Towards a Model of Twitter Network Formation in Global Education Policy

• In Event: New Agendas of Global Governance of Education

Proposal

Networks and digital communication platforms, such as Twitter, play a central role for global education and conviction processes (Junemann et al., 2016; Kolleck et al., 2017). Both state- and non-state actors use social networks as channels for the diffusion of educational innovations and practices (e.g. inclusive education) (True & Mintrom, 2001). Therefore, focusing on actor networks is an innovative approach to the study of global education policy. Although Twitter represents a relatively new form of social networks, it gains increasing importance for the diffusion of information and innovation, also in education policy (Conover et al., 2012; Dubois & Gaffney, 2014). Users apply Twitter to inform (themselves and others) and build connections to others using so-called retweets and mentions. Despite its growing use and the general relevance of this online communication for global education and conviction processes, there is only little knowledge about the formation of issue-specific communication networks to date.

Thus, the aim of this contribution is to find a model of the formation of an issue-specific, educational political Twitter network. We derive hypotheses based on empirical findings concerning the role of different organization types (e.g. non-governmental organizations (NGOs) or international organizations (IOs)) and on social network theory and test those using inferential network analysis. Our research subject is the Twitter network on the topic of inclusive education in the context of the UN Convention on the Rights of Persons with Disabilities (CRPD). Research on other UN conventions shows that especially NGOs (Zwingel, 2005) and IO-related actors (Jörgens et al., 2016) play a central role in the promotion of international treaties. As a consequence, we expect NGOs (and in particular disabled persons' organizations; DPOs) and IOs to take central positions within the Twitter network as well (H1). Furthermore, particularly in policy networks popular actors are often used as levers in order to increase the reach of information of less popular actors (leverage politics; Keck & Sikkink, 1998). As popularity on Twitter can be measured using the number of followers a user has, we assume that actors with a high number of followers often receive mentions and retweets (H2). Moreover, one basic network theoretical assumption is homophily which states that actors preferably build contacts to similar others. Following this, we expect to observe a tendency of actors in the network to communicate with other actors from the same organization type (H3). Studies about political offline networks also show that actors often use existing direct (reciprocal) or indirect (transitive) contacts to further disseminate their information (Ostrom, 1998; Ball & Junemann, 2012). Thus, we expect a similar pattern for the present Twitter online network (H4).

As data base we use Twitter data which were published in the context of the Conferences of States Parties to the CRPD in the years 2013 to 2017, with a specific filter for education-related tweets. These conferences bring together different state- and non-state actors to discuss issues related to disability rights, thus representing the whole range of actors involved in the debate about inclusive education. In order to test our hypotheses, we calculate Exponential Random Graph Models (Robins et al., 2007). This method offers the opportunity to inferentially test hypotheses concerning the formation of social networks. In line with our H1, the results show that especially IO- and DPO-related actors take central roles in the network under scrutiny. However, whereas DPOs use their central role to actively participate in the network, IOs remain rather passive. Moreover, there is a positive relation between the number of followers and the in-degree, that is, the frequency of being retweeted or mentioned (H2). Concerning homophily, the results show a general tendency of actors connecting with other actors from the same organization type (H3). This effect is especially strong and stable for NGOs and actors from research and is varying over time for IOs and private actors. Finally, the results contradict our assumption of reciprocal behavior within the network. In contrast, the tendency towards transitivity in the generation of contacts are especially strong, thereby partly supporting H4.

Overall, the results contribute to the understanding of education-specific Twitter network formation in the context of the CRPD by developing a statistical model. According to the results, the network is especially dominated by actors related to DPOs and IOs. Whereas IOs remain rather passive, DPOs seem to actively diffuse policy ideas concerning inclusive education on Twitter. Furthermore, particularly popular Twitter users are used as ‘levers’ to diffuse information published by less popular actors, suggesting the idea of leverage politics for Twitter policy networks as well. The general assumption of homophily cannot be confirmed entirely for the present network, probably due to the different objectives of the organization types. Whereas NGOs show their support by preferably retweeting and mentioning each other, IO-related and private actors seem to be also interested in the diffusion of information in the overall network. The high occurrence of transitive ties in contrast to a rather small number of reciprocal ties suggest that even in online policy networks social mechanisms, such as social capital and trust, are at work, at least to some extent. Regarding both empirical evidence and (political) network theory, these findings indicate a certain similarity of strategical mechanisms between offline and online policy networks. Thereby, they open up for further research at the intersection of Twitter networks and education policy and can thereby represent a valuable and innovative approach to the general understanding of global education policy.

Authors

• Johannes Schuster, Freie Universität Berlin

• Nina Kolleck, Freie Universität Berlin