While randomized experimental designs are the gold standard in education research concerned with causal inference, non-experimental designs are ubiquitous. For researchers who work with non-experimental data and are no less concerned for causal inference, the major problem is potential omitted variable bias. In this presentation, the authors provide an overview of a multi-year effort for the evaluation of a large-scale, school-based intervention for which a randomized design has not yet been possible to implement, but for which there are rich, longitudinal data and many opportunities to exploit design features to ask: does the intervention promote the achievement of children in high-poverty, urban schools? City Connects is a student support intervention that has demonstrated positive outcomes for elementary and middle school students. It is designed to address the out-of-school needs that can impact success in school for low-income, urban students. Having observed positive outcomes on academic achievement for students attending City Connects schools via quasi-experimental methods the authors sought to assess whether the findings can be considered a result of the intervention. In conclusion, the authors recommended that other school-based interventions that cannot feasibly use a randomized control design instead employ a range of methods aimed at reducing endogeneity and getting closer to understanding whether a causal relationship might exist between the intervention and its outcomes. Tables and Figures are appended.