Presentation is loading. Please wait.

Presentation is loading. Please wait.

“LiveJournal Libra!” The influence of the political blogosphere on political mobilisation in Russia in 2011-12 Shcherbak Andrey (LCSR, HSE SPb) Koltsova.

Similar presentations


Presentation on theme: "“LiveJournal Libra!” The influence of the political blogosphere on political mobilisation in Russia in 2011-12 Shcherbak Andrey (LCSR, HSE SPb) Koltsova."— Presentation transcript:

1 “LiveJournal Libra!” The influence of the political blogosphere on political mobilisation in Russia in 2011-12 Shcherbak Andrey (LCSR, HSE SPb) Koltsova Olessya (LINIS, HSE SPb) Supported by NES

2 Introduction The 2011-12 election: declining support for the ‘party of power’; mass protests The role of Internet? Blogs – in particular Focus on posts of the LiveJournal top-2000 bloggers, 3 periods Contribution: 1) blogs are the stronghold of opposition, 2) relationship between online politics and offline politics

3 Internet and protests Internet and politics. General discussion: Internet creates new opportunities for political communication and mobilization The Arab Spring; protests in Iran and Moldova YouTube, Twitter, social networks and blogs A descriptive approach prevails: focus on structure and functioning of Internet during protests Causality? Internet and voter turnout, share of votes. Internet accessibility, Internet consumption and protests Old Media vs. New Media; how to separate effects? The case of hybrid regimes

4 The 2011-2012 election in Russia: the role of Internet Russian Internet audience: 50 mln users. #1 in Europe. The Yandex audience exceeded the audience of the 1 st channel (2012) Relative autonomy of Internet in Russia Income growth and value change (Inglehart) Natalia Zubarevich: «4 Russias». «The First Russia» - residents of metropolitan areas, the most active Internet users Protests and political culture change (Makariin and Polishchuk). Access to new technologies and new opportunities for mobilization (Bodrunova and Litvinenko) Protests = middle class + access to Internet (?) Underestimated role of blogs We predict prevalence of protest attitudes in blogs

5 Data and Methods The choice of blog platform - LiveJournal Selection of top-2000 bloggers by rating 15th August - 15th September 2011 (the “quiet” period). 27th November – 27th December 2011 (the parliamentary election campaign and the subsequent protest actions). 1st February - 6th March 2012 (the presidential election campaign and protest actions). the ‘Big Data’ approach

6 Data and Methods an automatic topic extraction method this method assigns each text to each topic with a varying probability, thus performing a procedure akin to fuzzy clustering. We have selected a type of topic modelling algorithm known as Latent Dirichlet Allocation (LDA) with Gibbs sampling encoded in the Stanford Topic Modelling Toolbox software This algorithm assumes the existence of latent variables interpreted as topics in any corpus of texts. All words of the corpus are assumed to be distributed over these topics in a certain way The algorithm attempts to assess which words could most probably constitute each topic, thus forming lists of words with descending probabilities, which can easily be interpreted by a social scientist The advantage of this method of data selection over a keyword search lies, first and foremost, in the fact that the algorithm makes it possible to assign relevant texts to the topics in question, even if they do not feature the keywords that a researcher could think of. the algorithm makes it possible to reveal unexpected topics

7 Data and Methods 1 step: Each of the three samples was divided into 100 topics. Two coders manually selected topics connected with the current Russian politics and compared their results. In total, 55 political topics were identified. ‘Political’ means domestic politics, excluding historical topics (Soviet history), or international politics (Egypt or Ukraine), only I these events were not connected by author to Russian politics. 55 political topics: 11 August – September 2011, 24 – November - December 2011 and 20 – February-March 2012 Step 2. all texts assigned to the selected topics with a probability of more than 0.1 were united into one array. As a result, the reduced sample became, to a larger degree although not absolutely, related to domestic Russian politics. 3 step: This new database was divided into 13 separate weeks (four in the quiet period, four around the parliamentary elections, and five before the presidential elections), and each weekly sample was clustered into 20 topics. These topics were manually processed by two coders, and 123 topics predominantly related to domestic Russian politics were selected; the top 30 posts were taken from each. This produced a database containing 3690 posts with an extremely high probability of being classifiable as political.

8 Examples of clusterization масло1494 добавлять1126 вода933 мясо810 ложка788 минута777 сахар740 соль727 2725 рецепт708 вкусный698 г674 блюдо664 вкус650 мука642 яйцо639 перец601 сырой599 сок589 молоко588 участок 1712 избирательный1661 наблюдатель1573 голос 1491 голосование1160 результат 1071 кандидат 995 бюллетень981 избиратель960 комиссия 830 фальсификация807 4 735 март 714 голосовать702 нарушение574 опрос 556 проголосовать555 уик 531 партия 450 протокол 445 русский 6480 советский 2165 ссср 1513 язык 1016 народ 1010 сталин 912 еврей 806 латвия 666 национальный605 германий 584 немец 537 немецкий 525 союз 505 история 495 еврейский455 государство405 война 403 нация 389 население 376 гитлер 360

9 Data and Methods

10 4 step: the posts were manually coded by five coders who had obtained an intercoder reliability Kappa coefficient of not less than 0.73 in the pilot coding. Posts were coded according to several indicators : “politics” (classification of the content of a post as political), “government” (attitude to the government and the ruling elite), and “opposition” (attitude to the opposition). About 30% of posts were recognized as non-political, and in each of the political posts attitudes to the government and to the opposition were detected as two independent values: thus, a post could be both anti-government and anti-opposition simultaneously.

