Journal of Communication ISSN 0021-9916 ORIGINAL ARTICLE Networked Authoritarianism and Social Media in Azerbaijan Katy E. Pearce1 & Sarah Kendzior2 1 Department of Communication, University of Washington, Seattle, WA 98195-3740, USA 2 Department of Anthropology, Washington University, Saint Louis, MO 63130, USA The diffusion of digital media does not always have democratic consequences. This mixed- methods study examines how the government of Azerbaijan dissuaded Internet users from political activism. We examine how digital media were used for networked authoritarianism, a form of Internet control common in former Soviet states where manipulation over digitally mediated social networks is used more than outright censorship. Through a content analysis of 3 years of Azerbaijani media, a 2-year structural equation model of the relationship between Internet use and attitudes toward protest, and interviews with Azerbaijani online activists, we find that the government has successfully dissuaded frequent Internet users from supporting protest and average Internet users from using social media for political purposes. doi:10.1111/j.1460-2466.2012.01633.x Many assume that greater exposure to information technology leads to increased activism in authoritarian states (Kedzie, 1997; Zuckerman, 2008). From Reagan’s proclamation that the ‘‘Goliath of totalitarianism will be brought down by the David of the microchip’’ to Secretary of State Clinton’s ‘‘bet that an open Internet will lead to stronger, more prosperous countries,’’ this assumption has structured policy as well as scholarship. Faith in the Internet’s potential to drive activism in authoritarian states is often fueled by a belief that information technology ended the Cold War. The Internet has been compared to Soviet-era samizdat, self-published works that promoted dissident activity. Advocates of this analogy argue that protests are always worthwhile so long as you have a means to publicize them, as even suppressed protests create a sense of solidarity. Shirky (2007, p. 164) argued that documenting dissent in East Germany in 1989 created a dilemma for the government: ‘‘if the state didn’t react, the documentation would serve as evidence that the protesting was safe. If the state did react, then the documentation of the crackdown could be used to spur an international outcry.’’ He went on to claim that the same conditions apply to the Internet era more generally. Corresponding author: Katy E. Pearce; e-mail: kepearce@uw.edu Journal of Communication (2012) © 2012 International Communication Association 1 Social Media in Azerbaijan K. E. Pearce & S. Kendzior We argue the opposite: that greater documentation and publicizing of suppressed dissent can derail political protest. This is particularly evident in the authoritarian countries of the former Soviet Union—the very countries, ironically, that fuel the misguided Cold War analogy. Often neglected in analyses of the Internet, these countries have a unique approach to Internet regulation that represents a ‘‘middle path’’ between open access and censorship. Their approach exploits problems of cynicism, insecurity, and trust particular to post-Soviet political culture (de Waal, 2011; Gutbrod, 2011; Kendzior, 2011). This practice, commonly found in the former Soviet states, fits with our research finding that the Azerbaijanis most active online were the ones whose attitude to protest was most negatively affected by crackdowns. Networked authoritarianim occurs when ‘‘an authoritarian regime embraces and adjusts to the inevitable changes brought by digital communications’’ (MacKinnon, 2011, p. 33). States that practice networked authoritarianism do not strictly censor online dissent: they compete with it, making an example out of online dissenters in order to affirm the futility of activism to a disillusioned public. Using networked authoritarianism as a frame, this study looked at social media and online activism in Azerbaijani television, print, and Internet media sources (via World News Connection) from January 2009 to August 2011. We also examined public opinion data from 2009 and 2010, and found that between the 2 years, frequent Internet users became significantly less supportive of protests against the government, indicating that the government’s campaign against online activism was successful. Interviews conducted with Azerbaijani online activists validated these findings. While Azerbaijani government has not completely deterred citizens from using social media (although social media adoption is significantly less and growing at a slower rate than all of its poor but less authoritarian neighbors), they have changed the attitudes of frequent Internet users toward dissent. Our findings challenge the conventional wisdom that access to the Internet encourages support for dissent. Instead they imply that scholars should consider the political cultures of authoritarian systems before assuming the Internet offers an effective means to contest them. Our focus on former Soviet states is notable because, as Deibert and Rohozinski (2010) argue, their tactics are increasingly emulated by other countries. First, we outline the particularities of post-Soviet authoritarianism and describe online activism in Azerbaijan. Second, we describe the Azerbaijani government’s successful campaign against online activism and provide quantitative data linking increased Internet use with support for protests. Third, we detail the Internet regulation policies of post-Soviet authoritarian states: the networked authoritarianism that competes with online activism. We conclude with a description of how the Azerbaijani government has demonized not only Internet activism, but also social media use in general. Authoritarianism and digital media in Azerbaijan Authoritarian post-Soviet states are shaped by political practices specific to the region. Hale (2005) argues