Causality is an inextricable part of the educational process, as our understanding of what works in education depends on our ability to make causal attributions. Yet, the research and policy literature in education tends to define causality narrowly as the attribution of educational outcomes to intervention effects in a randomized control trial context. This reduction of the educational process to simple input-output relationships leaves important questions unattended about how change is produced in educational systems and how observed results can be predicted based on the propensity toward change in the baseline settings of those systems. This paper considers these questions from a Complex Dynamical Systems perspective, and concludes that answers to them can qualify the findings from at least some experiments.