A psychometric analysis of information propagation in online social networks using latent trait theory

The paper explores use of psychometric analysis based on latent trait theory to study quality of information propagation in online social networks. The collective intelligence of users of the network could be used to determine credibility of information. We use the latent trait of ability of users to distinguish between true information and misinformation as a measure of social computing in the network. Using repropagation features available in these networks as an affirmation of credibility of information, we build a dichotomous item response matrix which is evaluated using different models in latent trait theory. This enables us to detect presence of misinformation and also evaluate trust of users in the sources of information. Trust between users and sources of information is further used to construct a polytomous matrix. The matrices are evaluated using polytomous latent theory models to evaluate the types of trust and segregate possible collusion of users to spread misinformation. We show experimental results of psychometric analysis carried out in data sets obtained from `Twitter' to support our claim.