Wednesday’s post reported on a field experiment by Roland Fryer in which he imported a number of charter school techniques into low-performing Houston schools. The techniques made a very large difference in student outcomes in math (not in reading though).
The changes weren’t free. Are they cost-effective? Are they scalable?
The changes are clearly cost-effective for math. Here’s one way to put the comparison. One estimate from other studies is that lowering class size from 24 to 16 improves outcomes by 0.22 standard deviations over three years. That kind of decrease in class size increases instructional costs by about a third. Fryer’s bundle of interventions increased scores by about 0.16 standard deviations in one year. But the cost is only about $1,800 per student–far less than the cost of lowering class size.
Overall, Fryer estimates the internal rate of return for his interventions is about 13 percent. That’s a rate of return that would make most CEOs very, very happy. The most effective part of the intervention was intensive math tutoring, but the tutoring was also the most expensive. Without the tutoring component, the estimated rate of return rises to an incredible 22 percent.
The interventions were very clearly cost-effective. Fryer raises the separate issue of whether they are scalable. For example, Fryer’s team had to conduct over 300 principal interviews to find 19 principals they were happy to put in leadership positions in the treated schools. Similarly, more than a third of teachers were replaced. When you are treating a handful of schools, you can pull top staff from other schools or other school districts. But that doesn’t work when you need to change large numbers of schools. That requires a huge pool of top-notch educators.
So–and yes I’ve said this before–if we want to have a large number of teachers who can implement the kinds of reforms that really work, we’d better start paying the kind of money it takes to attract large numbers of the best and brightest into the profession.