that former Soviet states practice ‘‘patronal presidentialism,’’ in 2 Journal of Communication (2012) © 2012 International Communication Association K. E. Pearce & S. Kendzior Social Media in Azerbaijan which the presidency has both great formal powers and informal power based on patron–client relationships at the intersection of the state and economy. He notes that while politics in post-Soviet authoritarian states remains an elite affair, mass opinion should not be discounted, because the masses have an ‘‘independent and often important role in deciding outcomes’’ in political contestation (p. 162). Appearing legitimate to the public is an essential concern of post-Soviet leaders despite their massive power. After the Soviet Union collapsed, the presidents of the newly independent states cultivated nationalist ideologies that proclaimed their countries’ primordial roots and teleological progressions toward independence. March (2003) argues that this form of post-Soviet authoritarian legitimation is rooted in ‘‘a consistent rejection of the existence of conceivable ideological alternatives to the substantive orientation of the regime’’ (p. 210). The desire for political uniformity is echoed in the patronage networks of post-Soviet leaders, particularly in the Aliyev dynasty of Azerbaijan (Guliyev, 2012). The Internet, both unpredictable and a prime venue of unsanctioned content, threatens what post-Soviet authoritarian states value most: power through consistency, consistency through power. Azerbaijan has an economy dominated by fuel exports, allowing the regime to preempt any opposition (Guliyev, 2009), and it has been dominated by a father and son, Heydar and Ilham Aliyev, since 1993. The younger Aliyev has suppressed opposition groups and violated basic rights and civil liberties (Grono, 2011). Scholars describe Azerbaijani citizens as marked by a ‘‘pervasive bitterness and growing sense of deprivation’’ (Rasizade, 2003, p. 127), a general sense of apathy and fear (Abbasov, 2010), and a lack of trust in others (Gahramanova, 2009). In March 2009 presidential term limits were removed, further entrenching the regime and disillusioning the public and the opposition. The intimidation tactics the state uses to control the media in Azerbaijan have added to a self-censoring and fearful society (Gahramanova, 2009). Broadcast media is owned by or linked to the government and even commercial stations serve the needs of the state (Guliyeva, 2005; Oates, 2008). Due in part to restrictions on media freedom, public opposition to the government is rare. Citizens are further disinclined to engage in political protest against the government because they are preoccupied with economic concerns (Gahramanova, 2009; Rasizade, 2003). Many Azerbaijanis who lived through the chaotic 1990s are unwilling to risk jeopardizing the stability and economic well-being that they now enjoy. Meanwhile, the majority of Azerbaijanis live in poverty (Pearce, 2011). Our interview subjects noted that while ‘‘everyone’’ has complaints against the government, most Azerbaijanis depend on it directly or indirectly for employment and thus are more reluctant to oppose the system. By 2010, the Internet had slowly reached about a third of Azerbaijanis, but most of the frequent users were wealthier, better educated, younger urbanites, often with higher English proficiency (Pearce & Rice, forthcoming). Azerbaijan does little direct filtering of Internet content, although they have sporadically filtered opposition news sources, especially before elections (Grono, 2011; OpenNet Initiative, 2010). Social media users interviewed for this study report that both human and software Journal of Communication (2012) © 2012 International Communication Association 3 Social Media in Azerbaijan K. E. Pearce & S. Kendzior monitoring systems are watching ‘‘everyone’’ on Facebook in Azerbaijan. Users have had their Facebook accounts as well as Yahoo! and Gmail e-mail accounts compromised. Activists as well as politically disengaged citizens are monitored. The users to whom we spoke said that while posting ‘‘Traffic in Baku is bad’’ would not cause trouble, a status update such as ‘‘Traffic in Baku is bad because the traffic police are corrupt’’ might cause someone to mysteriously lose their job. Merely complaining is enough to feel the consequences of state monitoring. Making an example out of a donkey A vocal, if fractious, opposition existed in Azerbaijan prior to 2005 when the government violently cracked following parliamentary elections (Gahramanova, 2009). As opposition newspaper readership is low and limited largely to Baku, the Internet remains the only space for political activism. That said, the 4% of Azerbaijanis who use the Internet daily are not representative of the general population. Compared to other citizens, these young elites, most of whom live in the capital city of Baku, are better educated, more likely to have studied abroad, and more likely to have exposure to technology through globalized workplaces (Pearce & Rice, forthcoming). These early adopters of social media maintained close relationships online and were often involved in civil society organizations together offline. Although social media use has expanded their social ties, they have remained a tightly knit group. In early September 2009, two members of this group were arrested. Adnan Hajizada and Emin Milli were from middle-class families in Baku and had been educated in the United States and Europe. Both were employees of international organizations and had helped spearhead youth activist groups in 2005 and 2006. In 2009, they produced and posted a YouTube video parodying the government for spending hundreds of thousands of dollars to import donkeys from Germany. In the video, a group of solemn journalists interview a donkey (Hajizada in costume) and note that this donkey would be afforded more civil liberties than Azerbaijani citizens (the ‘‘donkey rights’’ video may be found, with English subtitles, at http://www.youtube.com/watch?v=Aaecvg7xCIk): ‘‘There will be someone to protect donkey rights,’’ the video concludes, ‘‘but what of human rights?’’ Two weeks later, Hajizada and Milli were having dinner with friends when they were attacked by two unknown men. They went to the police, assuming that the incident would be investigated. Instead they were arrested for ‘‘hooliganism’’ and sentenced to 30 and 24 months imprisonment, respectively, on November 11, 2009. Appeals by human rights groups and foreign governments in late 2009 and through most of 2010 were denied. Our interviewees speculated that the government was surprised at the international attention that the case received, but needed to save face an internal audience of online activists. After intense international pressure, they were released in November 2010. The ‘‘donkey blogger’’ arrest was widely reported on Web sites frequented by young Azerbaijanis. Over the months that followed, support for activism dropped precipitously among this population, despite their political affinity with Hajizada 4 Journal of Communication (2012) © 2012 International Communication Association K. E. Pearce & S. Kendzior Social Media in Azerbaijan and Milli. This reversal challenges assumptions that an ‘‘open Internet’’ and ‘‘trans- parency’’ increase support for activism. By publicizing the reprisals for even mild, humorous forms of dissent, the government provoked anxiety among Azerbaijanis insecure about the future and hesitant to engage in protest. The government would not have been able to instill this fear had it merely censored Milli and Hajizada. Only by making the Internet open could they reach the frequent Internet users who had become a source of consternation. As Milli remarked, ‘‘This is the way they function. . .They punish some people and let everyone else watch. To say, ‘This is what can happen to you.’ ’’ Networked authoritarianism The regime’s campaign against the ‘‘donkey bloggers’’ was perpetrated through a series of practices called ‘‘networked authoritarianism’’ (MacKinnon, 2011). In states which practice networked authoritarianism: the average person with Internet or mobile access has a much greater sense of freedom - and may feel that he has the ability to speak and be heard - in ways that were not possible under classic authoritarianism. At the same time, in the networked authoritarian state, there is no guarantee of individual rights and freedoms. Those whom the rulers see as threats are jailed; truly competitive, free, and fair elections are not held; and the courts and the legal system are tools of the ruling party. (MacKinnon, 2011, p. 33) Networked authoritarianism is manifest in political tactics that create selective social openings to create a semblance of transparency but in fact monitor and stifle dissent (Baogang & Warren, 2011). From a historical perspective, networked authoritarianism represents a third generation of strategies for censorship and control of the Internet (Deibert & Rohozinski, 2010). The first generation is characterized by widespread filtering and other attempts at direct censorship. These were rarely exclusively practiced in the former Soviet Republics (today’s Commonwealth of Independent States [CIS] countries). Second-generation controls involving the manipulation of law to regulate Internet content are used in the CIS countries. Specific tactics include redefining what is acceptable content within the national media space, and most notably for our purposes, ‘‘expanded use of defamation, slander, and ‘veracity’ laws, to deter bloggers and independent media from posting material critical of the government or specific government officials, however benignly (including humor)’’ (Deibert & Rohozinski, 2010, p. 25). Third-generation controls do not attempt to control Internet access, but to compete with it ‘‘through effective counter information campaigns that overwhelm, discredit, or demoralize opponents’’ (Deibert & Rohozinski, 2010, p. 27). Second- and third-generation controls were used to great effect in the ‘‘donkey blogger’’ affair in Azerbaijan. Journal of Communication (2012) © 2012 International Communication Association 5 Social Media in Azerbaijan K. E. Pearce & S. Kendzior The approach taken to Internet regulation in the former Soviet Republics that comprise today’s CIS may be distinct because their governments ‘‘officially laud Internet usage, but employ an iron fist against violators of these countries’ Byzantine regulations of ICTs’’ (Saunders, 2009, p. 14). Like all authoritarian states, CIS countries use the Internet to maximize surveillance of dissident populations and to block content from reaching the masses (Kalathil & Boas, 2003; Morozov, 2011). But they are unique in that they want to ‘‘effect cognitive change rather than to deny access to online information or services’’ (Deibert & Rohozinski, 2010, p. 28). That is, they exploit the pre-existing political culture of the population, which we argue is characterized by cynicism, apathy, and an aversion to risk (de Waal, 2011; Kendzior, 2011). Anticipating the role of the Internet in the 2011 Arab uprisings, Howard (2010) posed the question, ‘‘What is the regime countermeasure for the chilling effects of a plea from someone in your social network who has been a victim of police brutality?’’ (p. 10). One answer was provided by the Azerbaijani government’s tactics in the ‘‘donkey bloggers’’ case. Adnan Hajizada and Emin Milli in Azerbaijan were victims of police brutality, but their victimization served to dissuade rather than incite protest among their online peers. In 2009, before the arrest, one Azerbaijani frequent Internet user compared going on Facebook to being in the movie The Matrix, where the fog of apolitical ‘‘real life’’ was lifted and political problems were confronted. The government’s online campaign against social media activists and on-the-ground use of violence against them reminded frequent Internet users that there was no barrier between the virtual world and real life: both were in the control of the regime. Networked authoritarianism is particularly effective on a population that views the Internet as a refuge from their political reality instead of as a means to transform it. Internet use and attitudes toward political protest Azerbaijanis interviewed for this study agreed that the government’s handling of the affair made frequent Internet users afraid of supporting political protest, but we also examined survey data in search of broader evidence to support our belief that networked authoritarianism was effective in suppressing protest in Azerbaijan. Although there are no data available to demonstrate a direct causal relationship, we were able to compare the results of two surveys taken, one taken near the beginning of the ‘‘donkey blogger’’ case and one conducted a year later. Previous research, though limited, suggests that the same variables that predict Internet use also predict support for protests. Thus we consider the relationship that these variables have with both the willingness to support protests and frequent Internet use. Age, for example, is associated with the willingness to support protests (Hall, Rodeghier, & Useem, 1986; Rodeghier, Hall, & Useem, 1991) as well as with Internet use (Katz & Rice, 2002). Educational attainment is positively related to willingness to support protests because education increases commitment to civil liberties, decreasing support of use of violence, increasing knowledge of protestors’ 6 Journal of Communication (2012) © 2012 International Communication Association K. E. Pearce & S. Kendzior Social Media in Azerbaijan grievances (Hall et al., 1986; Rodeghier et al., 1991). Educational attainment is also positively related to Internet use (Katz & Rice, 2002; Rice & Katz, 2003). We also argue that economic consumption ability (specifically operationalized as the lack of material deprivation) has an influence on willingness to support protests because in the CIS, and in particular Azerbaijan, governments use economic well-being to appease their citizens and discourage opposition. Socioeconomic status is also related to Internet use (Chinn & Fairlie, 2006; Katz & Rice, 2002) and we extrapolate that lack of material deprivation would operate similarly. On the basis of these considerations, we examined the association of age, house- hold economic status (as consumption ability), and education with the frequency of Internet use and the willingness to support protests in late 2009 and again in late 2010. Specifically, we hypothesized that age would be negatively correlated with the frequency of Internet use (H1), that economic consumption ability would be positively correlated with the frequency of Internet use (H2), and that educational attainment would also be positively correlated with the frequency of Internet use (H3). In addition, we hypothesized that age (H4) and economic consumption (H5) would both be negatively correlated with support for protests, while educational attainment (H6) and the frequency of Internet use (H7) would both be positively correlated with support for protests. Method Sample and procedures Respondents were the heads of households in Azerbaijan answering a face-to-face survey, funded by the Carnegie Foundation, which has been administered by the Caucasus Research Resource Center once every fall since 2006. Participation in the nationally representative survey was voluntary and anonymous. The surveys were completed by 1,795 respondents in November-December 2009 and by 2,001 respondents in November–December 2010. Methodological details are available from the authors or from the CRRC Web site (http://www.crrccenters.org/). Measures Age Respondents were asked to report their year of birth. Age was calculated as the difference between their year of birth and the year in which the survey was taken (2009 or 2010). Respondents averaged just over 43 years of age in both the 2009 (M = 43.37, SD = 15.59) and 2010 (M = 43.13, SD = 15.62) samples. Economic consumption ability—Material deprivation Respondents were asked to select one of five phrases that best described their family’s financial situation. The phrases were ordered (1–5) to reflect increasing material and financial security: 1 = We don’t have enough money even for food; 2 = We have enough money for food but not for clothes; 3 = We can buy food and clothes, but not more Journal of Communication (2012) © 2012 International Communication Association 7 Social Media in Azerbaijan K. E. Pearce & S. Kendzior expensive things; 4 = We can buy some more expensive things like a TV or refrigerator; 5 = We can buy anything we want. Responses ranged across the entire scale, but, averaging across the two surveys, 27% reported not having enough money for food and 36% reported not having enough money for clothing. Scale mean values were 2.15 (SD = 0.98) for the 2009 survey and 2.22 (SD = 1.00) for the 2010 survey. Education Respondents were asked to self-report their education level using one of eight cate- gories (no primary education, primary education, incomplete secondary education, completed secondary education, secondary technical education, incomplete higher education, completed higher education, and postgraduate). The largest single group in each sample (49%) reported that they had completed secondary education. When the categories were ordered (1–8), the scale mean values were 4.56 (SD = 1.40) for the 2009 survey and 4.50 (SD = 1.42) for the 2010 survey. Internet frequency Respondents who indicated that they were aware of the Internet were asked how frequently they actually used it (never, less than monthly, monthly, weekly, and daily). Those who were unaware of the Internet were coded as responding ‘‘never’’ on the frequency scale. The majority of people in the 2009 (76%) and 2010 (77%) samples had not actually used the Internet, either because they were unaware of it or because they were aware of it, but did not use it. When the categories were ordered (0–4), the scale mean values were 0.45 (SD = 1.04) for the 2009 survey and 0.46 (SD = 1.05) for the 2010 survey. Support for protests Measuring support for protests was a significant challenge given Azerbaijanis’ hesitance to criticize the government. Due to the sensitive nature of the topic as well as the political environment, this measure was presented as a vignette, a cameo description of a hypothetical situation (King, Murray, Salomon, & Tandon, 2004; King & Wand, 2006) which allows for a specific interpretation of what the question is attempting to measure. Vignettes are less threatening because they are less personal (Hughes, 1998). The following three-step process was ultimately adopted as a result of pilot testing by the Caucasus Research Resource Center. First, respondents were given a privacy card in which they were asked to agree with one of two statements: (1) ‘‘People should participate in protest actions against the government, as this shows the government that the people are in charge’’ or (2) ‘‘People should not participate in protest actions against the government, as it threatens stability in our country.’’ For the statement they selected, respondents were next asked to indicate whether they agree or very much agree. Finally, responses in the first two steps were combined to produce a single scale. Respondents who very much agree with participating in protest actions were coded 5, those who agree with protest actions were coded 4, while those who very much agree with not participating in protest actions were coded 1, and those 8 Journal of Communication (2012) © 2012 International Communication Association K. E. Pearce & S. Kendzior Social Media in Azerbaijan who agree with not participating in protest actions were coded as 2. Respondents who refused to pick between items were coded as neutral (3). The result was a single continuous variable whose mean was 2.69 (SD = 1.34) in the 2009 survey and 2.59 (SD = 1.28) in the 2009 survey. Results Preliminary analyses Data quality and assumptions were assessed prior to the construction of a SEM model. Missing values analysis (Tabachnick & Fidell, 2007) and Little’s MCAR test revealed minimal missing data and no systematic pattern of missing values. Normality, kurtosis, and skewness were acceptable with no univariate or multivariate outliers, and little multicollinearity. Independent samples t-tests were conducted to determine if there were any significant differences between the 2 years. Hypothesis tests Hypothesized relationships were tested with structural equation models for each year in which age, economic consumption ability, and education were cast as direct predictors of both the frequency of Internet use and support for protests, and in which the frequency of Internet use directly predicted support for protests. Models were tested using Mplus (Muth´en & Muth´en, 2010). Fit tests for each year’s model were as adequate. The χ2 /df ratio (Bollen & Long, 1993; Browne & Cudeck, 1993) for 2009 was 401.041/7 and for 2010 was 453.616/7. RMSEA (Bentler, 1990) was 0.000 in both years. IFI (Bollen, 1989) and CFI (Bentler, 1990) were 1.000 for both years. Alternative models were also tested following Kline’s (2005) recommendation to ensure that the model reported in Figure 1 fit the data best. This model completely reversed all the causal order between attitude for protests and Internet frequency and the other variables. The fit of the resulting model dropped significantly, even after removing some nonsignificant paths. The SEM models with coefficients for both years are displayed in Figure 1 and the results of hypothesis tests are summarized in Table 1. The 2009 model explained 20% of the variance in use and <1% of the variance in protest attitudes, while the 2010 model explained 20% of the variance in use and 2% of the variance in protest attitudes. Frequency of Internet use Age (H1), Lack of material deprivation (H2), and Education (H3) predicted Internet use frequency in each year. Support for protests Age (H4) was weakly related to attitude toward protest in 2010. Lack of material deprivation (H5) predicted attitudes toward protest in both years. Education (H6) was only related to protest attitudes in 2009. Frequency of Internet use (H7) predicted protest attitudes only in 2009. Journal of Communication (2012) © 2012 International Communication Association 9 Social Media in Azerbaijan K. E. Pearce & S. Kendzior Figure 1 Fitted models of hypothesized relationships. Discussion Before summarizing and extending the results, it is useful to acknowledge several limitations. First, although we found little evidence of dishonesty and nervousness in a review of interview notes and analysis of paradata, it is possible that social desirability effects, which are especially common in both authoritarian and Muslim societies, may have colored respondents’ willingness to discuss their Internet use and attitudes toward protests. Second, interpretations regarding statistical significance of specific tests should be tempered by recognition of the large sample sizes involved. Finally, we examined only a few aspects of Internet use and a larger study focused more specifically on the links between Internet use and civic participation is necessary. Age, the lack of material deprivation, and education were significantly associated with Internet use in both 2009 and 2010. Younger people, those with greater income and other economic resources, and those with more education all used the Internet more often. Predictions about the level of support for protests, however, received mixed support and the overall models account for very little variance in attitudes toward protest (Table 1). Age and education were only minimally related to attitudes toward protest. Lack of material deprivation was significantly and negatively associated with support for protests in Azerbaijan in both 2009 and 2010. That is, respondents who were better-off were less likely to support protests, while those who were struggling to meet material needs were more likely to support protests. Controlling for the other factors, the frequency of Internet use was positively related to attitude toward protests only in 2009. 