Schooling often determines future earning potential and economic mobility. Yet educational choices are often fraught with confusion and uncertainty for students and parents, and for the policy makers who must make difficult decisions about controversial policies and scarce education resources. Are charter schools always a beneficial choice? Do vouchers provide more opportunity? What about expensive scholarships—do they really boost outcomes for students in need?
MIT’s School Effectiveness & Inequality Initiative (SEII), based within the Department of Economics, uses rigorous experimental and quasi-experimental methods to examine how education and economics are linked. SEII is helmed by faculty co-directors Joshua Angrist, David Autor, and Parag Pathak and is backed by more than 25 affiliated faculty, postdocs, and graduate students. SEII is the first research group to link the economic theory of market design—that is, the study of centralized school assignment schemes such as are now used in many American school districts—with questions of education policy. Theirs isn’t research in a vacuum: SEII’s studies have led to major educational reform and policy change since launching in 2011.
For instance, SEII was the first group in the country to produce credible, randomized, lottery-based evaluations of charter schools. Researchers found that some Boston charter schools outperform the city’s public schools and that charter schools boost performance of special-needs students and are linked to better MCAS results. These findings were central to the debate over an expanded cap on Massachusetts charter schools. SEII has conducted groundbreaking research on school lotteries, working closely with the former Menino administration in Boston to analyze zone-based and non-zone-based models.
“We don’t lobby or seek out politicians; we don’t try to influence policy directly. But people who make decisions read our work. We’re pro-research, in the rigorous MIT research tradition of putting science first,” says Angrist, who is Ford Professor of Economics.
This is important, because clear-eyed educational evaluation has long been overlooked, SEII’s founders say. According to Pathak, educational policy debates too often involve “people arguing on ideology instead of facts.”
“Other areas—health, energy, et cetera—have pretty well-established traditions of evaluation. In education, we often see people with the best of intentions trying out policies,” Pathak observes. “We’re trying to build a narrative around compelling case studies, using rigorous experimental methods—methods that allow you to have full confidence in what you’re studying, as you would in a medical trial.”
Take the question of scholarships in higher education. Much of American education policy is built around an expensive effort to subsidize college attendance for students deemed meritorious, in need of aid, or both. But do scholarships make it more likely that recipients will attend and complete college? Or would these motivated students do well, even without the money? To answer such questions, SEII has worked closely with the Susan Thompson Buffett Foundation (STBF), Nebraska’s largest financial aid provider, which gives away thousands of scholarships each year. In their pioneering randomized evaluation of the STBF aid program, the researchers discovered that STBF’s awards induced a clear shift from two- to four-year schools, as well as a marked increase in second-year college enrollment on the part of recipients, indicating that scholarships improve retention.
“We knew almost nothing about what actually helps those kids complete college and also whether it pays off for them,” says Autor. “The United States is the number-one country in the world at sending kids to college. But we’re fifteenth when it comes to those who actually earn the degree. We’re good at sending them to college, and not good at getting them through. Studies like the STBF research analyze not only how to get students into college but what keeps them there. This is central to thinking about long-term welfare for the American middle class.”
Pathak, meanwhile, has conducted pioneering research on voucher systems. His team evaluated the Louisiana Scholarship Program (LSP), a well-known voucher plan that provides public funds for disadvantaged students at low-performing state public schools to attend the private schools of their choice. These vouchers are allocated via random lottery at schools with more eligible applicants than available seats. His team estimated causal effects of voucher receipt by comparing lottery winners and losers in the first year after the program expanded statewide.
Vouchers might sound promising. But in a newsworthy twist, SEII’s comparison revealed that LSP participation markedly reduced academic achievement, lowering math scores by several standard deviations and increasing the likelihood of a failing score by 50%. SEII’s team found similar results for science, social studies, and reading, across income groups and geographic areas. Research revealed that the program might attract private schools struggling with enrollment—and thus indicated caution for expanding a voucher system.
“Now we have gold standard evidence of the effect of the voucher program, and you can imagine that the research was controversial,” Pathak says.
Education reform gets plenty of media attention, but it’s easy to be swayed by the next big trend or splashy idea. “The typical news story is this: A reporter is captivated by something a school is doing, goes there, describes it, says, ‘Isn’t that neat.’ Maybe it’s tech, maybe it’s child-centered learning, maybe it’s built around the arts. For whatever reason, it captures the imagination,” Angrist reflects. “But are these good things to be doing to students, does it benefit them in some measurable way that will have consequences? There are many examples where people have a strong instincts about what’s good, which somehow blinds them to the facts.” By applying quantitative analysis to theoretical ideas, SEII fills the role of fact-checker in a field that crucially needs it.