The Informational Segregation in the Social Medias: the Generational Factors

The algorithmic web personalization is rather developed and ubiquitous nowadays. However this process has negative consequences. For example, the emergence of filter bubbles [Pariser 2011]. The filter bubbles are the result of an algorithmic power, when users become separated from information that disagrees with their viewpoints, effectively isolating them in their own cultural or ideological web spaces. The filter bubbles could lead to echo chambers – somewhat closed social systems [Jamieson, Cappella 2008]. In the echo chambers, alike users repeat and reinforce similar beliefs and discrepant ideas are not discussed. As described in the literature, it is a state of intellectual isolation. This isolation leads to the informational and social polarization and segregation. In this research we focus on the informational segregation in the social media. Our target group are the millennials as the first digital natives. Some researches say that millennials are rather sustainable to the informational segregation and try to diversify their news flows [Bozdag et al. 2015; Zuiderveen Borgesius 2016]. Besides, others maintain that millennials are media-illiterate and do not aware of the quality and reputation of their informational sources sufficiently [Fletcher, Nielsen 2018]. In this paper we try to examine which sociodemographic attributes of millennials and other generations determine the sustainability to the informational segregation. The data are the information about subscriptions on the most popular political publics (5 ‘pro-government’ (Channel One Russia, RIA, Russia Today, Vesti, TASS) and 5 ‘opposition’ publics (Meduza, Lentach, Rain, Mediazona, The Ekho Moskvy) and sociodemographic attributes of users at the most popular Russian social media VKontakte. The data were collected by Python parsing on cloud computing. The total number of users after the data cleaning is 2 038 931 cases. We construct the index of the sustainability to the informational segregation as the lower boundary of the Wilson confidence interval for the ratio of proportions of ‘pro-government’ and ‘opposition’ subscriptions for each user. The choice of the parameter of the Wilson confidence interval seems a more scale resistant metrics than others. After that we run some statistical tests to compare means and medians and a pool of robust linear regression models with this index as the dependent variable and make binary classification via gradient boosting. We can conclude that the youngest generations – millennials and Z – are really more sustainable to the informational segregation than their predecessors. The age within generations is not so crucial. However the results are not unequivocal: for instance, the importance of generational feature is less crucial for division of the most sustainable and the most unsustainable users in gradient boosting. Females have fewer scores of sustainability than males that could be explained their less political concernment for example [Sindermann 2020]. Surprisingly such potentials factors of sustainability as education in the prestigious Russian universities and living in the big cities are negative. The existence of accounts in Instagram and Skype is associated with higher sustainability and the existence of accounts in Twitter, Facebook and LiveJournal - with less sustainability.