Exponential random graph models (ERGMs) for multilevel networks (Wang et al. 2013) offer a statistical framework that captures complicated multilevel structure through some simple structural signatures or network configurations based on assumptions that network ties within and across levels may be conditionally dependent (Wasserman and Pattison, 1996; Snijders et al, 2006; Robins et al, 2007; Pattison and Snijders, 2013). Combining nodal attributes and multilevel network structure, Wang et al (under review) proposed social selection models (SSMs) for multilevel networks that tests how nodal attributes may affect network structures across different levels. ERGMs and SSMs models the endogenous network processes with or without nodal attributes as covariates, auto-logistic actor oriented models (ALAAMs) (Robins et al, 2001; Daraganova and Robins, 2013) on the other hand tests how nodal attributes may be affected by attributes of other nodes and the given network structure. We extend ALAAMs to multilevel networks which allow hypothesis testing on how multilevel network structure may affect individual outcomes. We present ERGMs, SSMs and ALAAMs to a network of researchers and their affiliated laboratories (Lazega et al, 2008) to demonstrate the features of these models, and reveal the multilevel network processes and individual’s strategies in achieving greater performance based on the model interpretations.