Podczas Seminarium POLPAN 21 maja 2019 r. dr Michał Bojanowski (Akademia Leona Koźmińskiego) wygłosił referat pt. „Core discussion networks in Poland and America through the lens of Exponential-family Random Graph Models applied to egocentrically sampled data”.
People discuss certain important topics only with selected group other people. Members of these “core discussion networks” are probably very close and trustworthy. World Value Survey statistics on interpersonal trust report that around 40% of Americans agree with the statement that “most people can be trusted”. This percentage in Poland is around 20%, the lowest among the countries in European Union. Do the two countries differ in terms of with whom the people tend to discuss important matters? We pose the question about the structure of core discussion networks in Poland and America — what are the important factors explaining why certain discussion ties are but some others are not. We address this question by analyzing data from (American) General Social Survey and Polish “People in networks” survey. Using Exponential-family Random Graph Models fitted to egocentrically-sample data we (1) evaluate the importance of exogenous factors such as age, gender, or educational level and endogenous degree-related effects on the structure of discussion networks, and (2) compare the role of these factors in America and Poland.