Sunday, March 30, 2008

The gods must do randomized trials

Chris Blattman is in Liberia and he's happy with what he sees;

I hum and haw about comparison groups, going through my impact evaluation 101 schpiel. I have serious concerns that one would or could develop a control group, let alone randomize, for such a program. So I dance delicately around the subject.

"Wait a minute," interrupts the Imam, "Are you talking about a randomized control trial?"

I gape.

"Oh I see!" says one Reverend Minister, "We need a control group! This is a good idea."

It turns out his holiness was once an agronomist. "This is just like our control plots for fertilizer. But how are we going to control for spillover effects?"

An older Methodist leader frowns sitting in the corner glowers. "Please, a moment," he says. "I see a real problem here."

Here it comes. Here is the doubt and questioning I expected. We're talking about a peace building exercise, not fertilizer on a farm plot. Even I have my reservations. This man, of an older generation, clearly has other priorities.

"How," he asks "are we going to select a proper sample?"

No comments: