AoIR2018-#rezist Selected Papers of #AoIR2018: The 19th Annual Conference of the Association of Internet Researchers Montréal, Canada / 10-13 October 2018 Suggested Citation (APA): Mercea, D. (2018, October 10-13). The Political Information To Protest: An Assessment Of Topical Social Media Usage In Contentious Politics. Paper presented at AoIR 2018: The 19th Annual Conference of the Association of Internet Researchers. Montréal, Canada: AoIR. Retrieved from http://spir.aoir.org. THE POLITICAL INFORMATION TO PROTEST: AN ASSESSMENT OF TOPICAL SOCIAL MEDIA USAGE IN CONTENTIOUS POLITICS Dan Mercea University of London Introduction In Romania, the 2017 #rezist protests marked a high point of successful mobilisations that have followed a pattern of rolling demonstrations across the country and among Romanian communities abroad which attained their declared aims, i.e. to halt the licensing process for the largest open-cast gold mine in Europe, in 2013; to prompt the resignation of two governments, in 2012 and 2015; and in the case of the #rezist protests, to forestall the adoption of emergency ordinance no. 13 that would slacken the country’s anti-corruption legislation (Badescu and Jiglau, 2016). Attendant to this latest wave was a modest rise in levels of protest participation (Mercea, 2014). Against evidence that the utilisation of social media in the protest wave that has swept democratic countries since the turn of the decade (Biekart and Fowler, 2010) has primarily been instrumental to the orchestration of collective action, this article probes the circulation of political information among protestors on Facebook. As in other recent instances of protest events in the country (Mercea, 2017), protestors set up public Facebook event pages used for topical communication at various stages such as in advance, during or in the wake of a demonstration. A combination of network analysis and topic modelling are employed to describe the communication encountered on Facebook event pages associated with four Romanian cities where large anti-corruption protest locations transpired in January-March 2017. Survey data for the same four locations is analysed to measure the degree to which students who attended those protests exchanged political information on Facebook. Romanian students were previously reported as more likely to engage in acts of protests than the general population. Likewise, they were most likely to embrace social network services for both offline and online political participation (Burean and Badescu, 2014). The article grapples with the contention that the while the use of social media exposes people to diverse political content (Bakshy et al., 2015, Fletcher and Nielsen, 2017, Kim, 2011), it does not make them politically more knowledgeable. Critical observers have argued that social media users are becoming dangerously unaware of their level of political ignorance (Gil de Zúñiga et al., 2017). In this light, the aim to examine the degree to which protestors source political information on social media is further significant as they are ostensibly a cohort that is susceptible to taking political action on the basis of strategically manufactured online rhetoric designed to engineer a consensus for collective action (Bliuc et al., 2012). Data and Methods The Facebook dataset comprises 48 event pages associated with the four cities examined in this study—Bucharest, Cluj, Oradea and Timisoara—with data points covering the period 16 January to 6 March 2017. Secondly, a student survey probing participation in the #rezist protests was conducted in the four cities. These are university centres where 60% of the student population in Romania is concentrated according to the Romanian Ministry of Education. It provided the research team with the list of registered students in the country, in 2017. The survey was conducted from April to June 2017. It comprised 1658 respondents. The sampling strategy accounted for the proportion of students per higher education discipline. We first approached the social media—Facebook event page—data descriptively. To that end, we used the “five number summary” statistics (Luke, 2015). In addition, to gain an appreciation of how information circulated in the city event page network, we analysed the diffusion of information (González-Bailón and Wang, 2016) based on the distance between nodes or vertices in the network. The results are standardized falling between 0 and 1 with a high value indicative of a short distance between nodes and a low value suggesting the reverse. Moreover, we estimated the capacity for brokerage of actors in the network. As we expected to find structural gaps in the flow of information (González-Bailón and Wang, 2016), we calculated the modularity score of the network. Additionally, we conducted a topic modelling analysis using the Latent Dirichlet Allocation algorithm. This was done so as to examine the content of the communication transpiring in the event page network. Lastly, we used regression analysis to model the bearing of political information on protest participation while controlling for multiple factors previously shown to influence this outcome. Results and Discussion The regression analysis revealed that the likelihood of protest participation was influenced first by respondents’ political interest, measured as their penchant for sourcing news content from the internet; and by communication about the protests on social