In Education Reform, Effect Size Really Matters!
Interesting article about Baton Rouge today got me thinking about performance standards in education. The gist of it is that parents in a rich area of Baton Rouge are attempting to secede from the city to maintain the quality of their local public schools. Not only will property taxes in the new municipality be substantially higher per capita, but they can keep poorer kids out of the schools. The fact is that this is Louisiana, and so in their mind “our kids” are white and “those kids” are black, and so no doubt old-fashioned racism plays a strong role. But it’s also, I think, a case of responding to incentives.
Do you believe that educational results are driven primarily by teaching or by socioeconomic background? If it’s the former, than the move by Baton Rogue’s rich residents is perfectly rational – by concentrating their resources in a richer municipality, there will be more funds per capita which will translate into a higher quality of education. If it’s the latter, their move is no less rational – by keeping poor kids out of the school, they will have a higher-performing school purely by selection effects. The latter scenario is a much more disturbing one for people who want a more fair society, because it has a truly pernicious interaction with “merit-based” pay for teachers and school funding.
If education outcomes are dominated by socioeconomic background and teaching has a smaller effect, then the move to merit-based school funding will be disastrous for poor children. Under the current regime, schools which do poorly on standardized testing have resources cut, and schools that do well receive greater resources as an incentive scheme. If teaching effects dominate, this regime should improve overall outcomes – while events like East Baton Rouge’s secession will hurt, hopefully this will be outweighed by a rise in teaching quality. If socioeconomic background has larger effects, on the other hand, then we should expect a continual downward spiral in outcomes for poor children as their schools are deprived of incentive-based funding. Good teachers will also steer clear of these schools, knowing that they will be punished regardless of their actual skill. And as students are further segregated, they will be sorted into schools offering increasingly poor prospects.
The effect of socio-economic background doesn’t even have to be larger than the teaching effect for this downward spiral to occur. For incentive-based programs to be effective, the teaching effect must be very large indeed – there’s both the strength of the effect and there’s the question of how effective an incentive scheme you have put in place. If teaching effects are huge but your incentive system is poor, you might just get rampant teacher-led cheating alongside a continuing deterioration in standards for poor children.
The long and short of it is – if you believe socio-economic background matters at all in educational outcomes, you should be extremely skeptical of incentive-based education reform proposals that don’t have a clear empirical record behind them.