Galina Selivanova, Laboratory for Internet Studies, HSE, Saint Petersburg
For more than two decades researchers from various fields have been exploring how the new information and communication technologies influence politics, government, and political participation. They seek to understand how the digital media affect the dynamics of mobilization for protest events, recruitment of new supporters, organization and coordination of their activities. A visible body of literature has emerged recently that relates protest participation to internet consumption both in Russia and in other parts of the world, however, much less is known about the role of social movements’deliberate online activity and participation in those movements outside periods of mass mobilization. This paper aims at filling in this gap. Focusing on the case of ‘Observers for Fair Elections’movement in St.Petersburg, we obtain unique comparative data from 17 city districts and seek to find out what has driven high participation in some of them and low participation in others. While most conditions are the same in all districts, their group pages on a social networking site are dramatically different, and we seek to determine if this is related to the offline participation rates. We seek to answer the questions: How are features of online communities maintained by social movements related to offline participation rates? How does an individual’s activity or position in a movement’s online community predict his/her offline participation? Are online and offline leaders the same persons or some kind of division of labour may be observed?
Online data were retrieved from group accounts of each of the 17 district branches and the city-level group of Observers movement in VK.com with our (Lab’s) software VKMiner. The downloads included all publically available demographic data, group membership, friendship ties, comments, likes, and posts in groups produced by the group members and non-member VK.com users. We investigate activity of 12,778 participants who have ever been active in district groups who generated 2,326 posts, 11,249 comments and 33,375 likes during the entire period of the groups’existence (December 2011 ‘September 2014). This data was supplemented with information on offline movement members who actually attended the poll stations in the role of observers, counting committee members, and some other independent roles. This data was collected by the Observers’call centre during the elections day on September 14th, 2014 (in-coming and outgoing calls). At the next stage they were manually matched to their VKontakte accounts. 257 of around 300 offline participants have turned to be members of one or more online groups
At the district level, the number of people who participated in the elections in different offline ‘activist’capacities correlate with each other. Therefore, we use the overall number of all types of offline activists aggregated by district as the target variable. The entire dataset contains 69 variables, 51 of which have been used in the analysis divided into four major groups: (1) absolute numbers (e.g. number of members or likes in a group, 17 variables), (2) the same numbers weighted by the off-line district population (17 variables), (3) numbers weighted by the online group size (e.g. posts per user, 10 variables), and (4) network metrics such as density and modularity that also contain some weighting on group size (9 variables). At the first stage of the analysis we check how different network characteristics and online activities correlate to the offline activism. As determined by Pearson correlation coefficient, the offline participation rate (i.e. the number of participants per 10,000 of the district population) has a strong association with many features of online groups including network characteristics and online activity. Next, we run the cluster analysis to find out the common pattern of features inherent to districts that have more active citizens. Three downtown districts ‘Admiralteysky, Petrogrardsky and VO ‘clearly differed from the others by the number of offline participants weighted by the district population, and by a number of other variables.
At the following part of our research we investigated the individual level of online activity and offline participation. As the share of offline participants among online participants is small (2%) we ran a number of binary logistic regressions with penalized likelihood, including those with backward elimination. Models indicated that individuals producing more content in movements’online communities are not those who take part in offline electoral observation. However, a high number of friendship connections in the online communities points at the offline activists. Thus embeddedness in the community is more important than online communication. It should be noted that the variation between online-only participants is dramatically high and therefore on the individual level online data do not predict offline participation well. We then sought to determine which online properties best explain the informal leadership in the Observers movement. It turned out that leaders are much more active in their online communication than rank-in-file activists or online group members and are located in the centre of friendship networks in online communities. In addition they receive up to three times more attention and feedback to their entries than regular participants. Moreover, our findings confirm conclusions from the previous section on the individual activity: offline participants tend to produce less online content but on average have more friendship ties.
From the analysis above, we conclude that online footprints of social movements may be very informative, although these data should not be used straightforwardly. On the individual level it is evident that offline participants do not coincide much with online content producers, nonetheless, the friendships network analysis showed that activists and leaders take more central positions in online networks. However, the relation between online features and individual offline participation is weak, while it is much more visible at the group level. This suggests presence of an ecological effect: larger, more connected and active groups produce higher levels of offline participation, but not necessarily among active online. Simultaneously, online presence may reflect the existence of an active offline core produced by offline leaders rather than by online activity. With this case study we document the existence of a measurable relationship between online and offline realms of a social movement.