My research aims to evaluate and develop important insights around social policy interventions focused towards youth at risk of becoming involved in violence and the criminal justice system. The dissertation will consist of three essays that build on one another with the goal of answering closely related and complementary questions. The first essay seeks to evaluate a promising behaviorally-informed program for youth disconnecting from school in a large-scale randomized controlled trial (RCT). The second essay focuses on measuring any possible spillovers from this intervention on the peers of the study youth through a novel use of administrative data. The last essay seeks to learn how this program achieved longer-term behavioral change for youth through an additional field experiment run in schools. First, I seek to understand what type of programs can be effective for a population of youth who are disconnecting from schools and are unlikely to fully benefit from an in school intervention. My first essay provides evidence on how to engage and achieve long-term behavioral change for youth at risk of becoming involved in the criminal justice system outside of institutional settings. Specifically, I evaluate a community-based program called Choose to Change: Your Mind, Your Game that uniquely pairs an advocate from the community providing holistic mentorship with a behavioral health component (trauma-informed cognitive behavioral therapy) to help youth process their past traumatic experiences and develop decision-making tools that can help them navigate difficult situations. Evaluated through a large scale RCT in Chicago's most violent neighborhoods, results to date show the program to be highly effective at reducing total and violent arrests in number of arrests per person and having any arrest up to two years post random assignment, in addition to improving school engagement. Given the importance of peer connections among adolescents, my second essay seeks to understand how such programs can have ripple effects in the larger community and neighborhood. Specifically, we seek to measure if this crime-reducing intervention had spillovers on the peers of the study youth. Using exogenous variation in a peer's treatment status created by the crime-reducing RCT, we estimate peer effects using administrative co-offending data from the Chicago Police Department and cohabiting data from Chicago Public Schools. We find little evidence that the intervention influenced peers criminal behavior among the co-offending network of youth. We find some evidence that the intervention impacted the cohabiting peers of the study youth when the study person is older than the peer, by reducing the probability of having any violent arrest within 12 months. Future work is needed to better understand the spillovers of crime reducing interventions among adolescents. Lastly, my third essay seeks to understand why such crime-reducing programs can work for participants despite the strong peer influences among adolescents. Based off of qualitative work, this paper sought to explore if the crime-reducing RCT achieved such results because they helped youth increase their self-worth to respond differently to the peer pressures of their immediate communities. I test this idea in a field experiment in partnership with the Chicago Police Department (CPD) where C2C study youth are given a chance to sign up for CPD youth programming with randomly varied privacy assurances on the sign-up forms (the “public” forms suggested their peers might found about their interest in CPD). Given the stigma of working with CPD among these communities, I exploit the randomized experimental design of C2C to compare this public-private difference for the treatment versus control groups. I find suggestive evidence that a public form decreased take-up in the C2C control group, but significantly increased take-up in the C2C treatment youth, suggesting a response to peer pressure in a way that counteracts social norms.