Home Sociology Exponential Random Graph Models for Social Networks Theory, Methods, and Applications
Participants were asked about their gender (i.e., a visible attribute) and responded to items measuring social identity (i.e., an invisible disposition) at the beginning of the first semester. With regard to gender, male is coded as 0 and female as 1. The group identification scale (Karasawa, 1991) measured the strength of participants’ identification with the education department. This scale contains two subscales: the identification with member (IDMEMBER), including the feelings of closeness toward classmates (five items), and the identification with group (IDgrOup), including attachments with the department (seven items), all of which were rated on 7-point scales. The averaged score across all items was separately calculated for each subscale. Higher scores of IDMEMBER or IDgrOup indicate stronger identification with other classmates or the department, respectively. The attribute variables for these longitudinal models were measured at the first time period of the study.
Figure 19.1. Triadic configurations used in models. Labeling of configurations comes from standard triad census for directed graphs.
The longitudinal exponential random graph model (ERGM; Chapter 11) was used to analyze the data. Purely structural and actor-relation effects were included simultaneously in the model. The four types of friendship networks were each modeled separately and longitudinally, using the network at the first period (matriculation) as an initial state for the last period (end of the academic year). LPNet, a version of the PNet software (Wang, Robins & Pattison, 2009) for longitudinal data, was used for the parameter estimation.
|< Prev||CONTENTS||Next >|