Influence of networked partisans on perceptions of media bias

Media bias influences how voters perceive parties and policies, through print, broadcast, and social media. Individual's viewpoint on media bias is influenced both by consumption of media content and by the peer pressure of their networked political allies and opponents. This work models media bias perception through a network analysis approach, using a Bayesian probabilistic model to represent agents uncertain beliefs, which allows for evolving opinions based on external stimuli (e.g., editorials) and peer influence (e.g., opinion exchange).

We introduce partisans—agents who refuse to change their opinions—in networks of political allies and opponents. Even a single partisan can destabilise an allies-only network, causing agents’ beliefs to oscillate between the true bias and the partisan’s belief. The duration of stable beliefs increases with more partisans but decreases with opposing partisans. In opponents-only networks, asymptotic learning is achievable regardless of the presence of partisans. Mixed networks show complex dynamics where partisan placement and media signals lead to varying outcomes. Strongly balanced triads with partisans show intermittency, while unbalanced triads can either converge on the true bias or fluctuate, depending on partisan placement.