Angrist and Pischke

A nice review of their Mostly Harmless Econometrics book by someone I don't know.

The review is generally on target, though I grimaced at the use of experiment to refer to something other than deliberate random assignment.

I would add that the book, like all books, reflects the interests of its authors. For example, the section on longitudinal methods is not as rich as the sections on regression (including matching / weighting) or cross-sectional instrumental variables. This means that the book is a good complement to other more general books like Cameron and Trivedi's text.

Via marginal revolution.