Causes, Interactions

The context is the causes of unemployment. Tyler Cowen:
The general form of the argument is: “only x changed, therefore x is the cause.” A supply and demand graph, with the shift of one curve, shows that argument to be false. The net effect of the shift will depend, for one thing, on the slope of the other curve, plus whether the other curve has been shifting (more slowly) all along.
He's right, of course. Supply and demand, person and situation, genes and environment: the real world is all interactions. Yet the regression models you see in social science publications are almost all main effects. Ponder.

No comments: