the 90 percent confidence level. By and large, the first-difference estimate gave robustness to the results from the fixed-effects models.

 

Table 3: First-difference panel linear models predicting Facebook adoption.Note: N=543 (n=181, T=3); OLS coefficients with panel-corrected standard errors in parentheses.
#p≤0.1; *p≤0.05;**p≤0.01;a indicates a log-transformation.
Variables123456
Lagged FAIa.04 (.05).03 (.05).04 (.04).04 (.05).03 (.05).06 (.05)
Digital divide      
IDI-.14 (.12)   -.14 (.12)-.16 (.12)-.20# (.12)
GDPa-.39* (.19)   .22 (.26).20 (.26).36 (.26)
HDI.20 (.30)   -.14 (.31)-.04 (.31)-.18 (.31)
Tradea.03 (.02)   .04* (.02).03# (.02).04* (.02)
Leapfrogging      
Mobile phonesa .28** (.09) .34** (.12).29* (.10).30* (.10)
Urbanizationa 2.46* (1.22) 2.68* (1.23)2.94* (1.20)3.11** (1.18)
Youth population .08 (.11) .04 (.11).04 (.11).05 (.11)
Socio-political      
U.S. importsa  .15 (.10) .13 (.10).16 (.10)
Voice  .35* (.15) .36* (.15).45** (.15)
Stability  -.39** (.10) -.36** (.09)-.24* (.10)
Interaction      
Mobile phonesa × stability     -.19** (.07)
Urbanizationa × stability     .15 (.14)
U.S. importsa × stability     -.10 (.09)
Voice × stability     .18# (.11)
Intercept.24** (.03).18** (.02).21** (.02).20** (.03).19** (.03).18** (.03)
Adjusted R2.02.05.06.06.11.14

 

 

++++++++++

Discussion and conclusion

In predicting Facebook adoption at the country level, the results confirm the robust explanation of the leapfrogging and socio-political models. Of course, the existing cross-country disparities in economic affluence and infrastructural development seemed to linger on in the gap where people connect and interact with others via the digital networking channel. Yet the findings highlight that a developing country’s growth in mobile-phone adoption and urbanization could lead to its increased presence on the global SNS given its Internet penetration. Although the expansion of political rights and civil liberties promoted the adoption of the digital networking tool, its diffusion also accelerated in response to the accumulation of institutional failure to ensure political stability. More importantly, Facebook diffusion was enhanced by rapid mobile-phone adoption under increasing political instability. Also, in times of crisis and upheaval, the global SNS diffused at faster rates under institutions that were less democratic.

Admittedly, the expansion of a global digital network presents a headache for some regimes if they do not want political opponents or civil activists to have a channel for making their voices heard nationally and internationally. As seen in the Arab Spring, indeed, the use of SNSs became an integral part of political movements among thousands of protesters dissatisfied with their governments (Howard and Hussain, 2013). The diffusion of Facebook is thus undesirable for dictatorships that do not want people to be armed with the decentralized means of communication and organization. If regimes are capable of controlling network readiness, how can social media such as Facebook still serve as a means of connecting citizens in a wave of revolutionary demonstrations?

Certainly, institutional environments are important in the expansion of digital connectivity. Dictatorship has been indeed effectual in doing so by restraining people from physical Internet access or, alternatively, building firewalls (Deibert, et al., 2010; Milner, 2006). However, in keeping with the “dictator’s dilemma,” authoritarian power-holders have turned to co-opting ICTs for their own development cause. Meanwhile, the adoption of ICTs as an economic engine has complicated national and international business relationships within and between countries; therefore even dictatorships are finding it hard to keep disconnecting from the global network (Howard, et al., 2011). Doing so has made it possible to circumvent the underdeveloped telecommunications infrastructure, including fixed Internet penetration, caused by repressive regulatory environments as well as economic hindrance.

The so-called “leapfrogging” trend has enabled the diffusion of Facebook to transcend economic and infrastructural constraints. It is therefore surprising that existing inequalities in social development measured by HDI are not predictive of FAI in any model. Indeed, Facebook became a dominant platform for online activity in relatively laggard societies for education and life expectancy (for example, Algeria, Gabon, Guatemala, Haiti, Laos, Malaysia, Nigeria, Thailand, and Vietnam) rather than in developed countries. Within such contexts, the concentration of resources and facilities in urban areas has expanded available digital opportunities via wireless or shared Internet access at public stations such as cybercafés. The increasing affordability and accessibility of mobile-phone services in urban agglomerations have allowed the rapid diffusion of a new communication platform for social relationships among city dwellers rather than rural residents.

Even in closed regimes, the dramatic proliferation of mobile phones is no longer an idiosyncratic phenomenon. Rather, an up-to-date surge in wireless connectivity systems has enabled the expansion of digital networks to cut across regime types. The findings corroborate this trend where mobile-phone adoption accelerates the diffusion of Facebook as perceived stability of political institutions become weaker. Assuredly, not all autocratic regimes are troubled with a legitimacy crisis. Rather, some emerging democracies suffer lack of institutional consolidation and loss of public support for a government whose social and economic development policies are neither effective nor accountable. Nevertheless, the findings substantiate that the positive impact of political instability on FAI is found only in perceived less favorable circumstances to democracy and civil society. Thus, it is arguable that when authoritarian regimes expand cutting-edge wireless systems, their failures of stable governance connect more people to the global social network.

Why are such leapfrogging trends troublesome to dictators in spite of the enduring digital divide based on the fixed network technology? First, the relative affordability of wireless connectivity systems makes it difficult for governments to control the diffusion of mobile phones, since it becomes a cultural phenomenon unto itself in that people perceive such devices as a fashion tool (Katz and Sugiyama, 2006). Furthermore, the use of Web-based applications via mobile phones becomes increasingly accessible to anyone, even if personal ownership and use of networked fixed computers are still limited and skill-demanding. Also, Pearce (2011) found that, although many people lack access to the computer network in developing countries, they are given opportunities to use SNSs because of mobile-phone services with little or no access fee. Meanwhile, dictators will find it hard to separate the increasing availability of wireless connectivity in urban agglomerations from their disconnection from the global digital network.

Of course, Facebook adoption is hindered by a weak civil society and a lack of democratic institutions, conditional on the political willpower of the government. At the same time, the system of state controls has evolved into subtle but sophisticated techniques, including censorship and content filtering, based on highly restrictive regulations. The Great Wall of China is a well-known metaphor for a country that has been savvy in limiting the spillover effect of ICT development on the rise of networked civil society in cyberspace, which it considers pro-Western liberalism (Deibert, et al., 2010). Instead, China has driven its development in digital ICTs along with national companies and services. In light of this view, it seems inexorable that the diffusion of SNSs cannot help reinforce existing inequalities in digital connectivity along the lines of government institutions and policies.

Even so, authoritarian power does not contradict the findings of this study. Under restrictive state controls over political rights and civil liberties, people were found to use Facebook to a greater extent when they had a higher level of perceived political instability and social insecurity. This trend indicates how importantly the global SNS serves as a means of communication when widespread political unrest poses a major threat to the stability of authoritarian rule. Especially, SNSs are characterized by decentralization of the processes of message production and circulation rather than using traditional communication channels. The adoption of Facebook satisfies the demands of dissidents and activists for a channel of information and organization relatively independent of top-down regulatory controls over the media environment. In other words, when people are more subject to institutional restrictions on civil society development, the utility of the SNS is more amplified as an unprecedented means for collective action against dictatorship. On the contrary, without such a repressive environment, people’s higher-level perception of political instability does not lead to their increased use of Facebook — so much so that when institutional arrangements support a free flow of information and communication, a number of alternative media channels other than Facebook serve citizen reactions to socio-political unrest.

This paper advances our understanding of why dictators fail in their attempt to secure control over digital networks. Once authoritarian rules lack the stability of governance, that is, they lose the ability to maneuver in obstructing the way to the global SNS. The unstoppable march of mobile-phone distribution weakens further domination of the state over social relations. Is this trend, then, conducive to a wave of democratization such as the Arab Spring? Although people with access seem to co-opt the technological potential for communication and organization at odds with dictatorship, it is uncertain whether social-media use by itself promotes democracy or “freedom” in some abstract sense. Meanwhile, the authoritarian tradition of state intervention and control over digital connectivity has evolved into government policy to counter and contest the public discourse in cyberspace in an effort to discourage political participation and the development of civil society (Deibert, et al., 2012). Future studies should thus explicate the relationship between democratization and social-media adoption in the context of closed regimes.

