Dumping What Works: Austerity for Evidence-Based Programs

I confess there was something delicious in last Friday’s news that Republican legislators in the House were tentatively zeroing out a small federal initiative designed to expand a handful of programs that have been shown to work. The subtext of the article in the New York Times was that there was absolutely no public program that anti-government zealots would trust as effective or consider worthwhile. None.   

 

The initiative in question involves “evidence-based” programs that reflect the highest standards of performance evaluation. It includes a program in which nurses visiting teen-age mothers and other troubled parents have unequivocally produced improved health outcomes. It also includes grant competitions administered by the Department of Education and the Employment and Training Administration that employ the highest standards in evaluating applications.      

 

Originally an initiative of George W. Bush and embraced by the Obama Administration, the program is an outgrowth of a particular response to the anti-government campaign. To the charge that government is intrinsically incapable of constructively engaging with critical issues, some proponents of strong government policies have focused on “evidence based” practice to ratchet up the standards for determining whether programs work. They tend to echo the view that government tends to be wasteful and ineffective.  They then express a strong preference for programs that have met the highest standards of proof, preferably having been fielded with a control group determined by “random assignment” of the intervention (as is normally deployed in drug testing) . 

 

This is a high standard indeed, and is undoubtedly valid in some circumstances. The weakness of the standard is that random assignment usefully reveals the strength of an intervention only when the number of subjects in the experiment is substantial (people but not communities, interventions over time but not one-off events), and the environment of the experiment is stable. One could apply randomized trials to learn what would be the best diets to feed children in schools, but one would be hard pressed to subject to randomized trials efforts to develop urban farms to provide the food.  As Lizbeth Schorr has been arguing, one would hope that policy-makers have the courage to recognize that there are many other ways to learn what works. A more eclectic view on improving government performance is sponsored by the Center for American Progress.

 

Insistence on “evidence based” practice is valid up to a point, but it represents a pinched way to think about the public’s business. President Truman signed the G.I. Bill with no evidence that returning veterans would power the United States to prosperity. President Eisenhower lacked irrefutable proof that the Interstate Highway System would fundamentally change the country and open up unimagined opportunities. 

 

Nonetheless, the approach does respond to policy researchers’ appetites for controlled experiments, and politicians’ hopes to be able to show that they only endorse the highest standards for policy intervention.  

 

Budget support for the proven policies in question may yet be forthcoming. Meanwhile, the threat of defunding them provides a moment to reflect on how truly ideological the anti-government perspective really is.