Recent studies have reported a significant relationship between the use of social media and political engagement. However, there appear to be few comparative studies that explore the association between social networking site use and participation in different types of protest activities for the case of Latin America. The present study employs supervised and unsupervised data analysis techniques to explore this association for 9 different social networking sites and 5 types of protest using disaggregated data on 17 Latin American countries. Multiple correspondence analysis is applied to create proxy measures of the two phenomena, followed by a cluster analysis using these measures to classify individuals into different clusters in each country studied. These clusters indicated that there exists an interplay between the use of these sites and participation in protests. Decision rules were then induced to generate interpretable information on the clusters identified for each country. The results suggest that there is a high degree of heterogeneity in social networking site use and protest participation.
Latin America has been the setting for a number of civil protest movements over the last decade in which Internet social media have been used as a means of political expression (Guzman-Concha, 2012; Perez-Li ´ n˜an and Polga- ´ Hecimovich, 2016; Anselmi, 2017; Valenzuela, 2013; Harlow, 2011). This article explores the relationship between the use of different social networking sites and participation in protest activities in a series of Latin American countries. Recent comparative studies have shown that there is a positive relationship between social media use and protest participation (Kim and Chen, 2016; Boulianne, 2017; Dong et al., 2017; Gan et al., 2017). Meta-analyses, for example, have found evidence of a relationship between social media use and such participation in both online and offline environments (Skoric et al., 2016a; Boulianne, 2015, 2017; Skoric et al., 2016b). In one of these investigations, based on 133 cross—sectional studies, the author concluded that the effects of using social media on such participation are stronger for political expression and weaker for informational purposes (Boulianne, 2017). What was not established in these studies, however, was which social media was (were) associated with which type(s) of protest. There thus remains a need to identify the more specific relationships existing between the latter and the many different social networking sites that currently exist. The exploration of individuals’ use of such sites, or of the many types of protest they may participate in, is a complex, multidimensional task, however. To be sure, considerable work has already been done on the use by individuals of multiple social media channels. For example, in Hecking et al. (2018, 2017) the authors develop a tool and a strategy for analyzing the dissemination of content through Web pages, Twitter and Wikipedia. Another case is Farahbakhsh et al. (2016), which explores the cross-posting activity of users across three major social networking sites: Facebook, Twitter and Google+. But the research into this phenomenon is complicated by the sheer number of different social media, various problems to do with defining data models, anonymisation and privacy issues, and certain ethical questions (Henderson et al., 2013; Kaslow et al., 2011). Investigations have also shown that individuals engage in a variety of different types of protest (Mourao et al. ˜ , ۲۰۱۶; Fourcade et al., 2016; Harlow and Harp, 2012). Traditional protest modes and those that actually bring about changes in social situations still occur in offline environments but it is also true that much protest is now taking place online (Harlow and Harp, 2012). The latter type of activity is frequently hosted by social networking sites. But if, as we have just observed, tracking the use of multiple sites and the participation in different protest modes are complex tasks when tackled separately, the complexities of exploring the relationships between them are even greater. The present work therefore proposes to take the first step towards filling the gaps in our knowledge on these relationships by investigating the interplay between users of specific social media sites and their participation in specific protest activities in Latin America. Furthermore, our analysis will be conducted at a disaggregated level, that is, country by country. Thus, we will explore the association between participation in 5 types of protest activity and the use of 9 social media sites in 17 different Latin American countries. For each of these countries our objective will be threefold: (i) to explore the patterns of social networking site use and protest participation, (ii) to cluster individuals according to these two concepts, and (iii) to interpret the resulting clusters. To accomplish this we will bring to bear three techniques in turn: (i) multiple correspondence analysis, (ii) agglomerative hierarchical cluster analysis, and (iii) decision rules induction. The first two are data exploration techniques while the third is an automatic supervised learning technique we will apply to derive interpretable information from the clusters obtained using the second technique. This approach will allow us to carry out our analyses in a low-dimensional space.