career-ladder systems for teachers, strong teacher induction, and consistent, high-quality PD that is led by teachers themselves.

"There is a need to become much clearer about what is and is not good instruction," Jensen said via email. "U.S. districts should—and many do—have clear instructional models. But these should not be so granular that they lead to compliance responses," he added, saying that states and districts need to allow more room for educator expertise.

One case study from the Jensen report offers a glimpse at an elementary school in which the principal acts as the primary substitute teacher so that her faculty could spend time on classroom observation.

An August 2015 study by the teacher-training and advocacy group TNTP questioned the impact of PD activities in U.S. schools and criticized districts for overemphasizing workshops and other trainings that teachers often don't find helpful.

For all these types of models to operate effectively, channels for teacher collaboration are essential, the researchers emphasized. "This is a profound shift for many systems given the efforts to develop precise school performance measures over the past few years," the Jensen report says. "It requires faith and trust in the people making professional judgments."

To that end, advanced Shanghai teachers are turned into researchers, using the classroom to develop and test instructional approaches and interventions, the Zhang report shows.