In addition, the findings are not conclusive but require further study insofar as Facebook may not serve as an index of SNS adoption by country because of country-specific differences in connecting individuals with digital social networks. The presence of strong alternative SNSs in a country hampers its diffusion rate of Facebook adoption. This is likely to be one of the main reasons that the digital-divide model lacks explanatory power compared with the leapfrogging model. When countries lag behind in socioeconomic development and ICTs infrastructure, mobile-phone distribution drives the dominance of the U.S.-based SNS in the market without local competitors. In this regard, a qualitative comparison of similar country cases with regard to the SNS market could identify whether the leapfrogging factors are sufficient conditions for digitally enabled social relationships against government suppression. Moreover, within-case analysis may observe a different causal process in which governments control and manipulate social-media use.

Even though it is beyond the scope of the present study, future studies should expound the significance of imports from the U.S. in Facebook adoption. From the perspective of media imperialism, it calls for a more in-depth observation of whether economic connectedness leads to political and cultural spillovers. Such country-specific analyses might contradict the findings of this study, given their focus on the context-embedded process of causality rather than a generalizable systematic relation among the variables. Nevertheless, the Facebook diffusion pattern can be contextual to the extent that pre-existing social structures engage in its use. The small-N approach is thus necessary to complement the cross-country regression analysis in inferring the nature of causal order to explain Facebook diffusion as a global as well as a local phenomenon. End of article

 

About the author

Shin Haeng Lee is a Ph.D. candidate in the Department of Communication at the University of Washington, Seattle. He has a M.A. in Journalism from Indiana University-Bloomington.
E-mail: shinl2 [at] uw [dot] edu

 

Acknowledgements

The author thanks Professor Philip N. Howard for his helpful comments.

 

Notes

1. I plotted 182 countries to show the relation between their current Facebook and Internet penetration in 2012, according to each country’s political institution. The Freedom in the World Index, published by Freedom House (2012), provided a cross-country freedom index based on their institutional conditions for political rights and civil liberties.

2. Shih and Chang, 2009, p. 823; see also Ganesh and Kumar, 1996.

3. Howard, et al. (2011) have shown that authoritarian and even democratic regimes often engage in “blocking Internet access and disabling digital networks,” beyond mere censorship of digital content, in response to social and political unrest.

4. Howard, et al., 2011, p. 230.

5. For more discussion of the expression for calculating a value for TDI, see Howard, et al. (2009). The index scaled a country’s relative share of the global technology supply in proportion to its relative share of the global economic output.

6. According to Howard, et al. (2009), this index was based on the proportion of two ratios: a country’s relative share of the global technology supply and its relative share of the global economic output. Logarithmic transformation of the measure was also performed for normalization of the distribution.

7. Some might reasonably question the validity of voice and accountability in the WGI as the indicator of democratic institutions, because it measures subjective perceptions of governance at the aggregate level rather than assessing objective measures on the basis of formal laws and regulations (see Norris, 2002). Yet, Kraay, et al. (2010) argue that the WGI’s reliance on perceptions data is better to capture the reality of governance on the ground of actual experiences within the political context than the coding of laws. To accommodate potential perception biases, they also exploit a large number of individual data sources in constructing and standardizing the measures of governance. Indeed, the WGI’s composite indicator, voice and accountability, was highly correlated with the indices of the development of democratic institutions for 166 countries between 2009 and 2011 from Freedom House and Polity IV datasets (Spearman’s rho: .98 and .91, respectively).

8. Kraay, et al., 2010, p. 4.

9. Amelia II imputed five values, as default, for each missing cell in the panel data and thus generated five completed datasets. The program is available at http://gking.harvard.edu/amelia/.

10. Including a lagged dependent variable reduced the total N in analysis to 543 with 181 countries over three years, since it produced no data on the first observed year, 2010.

11. When it comes to China, its country-specific but unobserved effects are too consequential to be controlled for by including fixed effects at the country level. Indeed, despite its rapid development in digital connectivity, Facebook adoption in China remains near to zero given its Internet penetration. However, a post hoc analysis confirmed that the inclusion of the country would not have changed the results in any significant manner.

12. The Hausman test also endorses the preferred model of fixed effects rather than random effects, since the country-specific unobserved effects were correlated with the regressors. See Wooldridge (2002) for more discussion.

13. Note that Kraay, et al. (2010) warned against focusing on short-run year-to-year changes in the WGI instead of its long-term trends.

14. Pearson’s correlation coefficients were calculated using “demeaned” data, where every variable was transformed by subtracting the average over time within each country.

15. Croissant and Millo (2008) developed a function, Beck and Katz Robust Covariance Matrix Estimators, for panel models, which work with the plm package.

16. Technically, the first-difference estimator subtracted the previous within-unit data for each observation to assess any association between country-specific, one-period variations in dependent and explanatory variables. This method is especially effective when the country-specific unobserved effects that change over time are serially correlated.

