Education experiments

economics ▪ policy

The Economist ran a piece on a simple question of development that everyone needs to be asking: For all the investment in expanding public education, is our children learning?* The surprise is that, for the most part, most people (including government officials and donors) don't know. What's interesting, one charity organisation in India suggests an effective way to figure this out.

From the Economist:

Pratham's response to widespread illiteracy and innumeracy was to experiment (emphasis added). It tried various remedies in half the schools in a district or city, picking which half at random. The remaining schools provided a control group with which to compare the results of its efforts. One of its more successful ventures was to hire unqualified high-school graduates to provide remedial education for students falling behind. These balsakhis (which means “children's friends”) were cheap, paid about $10-15 a month, and quick to train, receiving only two weeks of prior instruction. Because they did their work in hallways or even under trees, there was nothing for governments or donors to build.

Nonetheless, the instruction they offered was surprisingly effective. In Mumbai it raised the chances of fourth-year pupils grasping first-year maths by 11.9 percentage points. It improved their chances of mastering second-year literacy by 9.9 percentage points. The gains in Vadadora (formerly known as Baroda) were smaller, but still worthwhile.

What I took from these paragraphs was not that "children-friends" is a good idea (it might or might not work in Indonesia), but that it is important to have a good way to infer which policies work best for Indonesia's context.

To be honest, I have a bias for randomized experiments because I am currently involved in one. But, although such experiments tend to be expensive, the kind of rigor offered by such experiments can allow reliable inferences about what policies really work - and what don't. Depending on the expected benefits, this is the kinds of thing that non-government organisations and the government needs to explore much more. There are also other, more statistically-demanding means to evaluate, and these need to be employed more as well.

The newspaper concludes:

Pratham's remedy may not apply to Africa, where fully trained teachers are still relatively cheap. But [Pratham's] approach — measure, experiment, evaluate — should...

Indeed.

---

Notes:
*The Economist article is a subscription-only article for about a week – then, it's free.
** The paper discussing the findings can be found here