Does Ability Grouping Harm Students?
Over the past few weeks, several publications have written about a new University of Sussex study purporting to show that ability grouping in math harms students.
This is news to those of us who follow education research. One of the biggest analyses of ability grouping to date (from James Kulik of the University of Michigan, surveying 23 major studies on grouping) found that when high-ability students receive accelerated classes, they advance as much as a whole year more than students of similar age and intelligence left in regular classrooms. Kulik's analysis found that specific subject grouping also helps slower students; low-achieving fourth graders put in a very focused group gained as much as two-thirds of an academic year over control subjects.
But the University of Sussex research, done by Prof. Jo Boaler and associated grad students, claims otherwise. The team followed 700 students over five years at three high schools. The press release states boldly that grouping kids by ability harms education. You can read an article about the research here. You can also read the report directly here.
Or you can read my take on it, rather than wading through the whole thing. I am always amazed, reading educational research, at how much the concept of "equity" excites some professors. Boaler studies math, and undertook her study to show that math classes could be used to advance the cause of democracy and caring for the least among us no differently than, say, civics class. She compared group-oriented, mixed-ability classes at a school poetically referred to as "Railside" (since it was near the railroad tracks) with two more suburban schools. The suburban schools used ability grouping. Though the students at Railside entered the school behind the suburban students, at the end of their four years, 41% of seniors were in advanced calculus or pre-calculus, v. 27% at the other schools.
That seems like a pretty clear win. But reading deeper, it becomes clear that the suburban schools were employing what I call the "boring" method of teaching mathematics. Teachers would lecture for 21% of the class; students would then work alone on problems in their textbooks for about half the class. Since this is the same thing they'd do for homework, I would not be surprised if kids spent whole periods watching the clock. When problems were discussed, teachers only spent about 2 minutes on each of them. There was very little opportunity for open-ended problem-solving.
At "Railside," on the other hand, when the teachers did discuss problems, they'd spend 5.7 minutes on a problem, and Boaler describes the process as far more interactive. The kids did a lot of group work -- approx. 70% of the class. There is nothing wrong with learning in groups. Math problem solving can often benefit from a shared approach. Of course, Boaler, being so into "equity," finds this very satisfying for a different reason. Students receive group grades! And best of all, they feel "responsible" for unmotivated students in their groups, rather than feeling that they are a "burden." The political implications are clear.
But I digress. What struck me, reading this study, is how well-trained the teachers at Railside must have been to make this set-up work. Since Boaler talks about "teachers" I am assuming that multiple ones employed this same group-work strategy. That means Railside had fairly strong quality controls in its math department. Teachers settled on a strategy and were committed to its implementation. They asked open-ended questions, and explained topics in depth. They roamed between groups during the problem solving sessions, meaning they were energetic enough to stay on-task for the entire class period. The beauty of the "traditional" approach (as Boaler refers to it) is that it allows the math teacher to read a copy of US Weekly for the 48% of the time the students are doing problems alone in their workbooks.
It begs the question. Given that Kulik's meta-analysis of major educational studies found that, in general, ability grouping helps both slower and faster learners learn more, isn't it possible that the better results achieved at Railside are the result of energetic, committed teachers, rather than the lack of ability grouping? Great teachers are highly correlated with great results.
Unfortunately, they're also rare. As I've written about ability grouping before on this blog, the benefit to that approach is that it fails better. Given a choice of a mixed ability class with a great teacher, and an ability grouped class with a lousy one, I'd take the former. But that is usually not the choice. This study compares intense, analytical, teacher-directed group learning with the "boring" method of teaching math. The former also happens to be heterogeneous in this case; the latter homogeneous. But when there are two big differences between things you're comparing, it's hard to know which one causes the results.
Two other quick critiques. First, while I think the teaching is more responsible for the gains seen at Railside than the ability grouping, the general type of ability grouping employed in American schools isn't finely tuned enough to challenge bright kids anyway. Creating three reading groups in a standard neighborhood school is slightly helpful. But it's far more helpful to create a gifted magnet program that draws students from, say, five counties. Then you can actually create classes for the top .1%. It wouldn't surprise me to find that most students in the 85-115 IQ range can do OK in a math class together. But if you try to throw, say, a Nate Bottman (2007 Davidson Fellow, topic: Analytically Determining the Spectra of Solutions of the NLS) in to a group-learning, mixed ability class, you will quickly find a problem with the approach. At Railside, "If they found that one student was standing out, by, for example, being faster in their mathematical thinking, they would find aspects of the work that the student was less good at and for which they needed more practice," Boaler writes. But a child who's learned calculus on her own at night is not going to need to practice more on any aspect of algebra.
And second, I assume the University of Sussex, and Stanford, where Boaler has also taught, aren't letting just anyone in these days. Their graduate programs and professorial hires are very much grouped by ability. This research was a group project between Boaler and her grad students. But this was hardly a true "mixed ability group." If Boaler really thinks ability grouping is harmful and inequitable, she should have pulled three people off the streets and asked them to work with her instead. I'm sure the results would have been just as good, and Boaler wouldn't have had to carry any extra weight in the computations at all...