Diversity and Student Achievement

Given that Democratic policymakers are cohering around school and community integration as a central policy agenda for helping the poor, it’s no surprise that John King, the acting U.S. Secretary of Education, tweeted the following yesterday:

It very well may be that increasing socioeconomic (and racial) diversity in schools is good for kids and for the country.  I did most of my teaching at an unusually integrated school, economically and racially, and I like to think we gave kids of all sorts a good experience.  But the evidence that diversity is specifically beneficial for student achievement is not at all unambiguous.

• We’ve discussed KIPP- the charter school model that perhaps has the strongest empirical evidence behind it, which is not at all integrated: KIPP’s students are almost all black or Hispanic and mostly low-income.
• We’ve also looked at North Carolina HBCUs— also not integrated, but apparently effective at encouraging college persistence and success among poorly prepared students.
• KIPP is almost certainly unrepresentative of patterns across regular K-12 public schools: in particular, these are settings that require a degree of buy-in and choice from parents and children and even more importantly can throw out obstreperous students.
• Nationally representative data, however, suggests that while black students in schools with very few black students have slightly higher performance than black students in majority black schools,  the gap between black and white students in the same schools does not narrow at all. (See Figure 7 here)
•  Chetty and his colleagues pick up large benefits (in terms of later-life earnings) to lower-income children who grow up in certain communities and not others. But as has often been noted, the place-based effects in these studies are largely confounded with race, which Chetty cannot directly observe in his data set.

The gold standard for the effects of socioeconomic integration on child outcomes is still the Moving to Opportunity study, a low-attrition RCT which used Section 8 housing vouchers to induce low-income families to move to more socioeconomically integrated areas:

The U.S. Department of Housing and Urban Development’s (HUD) Moving to Opportunity (MTO) for Fair Housing program is a unique experimental research demonstration designed to answer the question of whether moving from a high-poverty neighborhood to a lower-poverty community improves the social and economic prospects of low-income families. Authorized by the U.S. Congress in 1992, MTO made use of rental assistance vouchers, in combination with intensive housing search and counseling services, to assist low-income families to move from some of America’s most distressed urban neighborhoods to lower-poverty communities. A total of 4,600 low-income families with children, the vast majority of them headed by African- American or Hispanic single mothers, were recruited from high-poverty public housing projects in five participating cities between 1994 and 1998. These families were assigned by lottery to one of three research groups: A Traditional Voucher group, a Low Poverty Voucher group and a control group. Because of the random assignment design, the MTO study generates comparable groups of adults and children living in different types of neighborhoods, so that a comparison of outcomes across research groups can uncover the potential effects of neighborhood characteristics across a range of family and children’s outcomes. Among the households assigned to the Low Poverty Voucher group, 47 percent used a MTO voucher to relocate to a low-poverty neighborhood, while 62 percent of those assigned to the Traditional Voucher group relocated through MTO.
A follow-up study carried out 4 to 7 years after random assignment found that:

• MTO improved neighborhood outcomes. Assignment to either of the MTO mobility groups led participating adults to feel safer and more satisfied with their housing and neighborhoods.
• MTO had no effect on the labor market outcomes or social program participation of adults, but improved adults’ mental health as well as several important aspects of physical health.
• MTO improved outcomes for female youth, particularly their mental health, but on balance had deleterious effects on male youth risky behavior.
• MTO had no detectable effects on the math and reading achievement of children.

So we have one rigorous trial of the effects of deliberate school and community integration on low-income families, and true to the Iron Law, it found null effects on student achievement and mixed positive and negative effects on other child outcomes.

Richard Kahlenberg of the Century Foundation has long advocated for socioeconomic integration as a solution to low achievement by low-income students.  “Racial integration is very important,” he writes in one recent essay, “but if one’s goal is boosting student achievement, what really matters is economic integration.”

Kahlenberg is well aware of the results from KIPP and Moving to Opportunity. He argues that neither of these examples generalizes, and instead policymakers should focus on successful demonstrations of socioeconomic integration in La Crosse, Wisconsin, in Finland, and in Montgomery County, Maryland.  I don’t have a coherent argument for or against the policies he advocates, so here are ten scattershot points:

1. Kahlenberg is right that huge social forces have pushed Americans into more and more socioeconomically segregated schools and communities. Regardless of its effects on school achievement, this seems like a Bad Thing. The goal of social and cultural integration- the creation of a democratic polity capable of deliberation- was central to the creation of public schools.
2. The potential for reversing these forces seem more limited to me than to Kahlenberg. Montgomery County is one of the richest and most educated counties in the country. Most districts don’t have the same kind of headroom, financially and cognitively. The most economically integrated areas of the country are much more likely to be like Irving, Texas, or Ferguson, Missouri- stumbling through rapidly changing demographics rather than smoothly managing them.
3. The majority of school districts already have a large number of low-income students: American kids are poorer than the country as a whole, due to differential fertility. So while there are a smallish number of upper income redoubts resisting integration, in most cases the policy counterfactual is more like Moving to Opportunity than Moving to Montgomery.
4. It is very likely that middle-class parents privately overestimate the costs of economic and racial integration to their own kids. That doesn’t mean they are costless. See Hanushek’s paper on racial peer effects, and the paper on Katrina victims in the Houston schools. If you are a middle-class Black or Hispanic parent, in particular, you want the richest school you can find.
5. Given that negative peer effects so often work through the mechanism of disruptive misbehavior (see the Katrina paper), it is potentially self-defeating to mix a push for school integration with increasing federal oversight and regulation of school discipline. In particular, the federal government has often hailed in theory progressive and milder forms of discipline that prove ineffective in practice.
6.  Everything will work better for elementary schools than middle or high schools, perhaps because the gaps are smaller there.
7. In high schools, the more diversity you have, the more pressure for tracking you’ll experience. So more in-school segregation, less between-school segregation. Better or worse? Who knows.
8. Closing gaps in graduation rates is more achievable than closing gaps in test scores. That’s fine-in fact I think it makes sense to focus on graduation rates, college enrollment, and employment rates rather in policy, rather than juicing test scores. But Kahlenberg makes it seem like this is a conservative estimate of effects, when graduation rates often are affected by interventions even when test scores are not.
9. A small number of (high income) states have court-ordered integration of low-income housing by municipality. I believe that recent Supreme Court decisions are likely to expand this number.
10. My gut says that we should focus on task-based and service-based models of racial and economic integration rather than (solely) academic models (City Year being the one I know best.) But these are themselves very intensive and expensive.