When intuition isn't good enough

Recently I was at a talk given by Rachel Glennerster of the Abdul Latif Jameel Poverty

Action Lab (JPAL).  In speaking about cost-effectiveness, she gave examples from

several different education interventions in Africa:

1) merit scholarships for girls

2) free primary school uniforms

3) providing information to parents on how schooling increases income

4) deworming through primary schools

All the interventions sounded pretty good to me. I could see how all of them might increase the number of years children spend in school.

But JPAL compared randomized controlled trials of each intervention, and when they looked at the additional years of education you get for the cost, these are the results:

Source: JPAL

This isn't perfect information: the studies are from different countries, and as with any studies, there's some uncertainty. But it's a lot better than relying on guesswork.

In retrospect, it makes sense that parents would prioritize schooling if they know more about education's economic benefits to their children. But I can make up just-so stories for why any of these would be the most effective. You could have presented me with any of those interventions, told me it was a great method to get children more schooling, and I would have found it believable.

(In fact, the Madagascar study tested whether it works better to give parents statistics about children's expected earnings or to have "role models" speak to parents about how education benefited them. The role model intervention is more common, based on the theory that impoverished parents with poor literacy aren't able to understand statistics. But the study indicated that they are able to understand the information given, and that it actually works better than having role models speak to them.)

These studies go beyond asking "Does it work?" That's a good starting place — after all, some interventions don't work at all — but it's not enough. Everyone from small donors to government policymakers needs to know about cost-effectiveness. Unless we have infinite money, we need to know where our money will go the farthest so we can start there. If we have $100 to give, there's a big difference between buying school uniforms (which we expect to result in less than one year of additional education) and providing information to parents (which we expect to result in over twenty years of education!)

This is why research is important. I'm glad there are people out there like JPAL getting more information so we're not just guessing.  

You can hear Rachel Glennerster's talk and others from the Good Done Right conference here. Apologies for the occasional baby sounds in the background, which are my daughter.