Posts

Showing posts from May, 2016

Why I Work on Institutional Change at Home but Not Abroad

When the movement termed “effective altruism” started, one of the major currents was the movement in economic development toward conducting what are termed randomized controlled trials (RCTs) for poverty alleviation programs. A randomized evaluation is the way the FDA tests new drugs before they go on the market: a treatment is randomly provided to some of a given collection of individuals but not to others. Because the treatment is allocated randomly, there should not be any systematic differences between those who get it and those who don’t. So all sufficiently large differences between treatment and control should be due to the treatment and you can estimate, without any bias (in an ideal case) the effect of the treatment.
This is very different from the way many studies are done: longitudinal studies in medicine (which are usually how doctors get initial looks at the differences between different diets) suffer from selection bias: people who drink are likely to differ from people w…