10 Journal of Communication (2012) © 2012 International Communication Association K. E. Pearce & S. Kendzior Social Media in Azerbaijan Table 1 Results for Hypotheses, Paths, Errors, and Residuals 2009 2010 Parameter Estimate β(SE) β (SE) H1: Age → Internet Frequency −0.25∗∗∗ (.022) −0.01∗∗∗ (.001) H2: Economic Wealth → Internet Frequency 0.08∗∗∗ (.023) 0.10∗∗∗ (.021) H3: Education→ Internet Frequency 0.31∗∗∗ (.022) 0.23∗∗∗ (.014) H4: Age → Protest Attitude −0.05∗ (.028) −0.01∗∗ (.002) H5: Economic Wealth → Protest Attitude −0.13∗∗∗ (.027) −0.11∗∗∗ (.026) H6: Education→ Protest Attitude 0.08∗∗ (.029) 0.03∗ (.019) H7: Internet Frequency→ Protest Attitude 0.09∗∗ (.030) 0.02 (.026) Error in Internet frequency −0.10 (.110) −0.22∗ (.100) Error in protest attitude 2.10∗∗∗ (.134) 2.30∗∗∗ (.125) Residual in internet frequency 0.80∗∗∗ (.018) 0.80∗∗∗ (.016) Residual in protest attitude 0.97∗∗∗ (.009) 0.98∗∗∗ (.006) Note: Standardized coefficients and robust standard error reported in parentheses. ∗ p > 0.1. ∗∗ p > .01. ∗∗∗ p > .001. These factors explained little variance in attitudes toward protests. Closer exam- ination of Azerbaijan’s small group of daily Internet users, however, tells a different story. The government strategy against digital media use for political purposes worked, and the number of people agreeing that protest showed the government people were in charge plummeted between 2009 and 2010 from 53 to 27%. Similarly, only 38% of daily Internet users disagreed with protests, and this rose to 70% in 1 year. In other words, although there was little change among Internet users in general, support for protests fell dramatically among those who used the Internet the most. Azerbaijani social media users we interviewed in the summer of 2011 confirmed this shift. One said that there were 100 activists online before the ‘‘donkey blogger’’ case. He speculates that after the sentencing, 50 of the activists stopped being on Facebook, 20 became emboldened (to the point of stupidly risking themselves, he claimed), and another 30 became ambivalent. Hajizada also expressed the view that, while some became more invigorated, many Internet users simply left the movement. The relationship between Internet use and attitudes toward protest may be unique to Azerbaijan. In two neighboring countries, Georgia and Armenia, which share a history and similar social and economic conditions, but are less authoritarian than Azerbaijan, an analysis of the Caucasus Barometer dataset found no relationship between frequent Internet use and support for protests. Postscript: The demonization of social media since 2010 While the ‘‘donkey bloggers’’ sat in jail in 2010, most Azerbaijanis remained unaware of their existence. In our search of all Azerbaijani media content from June 2009 to August 2011, including print, television, and radio, we found no mention of the case in the mainstream government-controlled media, and only minimal, cautious Journal of Communication (2012) © 2012 International Communication Association 11 Social Media in Azerbaijan K. E. Pearce & S. Kendzior coverage in opposition print newspapers. Any substantial information about the case, therefore, was derived from the Internet. This was confirmed by our interviews with Azerbaijani social media users, one of whom remarked: ‘‘There is no way that a villager, a poor person in Baku, or someone in a regional city would have any idea that Emin and Adnan exist.’’ The arrest of the bloggers was thus targeted at an elite group: frequent Internet users who would hear of the arrests online and become afraid to use the Internet for activism. After the 2011 uprisings in Egypt and Tunisia, the Aliyev administration became wary of the potential of social media to leapfrog into public support, particularly since social media use in Azerbaijan had grown rapidly since 2010. Following the Egypt and Tunisia events, the government arrested a number of high-profile activists who used social media as their primary mode of communication. Unlike in the ‘‘donkey blogger’’ affair, the Aliyev administration trumpeted these new arrests to the public. This change represented a departure from networked authoritarianism, which made frequent Azerbaijani users feel like they could freely use the Internet only to be confronted with their fellow users’ shocking arrests. After the Arab Spring, the government began treating as threats not only online critiques of the government, but also a host of seemingly neutral social media activities. Azerbaijani mainstream press coverage of the Arab Spring emphasized the number of deaths and noted the dissimilarities between Azerbaijan and Arab countries, but did not mention the goals of activists or their use of social media. However, on the very day the Arab Spring began, state officials arrested former parliamentary candidate and online activist Baxtiyar Hajiyev. Two other online youth activists, Jabbar Salavan and Deyanat Babayev, were arrested in February and March 2011, respectively. Salavan was released on 26 December 2011 and Hajiyev was denied early release on 29 December 2011. While Babayev confirmed to the authors in December 2011, there has been no press coverage of a release. Unlike the ‘‘donkey blogger’’ case, the 2011 cases were covered by the mainstream