Education has no shortage of programs that work in a pilot and disappoint district-wide. Professional development is the clearest case. In the Kraft, Blazar, and Hogan meta-analysis, coaching posted the largest effects of any PD model — but the authors flagged the same catch other researchers keep finding: effects shrink noticeably when programs expand beyond their original scale.
Why Effects Shrink
- Dosage dilution: a coach who served 15 teachers well is stretched across 60 — and the individualized cycles that drove the effect become drive-bys.
- Fidelity loss: programs designed by researchers get handed to busy staffs with less training and less say in the design.
- The wrong problem gets scaled: districts scale the sessions and materials, not the mechanism — the relationships and feedback loops that actually caused the change.
As Education Next's analysis of scaling coaching concluded: the challenge is not proving personalized training works — it is making it standard practice without hollowing out what made it work.
What Leaders Can Do
The research points to a few protections: cap caseloads and give coaches real release time; prioritize the teachers for whom evidence shows the largest effects (novices and struggling teachers) instead of spreading everyone thin; and measure the mechanism, not just attendance — count coaching cycles completed, not seats filled. And teachers: if your school is expanding a program, ask for the version that was tested, not the condensed one.
