Indeed, there can be multiple assumptions, and, hence, multiple models, which demands for some assessment criterion. Network literature proposes at least two approaches. First, we can estimate likelihood-based measures, such as Akaike informational criterion or AIC (Hunter et al., 2009) comparing model’s AIC with some arbitrarily chosen baseline or across competing models. Second, there is so-called empirical goodness-of-fit approach, in which sample of simulated from a model networks is compared with observed one. If simulated networks resemble observed in many different respects (especially those which has not been directly modeled) there is, then, the evidence that proposed model properly fits to the data, and, therefore, may be assumed as reflecting underlying tie-formation processes (Robins et al., 2007; Goodreau, 2007).