A new study from Stanford’s Center for Research on Education Outcomes concludes that charter schools “post superior results with historically disadvantaged student subgroups.” Via Twitter, Neerav Kingsland asks, “can you run this study through your skepticism wringer?”
The full study is a couple hundred pages long, so the answer is no. What’s more, the main conclusion of the study is actually about the consistency of charter school results over time, not charter performance among minorities. Their primary takeaway is simple: charters that start out strong stay strong, and charters that start out weak stay weak. They don’t get better over time as they “work out the kinks.”
I actually find that reasonably plausible. Nonetheless, although I don’t have time to read the full report right now, I can read the executive summary and point out a few issues:
Short time frame: “Using the broad range of data that CREDO has developed in partnership with 25 state education agencies, we follow student-level performance in schools from their opening through their fifth year.”
Small sample size: “While the available data does not extend into the past, far enough to observe the birth of all [charter management organizations], a limited number of CMO ‘births’ are evident in the data window at our disposal and it is possible to observe their flagship school’s performance before and after replication.”
Confirms own biases: “Interviews with school staff along with our own observations of school activities and operations have formed the impression that the ‘rules’ of a school get set early on in the life of the school….If our admittedly limited, qualitatively-based conjecture is true and more generally supported, we conjectured that it should be possible to observe the phenomenon quantitatively and test the hypothesis statistically.”
These are reasons to treat CREDO’s conclusions with caution. Nonetheless, it’s worth pointing out that some of CREDO’s results aren’t super friendly toward the charter movement. Lousy charters stay lousy over time, for example, and that’s especially true for middle and high schools. Charter performance, on average, is pretty average. There are very few observable attributes that serve as signals of charter quality. So while their study confirms the bias they took into it—something that seems to be an absolute scourge in the ed reform field—it wasn’t all just confirmation bias. A few of their results were challenging for the charter school movement.
Overall, I don’t have a big issue with this study. It seems to be fairly modest in its claims, and its main policy recommendation is that since charters display a wide range of performance, they should be monitored closely and shut down early if they perform poorly. I’m generally in favor of lots of ruthless experimentation, so I guess this appeals to my biases.
As for whether charters really are especially strong among minority and low-income populations, that’s been the subject of dozens of studies, not just this one. The CREDO study seems to confirm that it’s possible for a charter to do especially well among these groups, but we still don’t how easy it is, how scalable it is, how replicable it is, or how expensive it is. The jury is still out.