Selection bias and Roger Clemens

Lots of studies conducted at CMF is based on a principle called randomized trial. This is a common research methodology in pharmaceutical science where patients are randomly divided into two groups and researchers give a new drug they want to test to one group, called treatment group, and a placebo to another, called control group. Since the drug is assigned randomly, on average and with a sufficient number of samples, we can assume that characteristics of patients in the two groups average out. Some of characteristics include factors like innate physical differences, general attitude towards health which may turn into a number of pro-health habits in everyday life which may affect the efficacy of the drug, etc. As a result, researchers can safely assume that the only difference between the treatment and control group is the fact that the treatment group received a real drug, and therefore, any difference between the two groups can be attributed to the drug.

The same logic applies in estimating impacts of microfinance. Previously it was farily common to look at a group of people who signed up for a microfinance program and compare their welfare with those who didn't. The problem with this approach was, those who signed up for a development program may have something inherently different from those who didn't. Maybe those are the ones with entrepreneurship. This is called selection bias.
So what researchers do in a randomization trial is look at an eligible population for a microfinance program, and then provide loans to randomly selected individuals.

It seems like the application of randomize trials, or at least the concept of it, is getting more popular. Knowledge@Wharton reports an interesting case where a randomized trial was called for during the doping scandals that rocked MLB last year.