In many applications of network analysis, it is important to distinguish\nbetween observed and unobserved factors affecting network structure. To this\nend, we develop spectral estimators for both unobserved blocks and the effect\nof covariates in stochastic blockmodels. On the theoretical side, we establish\nasymptotic normality of our estimators for the subsequent purpose of performing\ninference. On the applied side, we show that computing our estimator is much\nfaster than standard variational expectation--maximization algorithms and\nscales well for large networks. Monte Carlo experiments suggest that the\nestimator performs well under different data generating processes. Our\napplication to Facebook data shows evidence of homophily in gender, role and\ncampus-residence, while allowing us to discover unobserved communities. The\nresults in this paper provide a foundation for spectral estimation of the\neffect of observed covariates as well as unobserved latent community structure\non the probability of link formation in networks.\n