11 Variables The share of posts about politics (ratio of posts about politics to the total number of posts in the period studied); this varies from 0 to 100%; The share of posts about the elections (and protests) from among all political posts; this varies from 0 to 100%; The share of oppositional posts (the proportion of posts with a positive or neutral attitude to the opposition compared to the total number of political posts in the studied period); this varies from 0 to 100%; The share of “pro-government” posts (the proportion of posts with positive or neutral attitude to the government compared to the total number of political posts in the studied period); this varies from 0 to 100%; Attitude towards the opposition (average value of the variable “opposition” in the period studied); this varies from -1 to 2; Attitude towards the government (average value of the variable “government” in the period studied); this varies from -1 to 2;

12 Hypotheses Our assumption about the existence of a link between indicators of the politicisation of the blogosphere and the pre-election ratings was tested using Spearman’s correlation coefficient. H1: The share of positive oppositional posts will be significantly higher than the share of positive pro-government posts. H2. Attitudes towards the opposition will be significantly more positive than those towards the government. H4. The higher the political activity of bloggers (share of posts about politics, the share of posts about the elections, the share of positive oppositional posts, attitude towards the opposition, the difference between attitudes towards the opposition and towards the government), the lower the ratings of pro-government party and candidate (electoral rating of UR, Vladimir Putin, the confidence rating of Dmitry Medvedev). H4. The higher the political activity of bloggers (see H3), the higher the ratings of opposition (electoral ratings of the CPRF, LDPR, Gennady Zyuganov and Vladimir Zhirinovsky).

13 Political activity of bloggers

14

15 Results Blogosphere belongs to opposition-minded bloggers Mean score «The share of oppositional posts» – 0,47 «The share of “pro-government” posts » - only 0,26. «Attitude towards the opposition » - 0,56, but «Attitude towards the government» - 0,05. The difference between these variables (the difference between attitudes towards government and opposition) varied in certain weeks from 1,74 to -0,12, with mean value 0,60 (scale from -1 to 2). Overall these findings were consistent with our assumption concerning the domination of opinions of the “dissatisfied middle class” in blogs, with its unfulfilled demands for political freedom. Correlations with electoral ratings

16 Political activity and ratings RatingsPosts about elections, %Posts about politics, % The CPRF 0.872** (p=0.000) 0.571* (p=0.041) UR -0.172 (p=0.574) 0.091 (p=0.768) JR 0.551 (p=0.051) 0.321 (p=0.285) The LDPR 0.220 (p=0.471) 0.027 (p=0.931) Vladimir Putin -0.254 (p=0.402) -0.054 (0.860) Gennady Zyuganov 0.680* (p=0.011) 0.462 (p=0.112) Dmitry Medvedev -0.278 (p=0.358) -0.108 (p=0.724) Vladimir Zhirinovsky 0.106 (p=0.730) -0.155 (p=0.613)

17 Political activity and ratings The CPRF – the main benefiter from bloggers’ political activity? The effect of the ‘Navalny’s strategy’? To vote for any party but the United Russia. Only parliamentary parties could benefit from it: the CPRF, JR and the LDPR

18 Political activity and ratings RatingsOppositional posts, % of political posts Attitude twd opposition, weakly means Pro-government posts, % of political posts Attitude twd government, weakly means The CPRF0.310 (p=0.302) 0.026 (p=0.933) 0.089 (p=0.774) -0.140 (p=0.649) UR-0.042 (p=0.892) -0.333 (0.266) 0.733** (p=0.004) 0.636* (p=0.019) JR-0.696** (p=0.008) -0.123 (p=0.688) 0.106 (p=0.730) 0.095 (p=0.758) The LDPR-0.085 (p=0.783) 0.251 (p=0.408) -0.462 (p=0.112) -0.389 (p=0.188) Vladimir Putin0.401 (p=0.174) 0.043 (p=0.889) 0.196 (p=0.521) 0.152 (p=0.620) Gennady Zyuganov -0.378 (p=0.202) 0.132 (p=0.667) -0.053 (p=0.864) -0.171 (p=0.576) Dmitry Medvedev 0.449 (p=0.124) 0.004 (p=0.989) 0.158 (p=0.607) 0.094 (p=0.760) Vladimir Zhirinovsky -0.044 (p=0.887) 0.215 (p=0.480) -0.582* (p=0.037) -0.397 (p=0.179)

19 Results that appeals to the public online are capable of bringing benefits to politicians offline The “recipe for success” for the opposition seems to have been simple: the main thing was not what to write, but how much. The more posts about the elections, the higher the pre-election ratings of oppositional parties. The ruling party was associated with another strategy - an attempted increase in the share of pro-government posts.

20 Conclusions The relationship between online politics and offline politics The role of Internet in hybrid regimes. Do not ignore blogs! As a result we were able to answer questions both about the political activity inside the blogosphere itself and beyond its limits - in the world of “real”, offline politics.

21 THANK YOU!


Download ppt "“LiveJournal Libra!” The influence of the political blogosphere on political mobilisation in Russia in 2011-12 Shcherbak Andrey (LCSR, HSE SPb) Koltsova."

Similar presentations


Ads by Google