The literature on political socialization highlights the importance of parents and friends, but it is rare to find studies analyzing these socializing agents in the same model. In contrast, friends are often limited to one or a few friends that may not account for the actual effect of friends. The reason is that standard datasets do not collect information on the entire network of people's friends. Importantly, having an incomplete network can lead to biased estimates of network effects. To overcome this problem, we surveyed 419 students who recruited an additional 4500 social contacts who answered a shorter survey. Controlling for potentially endogenous network formation and using second-order peers to instrument for direct network effects, we find important political socialization from parents and friends on anti-immigrant sentiment and voting intentions among the students we survey. We also show that results differ if we only examine the impact of classroom peers, as is typically done in the literature. Surveying social contacts is a promising way to reach a complete social network, which overcomes data limitations in the current political socialization literature.