Autologistic actor attribute models are used to examine the probability of the presence of an attribute within the context of network data. A multi-level extention is proposed, following the hierarchical framework of (Koskinen & Snijders, 2023), allowing for multivariate extension and thus allowing multiple networks to be examined simultaneously. The model is applied to a data set stemming from the Stockholm Birth Cohort project, using a binary leadership variable as the response attribute of interest. In a Bayesian context, utilizing a multivariate exchange MCMC algorithm to overcome to troubles with double intractability for the models parameters, the results point in the right direction. While the presence of outliers show that each network does not converge smoothly, the group-constant parameters mix well. Further research into model evaluation, missing values imputation and additional experimentation with hyperparameters are recommended.