Orszag: More must be done to evaluate programs
Budget chief says more data systems and protocols for statistical analyses are needed.
The White House budget chief told statistics professionals Friday that more must be done to empirically evaluate the effectiveness of the nation's healthcare and education systems.
"We are making huge investments without doing enough to measure what works and what doesn't," said Office of Management and Budget Director Peter Orszag.
Speaking at a symposium on the use of statistics in federal programs, Orszag highlighted the administration's efforts to boost funding for "effectiveness research" in the $787 billion economic stimulus package.
But Orszag warned that more data systems, or protocols for statistical analyses, must be put in place to help policymakers determine the best ways to educate the nation's children and keep citizens healthy. "In health care, we are doing lots of things without measuring the impact," he said, noting that the costs of the same medical procedures sometimes vary wildly from hospital to hospital. Citing a Dartmouth College study showing that $700 billion is spent annually on healthcare services with no proven outcomes, he said the system is "wasting substantial amounts of resources on procedures that don't work."
Noting the administration's commitment to "evidence-based policy making," Orszag defended the elimination of programs in the president's fiscal 2010 budget details that were released Thursday. The Even Start program -- which supports family literacy projects and was protected from the Bush administration budget ax by its champions in Congress -- has simply performed too poorly on empirical tests to warrant funding, Orszag said. The administration's commitment to data analysis, Orszag later added, informs its commitment to evaluating teacher performance in the classroom rather than rewarding educators only on the basis of "upfront credentials."
He told the statisticians, "There could be a potentially very big payoff, but we're not going to be able to do that kind of analysis if we don't have a data system in place in the first place."