The problem of doing rigorous social science research

At the start of our meeting today I mentioned an interesting article about the difficulty of doing good social science research. It’s relevant to next week’s topic.

This article, written by Scott Alexander, currently my favorite  blogger/writer, reviews some of the social science research literature addressing this question:

Does the criminal justice system treat African-Americans fairly?

There’s plenty of evidence to support the argument that there’s unequal treatment. There’s also a lot of well-documented research that shows that when you adjust for confounding variables that the differences–or many of them–go away.

This response to the original article, and this follow-up  author of the first-reference post show that when you take confounding variables into account you may remove the data that you are trying to study.

So what’s a citizen to do? We can’t all do the kind of research that Scott Alexander does, and we’re all subject to the loud shouting coming from the liberal or conservative echo chambers of our choice–the ones that select the studies we’re likely to read–the ones carried out by the researchers who start with a point of view and then organize the data so that shows what they believed in the first place.

Confirmation bias anyone?

Leave a Reply