media, who ran stories linking social media use with mental illness and treason. Television shows, for example, described ‘‘family tragedies’’ and ‘‘criminal incidents’’ after young people join Facebook and Twitter. In March 2011, the Azerbaijani government’s Chief Psychiatrist said that social media users avoid real-life communication, have psychological problems and cannot maintain relationships. In May 2011, the Parliament discussed the bad influence that social media has on Azerbaijan and began proposing laws to curtail it. We argue that both the ‘‘donkey bloggers’’ and the online activism conducted by Hajiyev, Salavan, and Babayev threatened the government not only by its content, but also by virtue of its very existence. It represented a ‘‘conceivable ideological alternative’’ (March, 2003, p. 210) which threatened to mobilize mass opinion on a geographically ambiguous medium: a tactic doubly threatening to authoritarian legitimation. The administration cast the bloggers and social media activists as villains in a new national narrative that sent a harsh message about citizenship and Internet use. Initially they targeted frequent Internet users who had the greatest resources with which to question the legitimacy of the regime and, as young and educated citizens, 12 Journal of Communication (2012) © 2012 International Communication Association K. E. Pearce & S. Kendzior Social Media in Azerbaijan perhaps the most to lose by doing so. After the Arab Spring, the government began to head-off the next generation of potential activists by demonizing social media to the general public and allowing the newer online cases to be publicized in the mainstream press. The government’s campaign against social media has so far been unsuccessful. According to Facebakers, a commerical Facebook analytic tool, the number of Azerbaijanis using the site grew from 105,000 in January 2010 (1% of the population) to 279,000 in December 2010 (3% of the population). At the end of July 2011 there were 431,600 users (5% of the population), partially due to the introduction of the free mobile Facebook introduced in July 2011 after government efforts to prevent it. By December 2011, 604,160 Azerbaijanis (7% of the population) were on Facebook. Similarly, according to the 2011 Caucasus Barometer, 19% of Internet users (7% of the population) reported being on a social networking site. Meanwhile, according to the Caucasus Barometer, between 2010 and 2011, frequent Internet users only increased a few percentage points. This implies that existing Internet users are those joining Facebook. Conclusion We argue that there is no relationship between the frequency of Internet use and support for political protest after the ‘‘donkey blogger’’ affair in Azerbaijan. Unlike many of the countries in North Africa and the Middle East that experienced an Arab Spring, where the documentation of state crimes on social media mobilized the population, the arrest of Azerbaijani bloggers only demoralized frequent Internet users. We believe this can be explained by the government’s embrace of networked authoritarianism as a political strategy. Young Azerbaijanis, having grown up in a chaotic post-Soviet environment, value stability and are averse to political risks. The government capitalizes on this by making any political action online—even ones that are merely an expression of criticism—seem risky. Despite the low level of opposition in the country, the Azerbaijani government is threatened by Internet use. Their concern is rooted in a brand of authoritarianism particular to the CIS countries, whose leaders lack clear ideological motives and focus on maintaining power. Like its citizens, the government values stability above all else, and will go to great lengths to prevent the population considering alternatives to its rule. The government’s demonization of social media, in which they used mainstream media platforms to dissuade citizens from using Facebook and other social media platforms, was aimed at prohibiting an elite group of frequent Internet users from reaching the broader Azerbaijani public. This tactic seems to have failed, as more Azerbaijanis have joined Facebook since the campaign began. However, increased, albeit slow, Internet use means an increased likelihood that citizens will find the stories of how activists are punished for online activity. This approach has frightened Azerbaijani Internet users, most of whom valued the Internet as a means to discuss politics away from the state and did not aggressively pursue activism on-the-ground. Journal of Communication (2012) © 2012 International Communication Association 13 Social Media in Azerbaijan K. E. Pearce & S. Kendzior Although their negative opinion of the government may not have changed, they are now less likely to make their discontent known. Acknowledgments The authors acknowledge the helpful comments from the reviewers and editors. We also thank the Azerbaijani activists with whom we discussed this project, the Caucasus Research Resource Center, and Ron Rice. References Abbasov, S. (2010). Civil society in Azerbaijan: Under fire but still resisting. Caucasus Analytical Digest, 12. Retrieved from http://www.css.ethz.ch/publications/pdfs/CAD-12-13-14.pdf. Baogang, H., & Warren, M. E. (2011). Authoritarian deliberation: The deliberative turn in Chinese political development. Perspectives on Politics, 9, 269–289. doi: 10.1017/S1537592711000892. Chinn, M. D., & Fairlie, R. W. (2006). The determinants of the global digital divide: A cross-country analysis of computer and Internet penetration. Oxford Economic Papers, 59, 16–44. doi: 10.1093/oep/gpl024. Deibert, R., & Rohozinski, R. (2010). Liberation vs. control: The future of cyberspace. Journal of Democracy, 21, 43–57. doi: 10.1353/jod.2010.0010. de Waal, T. (2011). An anatomy of apathy. National Interest. Retrieved from