media, respectively. The effect of posting activity on social media, by way of calls for protest participation or commentary on the protests (Models 1 and 2) disappeared when controlling for students’ participatory experience. However, placing a #rezist hashtag overlay on one’s profile photo and joining a Facebook protest event page or group were two of the three variables with the highest predictive power in all models. Conversely, there was an inverse relationship between broadcast news consumption and protest participation. Both the effect of internet (positive) and broadcast(negative) news consumption disappeared, however, as soon as we introduced the indicator for protest experience. Next, we noted that neither political information nor knowledge had any bearing on participation. This additionally held true for organisational membership, voting or any of the socio-demographic variables. Thus, the chief predictors of protest participation were pro-active social media usage combined with protest experience and distrust of a prominent social (the church) rather than any political institution. Although initial models suggested political interest affected students’ inclination to participate, the effect vanished upon further testing. Table 1. Determinants of student participation in the #rezist protests Dependent variable: protest participation *<.05, **<.01, ***<001. Coefficients are standardized beta for logistic regression. Second, our study further points to the diffusion of information among members of event page networks on Facebook as being more robust than expected. Despite the existence of community structures in the network, and by implication of structural holes, the observed diffusion score and the capacity for brokerage among network actors led us to infer that information circulated reasonably effectively among and within groups in the Facebook network. The low density and long diameter of the protest event network alluded to its social fabric being at least in part heterogenous, with users many steps removed from each other. Network clustering suggested users were not confined to tight, localised groups. Representing the main public interface of the protests, the event Model 1 Model 2 Model 3 Model 4 TV or radio news consumption -.073*** -.066*** -.067 -.089 Used internet for political information .086*** .084*** .042 .039 Posted calls for protest participation on social media .071*** .069*** .066 .093 Posted a protest-related comment on social media .051*** .051** .072 .073 Placed #rezist hashtag on profile photo .069*** .076** .151* .193** Joined a Facebook protest event page or group .115*** .114*** .120** .158** Distrust in government .027 .046 .068 Distrust in parliament .039 -.111 -.161 Distrust in church .036* .105** .136 Distrust in National Agency for Combatting Corruption (DNA) -.074*** .009 -.007 Protest participation experience .157** .168** Vote .006 .011 Organisational membership -.019 -.022 Political knowledge .041 .056 Father’s education .004 Mother’s Education -.044 Income-money spent in a month .087 Gender .072 NagelkerkeR² .271 .302 .474 .488 Dependent variable: protest participation *<.05, **<.01, ***<001. The indicators are standardized beta for logistic regression. page network was populated with messages that displayed political information side-by- side with content pertaining to the orchestration, implications and outcomes of collective action. Table 2: LDA-modelled latent topics Selected References BUREAN, T. & BADESCU, G. 2014. Voices of discontent: student protest participation in Romania. Communist and PostCommunist Studies 47, 385-397. GIL DE ZÚÑIGA, H., WEEKS, B. & ARDÈVOL-ABREU, A. 2017. Effects of the News-Finds-Me Perception in Communication. Journal of Computer-Mediated Communication, 22, 105-123. GONZÁLEZ-BAILÓN, S. & WANG, N. 2016. Networked discontent: The anatomy of protest campaigns in social media. Social Networks, 44, 95-104. MERCEA, D. 2014. Towards a Conceptualization of Casual Protest Participation. East European Politics and Societies, 28, 386-410. Topic 1: Opposition to ordinance Facebook, Oradea, come, PSD, street, Unirii, participate, tell, ordinance, EU, judgement, government, OUG, resist, give, positions, penal, is, Cluj, protest Topic 2: Sunday anti- corruption protests Facebook, event, citizens, street, do, be, Sunday, public, government, right, protests, state, events, the people’s, see, want, government, corruption, kill, defend Topic 3: Protesting audio-visual regulator Victoriei, Facebook, audiovisual, event, play, generates, resistance, draw, law, code, do, television, constitution, CNA, regulation, disinformation, Sunday, resignation, see, protests Topic 4: Orchestration of protest banner, Facebook, protest, be, see, street, people, banners, Romania, can, meet, seeing, event, corruption, only, help, delimit, transmit, parliament, Victoriei Topic 5: Collective action and outcomes Protest, Facebook, are, do, change, public, see, corruption, be, kill, event, citizens, Sunday, want, begin, down, street, come, scandal, power Topic 6: Preserving the momentum of rolling protests Sunday, Facebook, square, flag, street, Cluj, fight, celebrate, resistance, win, Unirii, resist, Bucharest, initiative, Victoriei, resist, tri-color European, union Topic 7: Protest as civic participation Right, Sunday, liberty, citizens, do, European, civic, gather, needs, be, authority, politics, expression, rights, people, public, Opera, help, education, resist