This paper tests existing educational theory regarding the effectiveness of interactive pedagogy in order to determine the predictive ability of interactive pedagogy during ninth grade math and English classes towards graduation. This is done using a mixed methods approach which examines correlations between student teacher interaction and dropout through logistic regression models using data from the Philadelphia Educational Longitudinal Study (PELS), and qualitatively links interactive methods examined in the quantitative section (and some not examined) to pedagogy the sample of Philadelphia ninth-grade students find effective. Quantitative results provide strong evidence that low student-teacher interaction predicts a higher probability of dropout, as well as make suggestive connections between interactive methods and graduation. The qualitative results link both the non-interactive methods as part of the "pedagogy of poverty" and in-class discussion to pedagogy which students find effective in creating attachment to school, confirming the quantitative findings. Implications for educational theory are discussed.