17. In interpreting the findings, it should be noted that FAI weighs the number of Facebook users in each country against that of Internet users. Also, the fixed-effects analysis takes no account of any temporal cross-sectional relationship. This is the reason that some explanatory variables appear insignificant in their effect on FAI despite their presumable influence on the cross-country variation in the adoption of ICTs.

18. Adjusted R2 allows for standardized comparison of the explanatory power of regression models with different numbers of parameters.

 

References

Luis Andrés, David Cuberes, Mame Diouf, and Tomás Serebrisky, 2010. “The diffusion of the Internet: A cross-country analysis,” Telecommunications Policy, volume 34, numbers 5–6, pp. 323–340.
doi: http://dx.doi.org/10.1016/j.telpol.2010.01.003, accessed 3 May 2015.

Nathaniel L. Beck and Jonathan N. Katz, 1995. “What to do (and not to do) with time-series cross-section data,” American Political Science Review, volume 89, number 3, pp. 634–647.
doi: http://dx.doi.org/10.2307/2082979, accessed 3 May 2015.

Richard Beilock and Daniela V. Dimitrova, 2003. “An exploratory model of inter-country Internet diffusion,” Telecommunications Policy, volume 27, numbers 3–4, pp. 237–252.
doi: http://dx.doi.org/10.1016/S0308-5961(02)00100-3, accessed 3 May 2015.

W. Lance Bennett and Alexandra Segerberg, 2011. “Digital media and the personalization of collective action: Social technology and the organization of protests against the global economic crisis,” Information, Communication & Society, volume 14, number 6, pp. 770–799.
doi: http://dx.doi.org/10.1080/1369118X.2011.579141, accessed 3 May 2015.

Margarita Billon, Rocio Marco, and Fernando Lera-Lopez, 2009. “Disparities in ICT adoption: A multidimensional approach to study the cross-country digital divide,” Telecommunications Policy, volume 33, numbers 10–11, pp. 596–610.
doi: http://dx.doi.org/10.1016/j.telpol.2009.08.006, accessed 3 May 2015.

Francesco Caselli and Wilbur John Coleman, 2001. “Cross-country technology diffusion: The case of computers,” American Economic Review, volume 91, number 2, pp. 328–335.
doi: http://dx.doi.org/10.1257/aer.91.2.328, accessed 3 May 2015.

Menzie D. Chinn and Robert W. Fairlie, 2006. “The determinants of the global digital divide: A cross-country analysis of computer and Internet penetration,” Oxford Economic Papers, volume 59, number 1, pp. 16–44.
doi: http://dx.doi.org/10.1093/oep/gpl024, accessed 3 May 2015.

Vincenzo Cosenza, “World map of social networks,” at http://vincos.it/world-map-of-social-networks/, accessed 3 October 2012.

Javier Corrales and Frank Westhoff, 2006. “Information technology adoption and political regimes,” International Studies Quarterly, volume 50, number 4, pp. 911–933.
doi: http://dx.doi.org/10.1111/j.1468-2478.2006.00431.x, accessed 3 May 2015.

Edward M. Crenshaw and Kristopher K. Robison, 2006. “Globalization and the digital divide: The roles of structural conduciveness and global connection in Internet diffusion,” Social Science Quarterly, volume 87, number 1, pp. 190–207.
doi: http://dx.doi.org/10.1111/j.0038-4941.2006.00376.x, accessed 3 May 2015.

Yves Croissant and Giovanni Millo, 2008. “Panel data econometrics in R: The plm package,” Journal of Statistical Software, volume 27, number 2, pp. 1–43, and at http://www.jstatsoft.org/v27/i02/, accessed 3 May 2015.

Ronald Deibert, John G. Palfrey, Rafal Rohozinski, and Jonathan Zittrain (editors), 2012. Access contested: Security, identity, and resistance in Asian cyberspace information revolution and global politics. Cambridge, Mass.: MIT Press.

Ronald Deibert, John G. Palfrey, Rafal Rohozinski, and Jonathan Zittrain (editors), 2010. Access controlled: The shaping of power, rights, and rule in cyberspace. Cambridge, Mass.: MIT Press.

Larry J. Diamond, 1994. “Rethinking civil society: Toward democratic consolidation,” Journal of Democracy, volume 5, number 3, pp. 4–17.
doi: http://dx.doi.org/10.1353/jod.1994.0041, accessed 3 May 2015.

