With the 鈥淣o Child Left Behind鈥� Act of 2001 emphasizing rigorous research, calls for more randomized trials in education studies have gained a new momentum in Washington policy circles.
Long before 鈥渞igorous research鈥� and 鈥渞andomized trials鈥� became buzzwords in the field, however, the federal Department of Education quietly began setting aside more and more of its research dollars for experimental studies. According to the department, at least 16 such studies are in the works, nearing completion, or about to be funded over the next two fiscal years. That鈥檚 a big increase over previous years, the department鈥檚 research chief says.
鈥淚t鈥檚 very clear in No Child Left Behind that questions of what works in education will have high priority,鈥� said Grover J. 鈥淩uss鈥� Whitehurst, the director of the new Institute of Education Sciences, which oversees much of the research supported by the department. 鈥淨uestions of what works link naturally to randomized trials.鈥�
Common in research in medicine, pharmacology, and welfare reform, such studies entail randomly assigning subjects to either experimental or comparison groups. Such experiments are rarer and somewhat controversial in education.
For instance, of the 1,200 articles on mathematics and science education that were published between 1964 and 1998 in American Educational Research Journal, only 35 involved randomized trials, according to one study.
The apparent reluctance to use the methodology in education stems in part from concerns that such experiments can be expensive, unwieldy, and, in some cases, unethical. In addition, some experts contend, such studies often offer little help in understanding why an intervention works or doesn鈥檛 work.
鈥淕ood evidence is only as good as the theory that interprets it,鈥� said E.D. Hirsch Jr., the University of Virginia professor who created the Core Knowledge school improvement program. Even though a well-respected, randomized study showed that reducing class sizes improved student achievement in Tennessee, he noted, efforts to do the same in California were not as successful.
鈥淪o what good did randomization do in this case?鈥� Mr. Hirsch said.
Shrinking Pot?
The fear, other experts say, is that the Education Department鈥檚 new emphasis on randomized experimentation could shrink the pot of money available for studies using matched comparison groups, for descriptive studies, and for basic research.
鈥淚 don鈥檛 think that is likely,鈥� said Mr. Whitehurst, who was the department鈥檚 assistant secretary for research and improvement before the creation of the new institute this fall. 鈥淏y and large, these randomized trials will be funded out of evaluation and national- activities money, and that鈥檚 money that鈥檚 not previously been available to the research community.鈥�
What鈥檚 more, he said, 鈥渨e expect the pie to get larger.鈥� He pointed out, for example, that President Bush鈥檚 Education Department budget request for fiscal 2003 called for increasing funds for core educational research by 44 percent, or roughly $46 million.
The experimental studies that the department expects to release soon, most of which were begun under President Clinton鈥檚 administration, gauge the effectiveness of the 21st Century Community Learning Centers after-school program, the Even Start family- literacy program, and Upward Bound, which prepares students to attend college.
Two of the ongoing, federally financed experiments include a study of the Success for All schoolwide improvement program, which is expected to cost $12 million over five years in public and private money, and a new, five-year, $30 million study examining the effectiveness of six different types of preschool programs.
In addition, the department is planning to spend $47 million over the next two fiscal years for randomized studies in 11 areas. The subjects are: early reading instruction, preschool literacy instruction, after-school programs, family literacy, alternative certification of teachers, professional development, educational technology, English-language learning, vocational education, charter schools, and adult literacy.
Mr. Whitehurst said department staff members analyzed the percentage of federally supported studies examining cause-and-effect kinds of questions that were addressed through experimental鈥攖hough not necessarily randomized鈥攎ethodology over the past two years. He said that percentage increased from 32 percent in fiscal 2001 to 100 percent in fiscal 2002.
Still, Mr. Whitehurst said, the department has no intention of counting out other kinds of studies.
Experts say randomized experiments are important because they are the soundest way available to find out what works. A not-yet-published study by three researchers at Mathematica Policy Research Inc., a Washington-based group, suggests, in fact, that investigators may get answers in randomized experiments that are different from those they get when they test the same questions using quasi-experimental methods, such as demographically matched comparison groups.
Some experts trace the movement at the federal level to conduct more randomized education studies to the 1998 passage of the Reading Excellence Act. That was the first call in federal education law for 鈥渟cientifically based research.鈥�
Clear Preference
The same terms and their definitions have since appeared in two other important federal education laws: the Education Sciences Reform Act, which revamped the Education Department鈥檚 research arm, and the No Child Left Behind Act. The latter, which is a reauthorization of the Elementary and Secondary Education Act, uses the term more than 100 times. (鈥淟aw Mandates Scientific Base for Research,鈥� Jan. 30, 2002.)
While the definition does not exclude other types of education studies, it expresses a clear preference for randomized methodology.
鈥淕etting the definition into the law became a catalyst for a much broader movement that was about to burst on the scene,鈥� said Bob Sweet, who, as a staff member of the House Education and the Workforce Committee, helped draft that definition.
In some sense, the movement parallels one that took place in welfare reform in the 1970s and 1980s, according to Judith D. Gueron, the president of the Manpower Demonstration Research Corp., a New York City-based research organization. Since then, dozens of experiments have been conducted, leading to a major transformation in state and federal welfare programs.
鈥淭here were questions about the morality of doing this, questions about feasibility, and those questions exist in education,鈥� Ms. Gueron said. 鈥淭hey have been overcome, but I think it is harder in schools. It鈥檚 going to be a harder approach that should be used carefully.鈥�
The debate rages on in education. The National Research Council, in response to emerging concerns over what constitutes scientific research in education, published its own report on the subject last year.
This fall, in an effort to further that discussion among education researchers, the council convened a second group of prominent researchers. Among the topics on their agenda: randomized field trials and when it鈥檚 appropriate to use them.
鈥淲hat we have now is an opportunity鈥攁nd I鈥檓 hoping the education research community sees it as an opportunity鈥攖o respond to the call for better research and to think a little more rigorously about the relationship between the method and the question, and not to reject the idea of randomized trials,鈥� said Michael J. Feuer, the executive director of the National Research Council鈥檚 division of behavioral and social sciences and education.
鈥淚n a way,鈥� he said, 鈥渋t鈥檚 a rather healthy development to see the policy community expressing a real strong interest in more and better education research.鈥�
