If Northwestern University researcher Greg J. Duncan and his colleagues hadn鈥檛 used more than one research method to study an anti-poverty experiment in Milwaukee, they would never have known about the gangs and the sneakers.
That鈥檚 because the study鈥檚 quantitative results couldn鈥檛 explain a puzzling disparity that turned up between the boys and the girls they studied. The boys whose families took part in the program did better in school than boys whose families didn鈥檛. Girls fared about the same whether or not their families were enrolled in the experimental program, which provided a temporary safety net of wage supplements, insurance benefits, and child-care subsidies for working-poor families.
An explanation became clear when the field researchers who had been interviewing the families weighed in. The mothers they met had told them that local street gangs were trying to lure their sons by offering expensive sneakers and other items. To pre-empt those temptations, the mothers had used their additional income to buy their sons, but not daughters, those extras.
Mr. Duncan鈥檚 experience makes a good case for a push going on at the national level to shine a spotlight on 鈥渕ixed methods鈥� research in education鈥攊n other words, studies that blend different research strategies. Although various methodologies have always been part of researchers鈥� toolkits, much of the renewed attention to that strategy is a reaction to the U.S. Department of Education鈥檚 current emphasis on using randomized field trials to answer questions about what works in schools and classrooms.
Some experts say such studies, in which subjects are randomly assigned to either experimental or control groups, are the best way to answer questions about what works best. Many education researchers complain, however, that the Bush administration鈥檚 focus leaves out other valuable forms of research.
Yet while it seems common sensical that combining different research strategies could yield more complete answers, researchers also caution that the issue is not clear-cut. With limited funding for education studies and a small supply of researchers trained to use a variety of approaches, mixed methods aren鈥檛 always practical, according to these researchers.
Plus, within the field, philosophical divisions have for years separated researchers who favor qualitative or more descriptive kinds of studies from those who specialize in the quantitative, number-crunching variety.
鈥淭o say that mixed methods are always better would be naive,鈥� said Thomas D. Cook, a professor of sociology, psychology, education, and social policy at Northwestern University. 鈥淚t implies that we haven鈥檛 learned enormously from classical, single-method studies.鈥�
Subject of Debate
The topic of mixed-methods research was the focus of a Dec. 14 conference that was sponsored by four Washington-area national organizations that follow educational research: the National Research Council, an arm of the National Academies, a group chartered by the Congress to advise the federal government on scientific matters; the American Educational Research Association; the American Psychological Association; and the National Science Foundation, an independent federal agency. Sponsors had to draw up a waiting list for the meeting, which attracted a spillover crowd of 238 participants.
鈥淭here is unprecedented interest now in the methodological quality of studies in education,鈥� said Martin E. Orland, the director of the National Research Council鈥檚 center for education, which hosted the event.
He said that interest has grown since the Bush administration began its push for 鈥渟cientifically based research鈥� in education following passage of the No Child Left Behind Act in 2001.
Federal definitions of such studies favor randomized experiments over other study methods.
The problem, some education researchers contend, is that while randomized studies can determine whether an intervention works, they cannot answer key questions about why it works, they can鈥檛 tell whether it works better where it鈥檚 well implemented, and they can鈥檛 pick up on any unexpected side effects.
In comparison, mixed-methods research offers the potential for deeper understandings of some education research questions that policymakers need answered, some scholars said.
鈥淲e would not have come close to understanding the experimental impact had we been forced to rely on quantitative data alone,鈥� Mr. Duncan said of his study of the New Hope anti-poverty program. Researchers in that project surveyed 1,300 participants in the experiment, randomly assigning half to the program. They also studied a subset of 43 families in depth, interviewing them every six weeks for two years.
Breaking Down Barriers
Besides producing better research, mixed methods might also help heal professional rifts between qualitatively oriented researchers and quantitative-study proponents, said Rena F. Subotnik, the director of the American Psychological Association鈥檚 center for psychology in the schools.
鈥淲e have members whom we respect who are on the extreme edges of this qualitative-quantitative debate,鈥� she said. 鈥淭his is taking a more constructive approach.鈥�
Still, the payoff may not always be worth the expense involved in adding descriptive research to experiments, Mr. Cook of Northwestern University said. Known primarily as a quantitative expert, he collaborated with qualitative researchers to study a school improvement program in Chicago. In the end, he said, both the quantitative and qualitative researchers reached nearly identical conclusions on which schools were implementing the program well and which were not.
鈥淲e learned in our particular case that we need not have done that,鈥� Mr. Cook said. 鈥淗owever, we could not have known that before we began the study.鈥� The problem, he added, is that the study of mixed-methods practice is still too underdeveloped to guide researchers on which methods to use for what questions.
One qualitative researcher, Joseph A. Maxwell, an associate education professor at George Mason University in Fairfax, Va., pointed out that philosophical barriers between researchers can also get in the way of collaboration.
鈥淨uantitative researchers think about the world in different ways than qualitative researchers,鈥� he said. He worries that the statistically oriented researchers, rather than viewing their qualitative-study colleagues as equal partners, might just use the quotes and anecdotes the field researchers collect to buttress the numbers.
Training Novice Researchers
Such schisms might be overcome, researchers said, if more education schools exposed budding researchers to a variety of approaches. The American Psychological Association, the American Educational Research Association, and the federal Education Department鈥檚 Institute of Education Sciences have all begun pilot programs in the past five years that aim to help education researchers in training expand their methodological and disciplinary expertise.
鈥淭he challenge is one of having a breadth of understanding and yet to bring one鈥檚 own expertise to the table,鈥� said Felice J. Levine, the executive director of the Washington-based AERA. 鈥淲e do not have to make a jack-of-all-trades out of everyone.鈥�
