ABSTRACT
In political regimes where traditional mass media are under state control, social networking sites may be the only place where citizens are exposed to and exchange dissident information. Despite all the attempts, complete control of social media seems to be implausible. We argue that the critical information that people see, read and share online undermines their trust in political institutions. This diminishing trust may threaten the legitimacy of the ruling regime and stimulate protest behaviour. We rely on original survey data of Kazakhstani college students to confirm these expectations. The data are unique in that they directly measure exposure to critical/dissident information, as opposed to simply assuming it. The analysis leverages Coarsened Exact Matching to simulate experimental conditions. This allows us to better identify the consequential mechanism and the attitudinal precursor by which social media influence protest in an authoritarian context.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1 Freedom House ranking: Not Free(https://freedomhouse.org/country/kazakhstan).
2 CEM is a statistical approach to address the possibility of endogeneity (see Esarey Citation2015 for a description of the logic and limitations). A more detailed description of CEM is in the supplemental online material; see also Blackwell et al. (Citation2010) and Iacus, King, and Porro (Citation2012).
3 We also created two separate indices here for a validity check on our model estimates, one based on passive critical exchange, and one representing active critical exchange.
4 We checked the robustness of these results by estimating a model of protest behaviour using a subset of the Kazakhstan 2011 World Values Survey. The operationalization and models are included in the supplemental online material.
5 It is entirely plausible that the relationship between our measure of critical SNS use and trust is driven by the index components focused on active critical exchange, as opposed to passive critical exchange (or critical exchange overall for that matter). We estimated these models with separate indices for each to test for this possibility, and the substantive results did not change.