Jan van Dijk, 2005. The deepening divide: Inequality in the information society. Thousand Oaks, Calif.: Sage.

Gili S. Drori and Yong Suk Jang, 2003. “The global digital divide: A sociological assessment of trends and causes,” Social Science Computer Review, volume 21, number 2, pp. 144–161.
doi: http://dx.doi.org/10.1177/0894439303021002002, accessed 3 May 2015.

Nicole B. Ellison, Charles Steinfield, and Cliff Lampe, 2007. “The benefits of Facebook ‘friends:’ Social capital and college students’ use of online social network sites,” Journal of Computer-Mediated Communication, volume 12, number 4, pp. 1,143–1,168.
doi: http://dx.doi.org/10.1111/j.1083-6101.2007.00367.x, accessed 3 May 2015.

Bruce Etling, John Kelly, Robert Faris, and John Palfrey, 2010. “Mapping the Arabic blogosphere: Politics and dissent online,” New Media & Society, volume 12, number 8, pp. 1,225–1,243.
doi: http://dx.doi.org/10.1177/1461444810385096, accessed 3 May 2015.

Facebook, 2012. “Advertise on Facebook,” at http://www.facebook.com/advertising, accessed 2 October 2012.

Geoffrey A. Fowler, 2012. “Facebook: One billion and counting,” Wall Street Journal (4 October), at http://online.wsj.com/article/SB10000872396390443635404578036164027386112.html, accessed 4 October 2012.

Jaishankar Ganesh and Vipin Kumar, 1996. “Capturing the cross-national learning effect: An analysis of an industrial technology diffusion,” Journal of the Academy of Marketing Science, volume 24, number 4, pp. 328–337.
doi: http://dx.doi.org/10.1177/0092070396244004, accessed 3 May 2015.

Mauro F. Guillén and Sandra L. Suárez, 2005. “Explaining the global digital divide: Economic, political and sociological drivers of cross-national Internet use,” Social Forces, volume 84, number 2, pp. 681–708.
doi: http://dx.doi.org/10.1353/sof.2006.0015, accessed 3 May 2015.

Eszter Hargittai, 1999. “Weaving the Western Web: Explaining differences in Internet connectivity among OECD countries,” Telecommunications Policy, volume 23, numbers 10–11, pp. 701–718.
doi: http://dx.doi.org/10.1016/S0308-5961(99)00050-6, accessed 3 May 2015.

Eszter Hargittai and Yuli P. Hsieh, 2013. “Digital inequality,” In: William H. Dutton (editor). Oxford handbook of Internet studies. Oxford: Oxford University Press.
doi: http://dx.doi.org/10.1093/oxfordhb/9780199589074.013.0007, accessed 3 May 2015.

James Honaker, Gary King, and Matthew Blackwell, “Amelia II: A program for missing data,” at http://gking.harvard.edu/amelia/, accessed 10 October 2012.

Philip N. Howard and Muzammil M. Hussain, 2013. Democracy’s fourth wave? Digital media and the Arab Spring. New York: Oxford University Press.

Philip N. Howard and Nimah Mazaheri, 2009. “Telecommunications reform, Internet use and mobile phone adoption in the developing world,” World Development, volume 37, number 7, pp. 1,159–1,169.
doi: http://dx.doi.org/10.1016/j.worlddev.2008.12.005, accessed 3 May 2015.

Philip N. Howard, Sheetal D. Agarwal, and Muzammil M. Hussain, 2011. “When do states disconnect their digital networks? Regime responses to the political uses of social media,” Communication Review, volume 14, number 3, pp. 216–232.
doi: http://dx.doi.org/10.1080/10714421.2011.597254, accessed 3 May 2015.

Philip N. Howard, Ken Anderson, Laura Busch, and Dawn Nafus, 2009. “Sizing up information societies: Toward a better metric for the cultures of ICT adoption,” Information Society, volume 25, number 3, pp. 208–219.
doi: http://dx.doi.org/10.1080/01972240902848948, accessed 3 May 2015.

Shanthi Kalathil and Taylor C. Boas, 2003. Open networks, closed regimes: The impact of the Internet on authoritarian rule. Washington, D.C.: Carnegie Endowment for International Peace.

James E. Katz and Satomi Sugiyama, 2006. “Mobile phones as fashion statements: Evidence from student surveys in the US and Japan,” New Media & Society, volume 8, number 2, pp.321–337.
doi: http://dx.doi.org/10.1177/1461444806061950, accessed 3 May 2015.

