Prossimo seminario permanente con format "joint talk" di Giovedì 27 Giugno 2024 con il titolo "Computational Social Science Methods That May Help With the Conundrum of Causality."

 

 

 

Abstract: 

 

Computational social science means the use of various computational methods to explain empirically-detected social patterns and dynamics. In this joint talk, we will focus on two cases that exemplify how these methods can help to solve problems of inference from data. In the first example, Veltri will explore the heterogeneity of treatment effects from an empirical study and in two scenarios of RCT(Randomized Control Trial)-generated data using causal tree and causal forest methods. Additionally, he will examine the use of causal discovery algorithms in the context of Bayesian networks and how they can be beneficial in developing complex causal models from observational data. In the second example, Squazzoni will explore the use of agent-based models to explore multiple causal paths of advice-seeking network formation and dynamics, by fitting existing network data with simpler assumptions than in previous Stochastic-Actor-Oriented Models (SAOM) network models. He will show that different generative causal processes can fit equally the same network data, thus arguing that we need a many-model thinking whenever treating the concept of causality.

 

 

Speakers:  Flaminio Squazzoni (Department of Social & Political Sciences, University of Milan) & Giuseppe Veltri (Department of Sociology and Social Research, University of Trento)