Using mixed methods to interpret the findings of a quantitative program evaluation in the context of a qualitative assessment of program implementation, this paper addresses the effect of a recent policy change that allows TANF participants to retain more of the child support paid on their behalf. The proposed paper considers lessons for program implementation and evaluation. Most families participating in TANF are also engaged in multiple other public programs, but coordinating and rationalizing administrative, staffing, and benefit rules across programs has been a persistent challenge. Improving collaboration across these public programs provides the opportunity for increased efficiency for agencies and more streamlined experiences for program participants, which is critical during a time of scarce public resources.  However, the success of cross program collaboration is largely dependent on the success of its implementation. The proposed paper will discuss the results of the Collaborative Strategies Project (CSP), a demonstration project designed to support collaborative partnerships among State Child Support Enforcement and TANF agencies and university scholars. Previous research found that incorrect information was being provided to TANF applicants about the interaction of TANF participation and child support receipt (Ybarra and Noyes, 2009). In addition, recent child support policy changes increased the amount of child support received by current and past TANF participants, thus enhancing the potential gains for better informed TANF applicants. By increasing collaboration between child support and TANF workers and providing them with opportunities to learn about new policies related to child support and TANF receipt, this project aimed to improve the economic well-being of low-income children and their families by increasing their access to accurate and complete information about child support policy and the interaction of the child support and TANF systems. Program and policy implementation research indicates that caseworkers are an important source of information for program participants and that the discretion caseworkers wield actually creates policy at the ground level; thus the CSP developed and implemented a state-wide training of child support and TANF workers that would promote collaboration and inform workers of the new child support policies. Observations of caseworker and client interactions before and after the state-wide training suggest that the training did little to change the behavior of TANF caseworkers and that workers were not applying what they learned in the training to their client interactions. This paper highlights the results from a Difference-In-Difference analysis (Cancian and Selekman, 2012) which assessed the impact of the policy change on TANF applicant participation decisions. Drawing on implementation literature, the discussion of these quantitative results will be couched in the context of TANF caseworker knowledge and working conditions. This mixed methods analysis focuses on implementation as a means to improve policy outcomes and a way to interpret the quantitative evaluation. This project contributes to understanding policy development and implementation at the ground level. Using observational data to interpret the findings of the quantitative analysis using administrative data allows for a deeper, more nuanced interpretation of the results of this collaboration project.