Modeling Homophily: A Computational Test of Merton and Lazarsfeld's Thought Experiment and Its Extension.

Monday, 11 July 2016: 09:45
Location: Hörsaal 27 (Main Building)
Oral Presentation
Carmelo LOMBARDO, Sapienza University of Rome, Italy
Enrico NERLI BALLATI, Sapienza University of Rome, Italy
Pasquale DI PADOVA, Sapienza University of Rome, Italy
This paper focuses on the results of a computational experiment conducted with an agent-based model grounded on Merton and Lazarsfeld’s homophily theory. In “Friendship as a social process” the authors described homophily as “a tendency for friendship to form between those who are alike in some designated respects”, considering both status homophily and value homophily. In that essay, two causal micro-mechanisms were presented to account for the observed macro-patterns of value homophily: selection and adjustment. Such theory was a pioneering attempt to show how macro-consequences originate from individual preferences and the two micro mechanisms have played a pivotal role in the history of research on social influence and dynamic networks. Lazarsfeld developed a thought experiment to explain how homophily emerges, supposing 800 social actors linked together by 400 social ties. Using NetLogo language, we formalize and simulate an agent-based model in order to dynamically generate the process suggested in that work. Therefore, we simulate four scenarios to stress the original theory. The first scenario is from our previous formalization of Lazarsfeld’s thought experiment and his description of the system of action. Here, we introduce counterfactual hypotheses formulated by the authors concerning the above-mentioned causal mechanisms. In the second scenario, agents are assumed to interact in a dynamic network more complex than the one conceived by the authors. In the further two scenarios, on the basis of Blau’s theory of social structure, we introduce two hypotheses concerning the impact of status homophily and actor’s social position on selection and adjustment. Reading the simulation results through the four-fold typology of social ties proposed by Lazarsfeld, we can clearly recognize four macro patterns generated from the micro level by the interactions between individual agents.