Tyler Cowen has an interesting comment on this [0]. Generally, it's good policy to interpret these studies cautiously.

[0] http://marginalrevolution.com/marginalrevolution/2012/02/how...