Tuesday, 4 November 2014

Charter schools - could do better

As NZ's charter partnership school experiment gets cracking, now under the guidance of David Seymour, a new study(pdf) has emerged looking at 10 years of data from the Texas experiment.

The big news is that it shows that Charter Schools Get Better Over Time, a line that has reverberated around social media today.

Well I should hope so. Isn't that what we expect of all schools? It seemed so odd to trumpet such a weak result that I had a wee peek at the paper itself and found that

  • the result is actually much stronger that just "get better"
  • but highlighting its strength would be embarrassing to charter school fans. 
What they study actually found is that charter schools were really terrible at the start, but are now nearly as good as normal schools. Texas started their experiment in 1995. The study period was 2001-2011, so there were 6 years excluded from the start. Even so, by 2001 student achievement results were still appalling. Here are the distributions of "value added" in reading and mathematics across the sample for charter and normal (TPS) schools.

What you want is a nice tight (peaked) distribution indicating that you're doing pretty well for almost all students. The reading distributions improved a bit faster than the maths ones but even at the end, of the sample they aren't quite as good are they?

I get the idea nicely described by Donal Curtin that kids are so heterogeneous that more variety in schools could be useful. However if that really was the objective, it's far from obvious that the charter school system is the best approach. They seem hella costly (though the Minister has no figures), and the Texas experiment suggests that many students got a bum deal for many years. That is a massive cost to future citizens. I'd love to see the analysis behind this policy.

No comments:

Post a Comment