http://www.carnegieendowment.org/2011/04/29/anatomy-of-apathy/3l2. Gahramanova, A. (2009). Internal and external factors in the democratization of Azerbaijan. Democratization, 16, 777–803. doi: 10.1080/13510340903083919. Grono, M. (2011). Nations in transit: Azerbaijan. Freedom House. Washington, DC. Guliyev, F. (2009). The Trans-Caspian energy route: Cronyism, competition and cooperation in Kazakh oil export. Energy Policy, 37, 3171–3182. doi: 10.1016/j.enpol.2009.04.009. Guliyev, F. (2012). Political elites in Azerbaijan. In A. Heinrich & H. Pleines (Eds.), Challenges of the Caspian resource boom: Domestic elites and policy-making. Houndmills, NH: Palgrave Macmillan. Guliyeva, G. (2005). Democratization and the rule of law in Azerbaijan: Europe’s relevance. In C. P. M. Waters (Ed.), The state of law in the South Caucasus (pp. 41–58). Houndsmill, NH: Palgrave Macmillan. Gutbrod, H. (2011). An assessment of social capital in Georgia. Caucasus Research Resource Center Tbilisi. Retrieved from http://scr.bi/oi6qsu. Hale, H. E. (2005). Regime cycles: Democracy, autocracy, and revolution in Post-Soviet Eurasia. World Politics, 58, 133–165. doi: 10.1353/wp.2006.0019. Hall, R. L., Rodeghier, M., & Useem, B. (1986). Effects of education on attitude to protest. American Sociological Review, 51, 564–573. doi: 10.2307/2095588. Howard, P. N. (2010). The digital origins of dictatorship and democracy: Information technology and political Islam. New York, NY: Oxford University Press. Howard, P. N., Agarwal, S. D., & Hussain, M. M. (2011). When do states disconnect their digital networks? Regime responses to the political uses of social media. The Communication Review, 14, 216–232. doi:10.1080/10714421.2011.597254. 14 Journal of Communication (2012) © 2012 International Communication Association K. E. Pearce & S. Kendzior Social Media in Azerbaijan Hughes, R. (1998). Considering the vignette technique and its application to a study of drug injecting and HIV risk and safer behaviour. Sociology of Health and Illness, 20, 381–400. doi: 10.1111/1467-9566.00107. Kalathil, S., & Boas, T. C. (2003). Open networks, closed regimes: The impact of the Internet on authoritarian rule. Washington, DC: Carnegie Endowment. Katz, J. E., & Rice, R. E. (2002). Social consequences of Internet use: Access, involvement, and interaction. Cambridge, MA: MIT Press. Kedzie, C. (1997). Communication and democracy: Coincident revolutions and the emergent dictators. Los Angeles, CA: Rand. Retrieved from http://www.rand.org/pubs/rgs_dissertations/RGSD127.html. Kendzior, S. (2011). Digital distrust: Uzbek cynicism and solidarity in the Internet age. American Ethnologist, 38, 559–575. doi: 10.1111/j.1548-1425.2011.01323.x. King, G., Murray, C. J. L., Salomon, J. A., & Tandon, A. (2004). Enhancing the validity and cross-cultural comparability of measurement in survey research. American Political Science Review, 98, 191–207.doi: 10.1017/S000305540400108X. King, G., & Wand, J. (2006). Comparing incomparable survey responses: Evaluating and selecting anchoring vignettes. Political Analysis, 15, 46–66. doi: 10.1093/pan/mpl011. Kline, R. B. (2005). Principles and practices of structural equation modeling. New York, NY: Guilford Press. MacKinnon, R. (2011). Networked authoritarianism in China and beyond: Implications for global Internet freedom. Journal of Democracy, 22, 32–46. doi: 10.1353/jod.2011.0033. March, A. F. (2003). State ideology and the legitimation of authoritarianism: The case of post-Soviet Uzbekistan. Journal of Political Ideologies, 8, 209–232. doi: 10.1080/13569310306088. Morozov, E. (2011). The Net delusion: The dark side of Internet freedom. New York, NY: Public Affairs. Muth´en, B., & Muth´en, L. (2010). Mplus. [Computer software]. Los Angeles, CA. Oates, S. (2008). Comrades online? How the Russian case challenges the democratizing potential of the Internet. Retrieved from http://newpolcom.rhul.ac.uk/politics-web-20- paper-download/oates%20comrades%20online%20revised%20april%202008.pdf OpenNet Initiative. (2010). Azerbaijan. Retrieved from http://opennet.net/research/profiles/azerbaijan. Pearce, K. E. (2011). Poverty in the South Caucasus. Caucasus Analytical Digest, 34. Retrieved from http://www.isn.ethz.ch/isn/Digital-Library/Publications/Detail/?ots591=0c54e3b3- 1e9c-be1e-2c24-a6a8c7060233&lng=en&id=135319. Pearce, K. E., & Rice, R. E. (forthcoming). Digital divides of Internet awareness, adoption, and use due to socioeconomic status and foreign language skill: The cases of Armenia, Azerbaijan, and Georgia. Rasizade, A. (2003). Azerbaijan in transition to the ‘‘new age of democracy.’’ Communist and Post-Communist Studies, 36, 345–372. doi: 10.1016/S0967-067X(03) 00043-6. Rice, R. E., & Katz, J. E. (2003). Comparing Internet and mobile phone usage: Digital divides of usage, adoption and dropouts. Telecommunications Policy, 27, 597–623. doi: 10.1016/S0308-5961(03) 00068-5. Rodeghier, M., Hall, R. L., & Useem, B. (1991). How education affects attitude to protest: A further test. The Sociological Quarterly, 32, 277–288. doi: 10.1111/j.1533-8525.1991.tb00358.x. Journal of Communication (2012) © 2012 International Communication Association 15 Social Media in Azerbaijan K. E. Pearce & S. Kendzior Saunders, R. A. (2009). Wiring the second world: The geopolitics of information and communications technology in post-totalitarian Eurasia. Russian cyberspace, 1, 1–24. Retrieved from http://www.digitalicons.org/issue01/pdf/issue1/Wiring-the-Second- World_R-A-Saunders.pdf. Shirky, C. (2007). Here comes everybody: The power of organizing without organizations. New York, NY: Penguin. Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics (5th ed.). Boston, MA: Allyn & Bacon. Zuckerman, E. (2008). The cute cat theory talk at ETech. Retrieved from http://www.ethanzuckerman.com/blog/2008/03/08/the-cute-cat-theory-talk-at-etech/. 16 Journal of Communication (2012) © 2012 International Communication Association