If you thought the class size debate should have been over and done with years ago, researchers Thomas Dee and Martin West should give you pause. Their new study looks at the non-cognitive effects of smaller class sizes--things like engagement, motivation, self discipline, homework completion--and whether or not these qualities help to improve educational outcomes later in life (e.g., high school graduation) and success in the labor market.

Reanalyzing data from the well-known Tennessee STAR study of class size as well as from a nationally representative survey of older students, the National Educational Longitudinal Study of 1998 (NELS), Dee and West find that smaller class sizes do make a difference on non-cognitive behaviors. The trick is when to reduce class size in order to see the most gains. They argue that while any effects of class size in the early grades (1-3) quickly dissipate, class size reductions in the later years may be much more effective in producing returns, particularly when targeted at urban schools.

Though the authors didn't use specific definitions of small and large classes--making any sort of policy takeaway a bit fuzzy--they define a class size reduction of about 6 students as significant. The average class size in the sample was 24.5 students and the effects of smaller class sizes were seen in classes with 18 students. So advice to state legislators who love to drop class sizes by a student or two: either do it right or don't do it at all.