This paper addresses the identification of peer effects in social networks, extending the linear-in-means model to allow for individual-specific reference groups. The authors provide necessary and sufficient conditions for identifying endogenous and exogenous effects, both in the absence of correlated effects and when controlling for correlated effects at the component level. They show that endogenous and exogenous effects are generally identifiable under network interaction, although identification may fail for certain network structures. The paper also discusses the implications of network characteristics, such as density and intransitivity, on the quality of estimates of social effects. The results are illustrated through Monte Carlo simulations and applied to various models studied in the literature, including those by Manski (1993), Moffitt (2001), and Lee (2006). The findings highlight the importance of considering network structure in the identification of peer effects and provide a theoretical foundation for empirical studies on social interactions.This paper addresses the identification of peer effects in social networks, extending the linear-in-means model to allow for individual-specific reference groups. The authors provide necessary and sufficient conditions for identifying endogenous and exogenous effects, both in the absence of correlated effects and when controlling for correlated effects at the component level. They show that endogenous and exogenous effects are generally identifiable under network interaction, although identification may fail for certain network structures. The paper also discusses the implications of network characteristics, such as density and intransitivity, on the quality of estimates of social effects. The results are illustrated through Monte Carlo simulations and applied to various models studied in the literature, including those by Manski (1993), Moffitt (2001), and Lee (2006). The findings highlight the importance of considering network structure in the identification of peer effects and provide a theoretical foundation for empirical studies on social interactions.