Christopher Kedzie, 2002. “Coincident revolutions and the dictator’s dilemma: Thoughts on communication and democratization,” In: Juliann E. Allison (editor). Technology, development, and democracy: International conflict and cooperation in the information age. Albany, N.Y.: SUNY Press, pp. 105–130.

Aart Kraay, Daniel Kaufmann, and Massimo Mastruzzi, 2010. “The worldwide governance indicators: Methodology and analytical issues,” World Bank Policy Research Working Paper, number 5430.
doi: http://dx.doi.org/10.1596/1813-9450-5430, accessed 3 May 2015.

Vijay Mahajan and Eitan Muller, 1994. “Innovation diffusion in a borderless global market: Will the 1992 unification of the European community accelerate diffusion of new ideas, products, and technologies?” Technological Forecasting and Social Change, volume 45, number 3, pp. 221–235.
doi: http://dx.doi.org/10.1016/0040-1625(94)90047-7, accessed 3 May 2015.

Helen V. Milner, 2006. “The digital divide: The role of political institutions in technology diffusion,” Comparative Political Studies, volume 39, number 2, pp. 176–199.
doi: http://dx.doi.org/10.1177/0010414005282983, accessed 3 May 2015.

Miniwatts Marketing Group, 2012. “Facebook users in the world,” at http://www.internetworldstats.com/facebook.htm, accessed 2 October 2012.

Evgeny Morozov, 2011. The Net delusion: The dark side of Internet freedom. New York: Public Affairs.

Pippa Norris, 2002. Democratic phoenix: Reinventing political activism. New York: Cambridge University Press.

Pippa Norris, 2001. Digital divide: Civic engagement, information poverty, and the Internet worldwide. New York: Cambridge University Press.

Katy E. Pearce, 2011. “Convergence through mobile peer-to-peer file sharing in the Republic of Armenia,” International Journal of Communication, volume 5, pp. 511–528, and at http://ijoc.org/index.php/ijoc/article/view/800/536, accessed 3 May 2015.

Katy E. Pearce and Ronald E. Rice, 2013. “Digital divides from access to activities: Comparing mobile and personal computer Internet users,” Journal of Communication, volume 63, number 4, pp. 721–744.
doi: http://dx.doi.org/10.1111/jcom.12045, accessed 3 May 2015.

Katy E. Pearce and Sarah Kendzior, 2012. “Networked authoritarianism and social media in Azerbaijan,” Journal of Communication, volume 62, number 2, pp. 283–298.
doi: http://dx.doi.org/10.1111/j.1460-2466.2012.01633.x, accessed 3 May 2015.

Robert D. Putnam, 2000. Bowling alone: The collapse and revival of American community. New York: Simon & Schuster.

Kristopher K. Robison and Edward M. Crenshaw, 2010. “Reevaluating the global digital divide: Socio-demographic and conflict barriers to the Internet revolution,” Sociological Inquiry, volume 80, number 1, pp. 34–62.
doi: http://dx.doi.org/10.1111/j.1475-682X.2009.00315.x, accessed 3 May 2015.

Everett M. Rogers, 2003. Diffusion of innovations. Fifth edition. New York: Free Press.

Hsin-Yu Shih and Tung-Lung Steven Chang, 2009. “International diffusion of embodied and disembodied technology: A network analysis approach,” Technological Forecasting and Social Change, volume 76, number 6, pp. 821–834.
doi: http://dx.doi.org/10.1016/j.techfore.2008.09.001, accessed 3 May 2015.

Jeffrey M. Wooldridge, 2002. Econometric analysis of cross section and panel data. Cambridge, Mass.: MIT Press.

Jonathan Zittrain and Benjamin Edelman, 2003. “Internet filtering in China,” IEEE Internet Computing, volume 7, number 2, pp. 70–77.
doi: http://dx.doi.org/10.1109/MIC.2003.1189191, accessed 3 May 2015.

 

Appendix

 

Correlation coefficients for all variables

 

 


Editorial history

Received 16 June 2014; revised 5 April 2015; accepted 10 May 2015.


Creative Commons License
This paper is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Can regimes really discourage social networking? Urbanization, mobile phone use, and the dictator’s plight
by Shin Haeng Lee.
First Monday, Volume 20, Number 5 - 4 May 2015
http://www.firstmonday.dk/ojs/index.php/fm/article/view/5420/4486
doi: http://dx.doi.org/10.5210/fm.v20i5.5420





A Great Cities Initiative of the University of Illinois at Chicago University Library.

© First Monday, 1995-2017. ISSN 1396-0466.