If education is sensitive to the adaptive group contexts, then it stands to reason that you wouldn't see "total" reforms working in groups lacking characteristics of a strong coherent (adaptive) group. Thus one should see total reforms working in conditions where groups have strong norms, strong leaders, strong sense of purpose, etc.
Indeed this is what one finds. The large, ambitious New American School Reform project of the late 90's found that total reform depth was correlated with strong principals and high levels of teacher self-efficacy (I feel I can succeed with any student).
This fits with what would expect from a multi-level selective perspective. While this is certainly not evidence of the necessity of multi-level selective framing, it does show that yet another small piece of the empirical world doesn't negate such framing.