Séminaire GT C, Pedro Ramaciotti Morales – Jeudi 31 mars 2022 à 10h

Titre: Inferring ideological positions from social graph data: methodological challenges and applications in sociology, political sciences, and internet governance

Résumé: Despite the critiques of which they are a target of, polls and surveys remain the standard instruments for measuring « public opinion ». But recent developments in social network analysis have paved the way for the estimation of opinions using digital behavioral traces, such as digital social networks of friendship, the sharing of contents, and other choice data that can be exploited with psychometric models and network embedding methods. In this talk, I will present advances in tackling the challenges of these methods, in particular in adapting them for social settings beyond those which gave them birth, in the US socio-political system. European settings, for example, contrast with that of the US in that they do not lend themselves to descriptions using binary ontologies (such as Democrat- and Republican-leaning actors) nor with one-dimensional spatial models (e.g., ranging from most liberal to most conservative). General ideological and attitudinal estimation of individuals and entities of the digital public spaces (such as media outlets or social groups) demand multi-polar, multi-dimensional empirical spatial opinion models, which are the central topic of this talk. I will also show some applications that can be developed by tackling these challenges, including the study of the ideological dynamics of social movements, the role of ideology in mistrust and misinformation phenomena, the role of news media and influence dynamics, and in advancing the understanding of phenomena that are difficult to conceptualize and measure, such as polarization.