Home Sociology Exponential Random Graph Models for Social Networks Theory, Methods, and Applications
The use of dyadic covariates in ERGMs is often relatively straightforward. The interest lies in whether the dyadic covariate is associated with the
Table 8.3. Dyadic covariate configurations for ERGMs
presence of an observed tie. In Table 8.3, we present an “entrainment” effect that captures such an association for both directed and undirected networks. For example, one of the research questions might be to explore whether people from the same department are more likely to form an advice relationship. Here, “same department” is a dyadic covariate. A continuous dyadic covariate might be the amount of time two individuals work together on the same project.
If the dyadic covariate is another network, and something more than a simple control is of interest, then it is possible to extend these effects by using a bivariate ERGM to model the two networks simultaneously, as presented in Chapter 10. Sometimes, of course, a dyadic covariate is not another network but involves a complex conceptualization in its own right. An example is when dyadic covariates are derived from geographic space to which we now turn. (Various triadic effects are also possible as presented in Chapter 10 on multiple networks, such as i is friends with j, j is friends with k, and therefore i trusts k.)
|< Prev||CONTENTS||Next >|