Enormous variation exists in animal sociality, and understanding the evolution and maintenance of this variation is one of the major themes in behavioral ecology. Interaction among group members is often a major component of social living, yet detailed studies of the causes and consequences of social interactions are relatively rare. Network analysis offers a broad framework for studying patterns of interactions, but its general utility remains to be tested in animal systems. My dissertation work explores social network analysis as a general methodology for studying animal behavior and applies it to examine biological causes and consequences of individual social attributes in yellow-bellied marmots (Marmota flaviventris). I first reviewed existing network literature in many fields of study and discuss potential applications and limitations for social network analysis in animal behavior. The network approach has the strengths in providing quantifiable measures of social attributes at multiple levels of analysis, and if applied properly, can offer diverse insights into animal social behavior. Next, I examined the ontogeny of individual social attributes and the role of age, sex, and kinship in structuring marmot social networks. Patterns of development in affiliative and agonistic attributes with age indicate that younger marmots contribute more to maintaining social cohesion, while older marmots play a more prominent role in competitive networks. Additionally, affiliative networks were significantly structured by age and kinship, suggesting that these factors strongly shape marmot social structure. Finally, I tested hypotheses about the relationship between individual social attributes and multiple fitness measures (annual reproductive success, number of ectoparasites, diversity of endoparasites, and basal stress). In general, there were more benefits than costs associated with high attribute values for both affiliation and agonism, implying a large potential adaptive value of sociality even in this facultatively social species. However, patterns of association varied considerably between age-sex groups, indicating that the adaptive value of sociality changes with life stage and differs between females and males. In marmots, attributes measuring absolute amounts of direct interactions had more observable fitness correlates than attributes measuring indirect interactions or relative amounts of interaction. Overall, this network attribute-based approach was revealing in marmots and should also have broad utility across other animal systems.