Social Network: A New Paradigm for Modeling Human Interaction: Implications for Statistical Inferences

A broad of spectrum of disciplines have adopted social network data to examine relevant contextual issues in a wide array of fields. Yet, statistical methods to address biases in statistical inference introduced by the between-subjects relationship within the context of node, or subject, interaction in social networks are underdeveloped. Traditional statistical models define relationships among measures of within-subject attributes, i.e., measures of attributes from each subject. The between-subject attribute for node (subject) interaction in social networks is both conceptually and analytically different from the within-subject attribute. As a result, conventional statistical methods such as t-test and linear regression models are fundamentally flawed when applied to model between-subject attributes in social network settings. We illustrated fundamental differences of the betweenand within-subject attributes and resulting implications for social network data analysis of social network densities. We also proposed a new paradigm to model between-subject attributes and illustrate the approach with the analysis of social network density.