New Regression Models with Egocentric Social Network Data: An analysis of political party preference in Japan

This paper introduces some regression models for a categorical dependent variable with data from egocentric social networks as covariates to analyze the determinants of the outcome for the subject and those of the extent of agreement or disagreement between the outcomes for the subject and the persons to whom he/she is directly connected. The merits of those models are their wide applicability to surveys that collect data on egocentric social networks and their capacity to identify the determinants of agreement between the subject's and his/her friends' attitudinal or behavioral outcomes, controlling for the tendency to agree due to homophily in the choice of friends. An illustrative application using data from the 2004 Japanese General Social Survey shows that several substantively distinct characteristics of egocentric networks affect the subject's political party preference and the extent of agreement in the preference between the subject and that of his/her significant others.