Class-Based Integration

Emily Bazelon wrote a New York Times article recommending class-based school integration (much of her article resembles previous work by Rick Kahlenberg). She suggests that class-based integration would improve test scores:

Test scores may not be the best way to assess the quality of a teacher or a school, but the pressure to improve scores, whatever its shortcomings, is itself on the rise. And if high test scores are the goal, it turns out, class-based integration may be the more effective tool.

Researchers have been demonstrating this result since 1966, when Congress asked James S. Coleman, a Johns Hopkins sociologist, to deliver a report on why the achievement of black students lagged far behind that of white ones. The expected answer was that more than a decade after Brown, black kids were still often going to inferior schools with small budgets. But Coleman found that the varying amount of money spent on schools didn’t account for the achievement gap. Instead, the greater poverty of black families did. When high concentrations of poor kids went to school together, Coleman reported, all the students at the school tended to learn less.

How much less was later quantified. The Harvard sociologist Christopher Jencks reanalyzed Coleman’s data in the 1970s and concluded that poor black sixth-graders in majority middle-class schools were 20 months ahead of poor black sixth-graders in majority low-income schools. The statistics for poor white students were similar. In the last 40 years, Coleman’s findings, known informally as the Coleman Report, have been confirmed again and again. Most recently, in a 2006 study, Douglas Harris, an economist at the University of Wisconsin, found that when more than half the students were low-income, only 1.1 percent of schools consistently performed at a “high” level (defined as two years of scores in the top third of the U.S. Department of Education’s national achievement database in two grades and in two subjects: English and math). By contrast, 24.2 percent of schools that are majority middle-class met Harris’s standard.

A few points: 1) The problem with using Jencks' analysis here is that it doesn't account for selection effects -- poor black sixth-graders in the 1970s who managed to attend majority middle-class schools may have had very different family backgrounds, attitudes, and motivations than the ones stuck in impoverished schools.

2) Citing Douglas Harris's report is even less convincing. As Bazelon's text itself suggests, Harris merely compared the test scores in high-poverty schools to those in low-poverty schools. This is just a static picture; it tells us nothing about what would happen if you took an impoverished child and set him down in a low-poverty school. Thus, Harris's finding doesn't "confirm" the Coleman Report's finding as to how poor children did when they attended middle-class schools (Harris doesn't deal with that point at all).

3) A more enlightening and relevant study -- one that Bazelon doesn't mention -- comes from the Moving to Opportunity experiment, which took families in public housing and randomly assigned them to three groups: a control group, a group that got regular housing vouchers, and the experimental group that got housing vouchers to be used only in neighborhoods with a poverty rate under 10%. Researchers who expected to find that the experimental group did better academically were disappointed:

Lisa Sanbonmatsu, Jeffrey R. Kling, Greg J. Duncan, and Jeanne Brooks-Gunn, Neighborhoods and Academic Achievement: Results from the Moving to Opportunity Experiment.

Although we had hypothesized that reading and math test scores would be higher among children in families offered [housing] vouchers (with larger effects among younger children), the results show no significant effects on test scores for any age group among over 5000 children ages six to 20 in 2002 who were assessed four to seven years after randomization. Program impacts on school environments were considerably smaller than impacts on neighborhoods, suggesting that achievement-related benefits from improved neighborhood environments alone are small.

To be sure, that last sentence is important: many of the people given housing vouchers to use in low-poverty neighborhoods ended up with their children attending new schools that were not a whole lot better than before; some even kept their children at the same schools (as this followup article reveals). So while the MTO experiment doesn't fully predict what would happen if you took the most impoverished inner-city kids and somehow transplanted them into the highest-performing suburban schools, it does give some idea of how hard it would be to pull off that kind of school reassignment.


UPDATE: I dug up my copy of the book "On Equality of Educational Opportunity," edited by Frederick Mosteller and Daniel P. Moynihan, in which (as I recalled) Christopher Jencks had published his reanalysis of the Coleman Report data. Sure enough, after noting that poor black sixth graders in middle-class schools did well, Jencks adds, "poor families who send their children to middle-class schools may also be more achievement-oriented and more competent than poor families who send their children to lower-class schools. The EEOS [Equality of Educational Opportunity Survey, or the Coleman Report] does not provide adequate data for testing this theory."

Jencks makes another interesting point, which I take to imply that the most educational improvement could be gained from improving schools and curriculum, not from merely rearranging students within existing schools:

P. 86:

Contrary to popular belief, the students who performed best on these tests were often enrolled in the same schools as the students who performed worst. . . . In some ways this is the most important and most neglected single finding of the EEOS. It means that if our objective is to equalize the outcomes of schooling, efforts to reduce differences between schools cannot possibly take us very far. If by some magic we were able to make the mean achievement of every Northern urban elementary school the same, we would only have reduced the variance in test scores by 16-22 percent. If, on the other hand, we left the disparities between schools untouched but were somehow able to eliminate all disparities within schools, we would eliminate 78-84 percent of the variation in 6th-grade competence.

The implications of this are in many ways more revolutionary than anything else in the EEOS. In the short run it remains true that our most pressing political problem is the achievement gap between Harlem and Scarsdale. But in the long run it seems that our primary problem is not the disparity between Harlem and Scarsdale but the disparity between the top and the bottom of the class in both Harlem and Scarsdale.