Mental health among students on college campuses has been deteriorating for the last two decades. Objective social connection, or the amount of time spent with others, is thought to be related to mental health levels and of critical importance to college students, a group in which establishing relationships and socializing with others is a primary motivator. Yet, in this population, we know very little about how objective indicators of social connection relate to mental health which is likely due, at least in part, to how difficult it is to measure such a complex phenomenon. Overcoming measurement limitations in past research, the present thesis was able to naturally capture objective social connection by sampling conversation via smartphone microphones in a sample of college students. Leveraging this unbiased measure of social interaction, in combination with computational methods designed to identify fine-grained temporal and spatial relationships between variables, provided novel insights into the dynamic relationship between social interaction and two mental health constructs relevant to college students -- stress and anxiety. In Study 1, we used daily levels of conversation and a network-based approach to show that stress was negatively related to concurrent and subsequent levels of social interaction. Importantly, we did so while accounting for other measures also captured by passive sensing (e.g., sleep, movement) known to be related to both stress and social interaction. Study 2A examined conversation at an even finer level and leveraged techniques commonly used in ecology to map temporal and spatial characteristics of conversation patterns, providing perhaps the first concrete look at the social landscape of a college campus. With the social structure mapped, Study 2B was able to demonstrate that social patterns differed as a function of anxiety. Namely, those with lower anxiety socialized more in dining and Greek locations. Collectively, these results demonstrated how an objective marker of social connection -- amount of conversation -- relates to mental health and illustrated multiple modeling approaches to fully capitalize on ever prevalent rich, intensively collected datasets. [The dissertation citations contained here are published with the permission of ProQuest LLC. Further reproduction is prohibited without permission. Copies of dissertations may be obtained by Telephone (800) 1-800-521-0600. Web page: http://www.proquest.com.bibliotheek.ehb.be/en-US/products/dissertations/individuals.shtml.]