Douglas B. Downey

In a nutshell

This book is about understanding how schools influence achievement gaps. The traditional story is that schools are part of the problem and that large differences in school quality make achievement gaps worse. This view is so widely accepted that it is now asserted without evidence. But if we do look at the evidence, we find some surprising patterns.

First, socioeconomic gaps in math and reading skills are already large at kindergarten entry, and then do not appear to increase during the school years. This pattern is better known now that we have high-quality data that begin at kindergarten, follow children for many years, and use scales of math and reading skills that allow an understanding of how gaps change over time.

Second, schools serving mostly advantaged children (high-socioeconomic status and white) do not produce any more math and reading learning than schools serving mostly disadvantaged children (low-socioeconomic status and black). Of course, schools serving advantaged children tend to have higher test scores than schools serving disadvantaged children, but careful analyses indicates that these differences are due to the skills the students develop outside of school. Once in school, both advantaged and disadvantaged children learn at roughly the same rate. This pattern replicates across many datasets but, surprisingly, is hardly talked about in the policy world.

Combined, these two patterns suggest that, when it comes to achievement gaps, schools are more part of the solution than the problem. Rather than focus our attention on schools, these patterns prompt us to consider the way that highly unequal early childhood conditions generate and maintain achievement gaps.

There are many policymakers interested in reducing achievement gaps, but most of their solutions focus around school reform. Those attempts are not without merit, but they are unlikely to do much to reduce societal-level achievement gaps, because the gaps are mostly developed when children are not in school. If we really want to reduce these achievement gaps we will need to think bigger than school reform. We will need to consider reforms that reduce the widely unequal family and neighborhood conditions children experience when not in school.

The wide angle

I began exploring the relationship between schools and inequality after reading the 1992 article entitled “Summer Setback” by Doris Entwisle and Karl Alexander. The authors showed that gaps in math skills between advantaged and disadvantaged children in Baltimore grew faster when school was out in summer than when school was in during the rest of the year. The implication was shocking: schools were reducing the gaps we would likely observe if children were not there. Schools were compensatory.

At first, I thought there was something odd with the Baltimore study. But when my colleagues and I replicated this pattern with the nationally representative Early Childhood Longitudinal Study—Kindergarten Cohort of 1998 (ECLS-K:1998), I was forced to rethink my earlier assumption that schools were generating significant inequality in math and reading skills.

The ECLS-K:1998 data were collected by the Department of Education and are among the highest quality education data available. The study consists of over 20,000 children attending kindergarten in roughly 1,200 schools in the fall of 1998-99. One unusual feature of the data is that children were tested in both fall and spring of kindergarten and first grade, allowing researchers to observe how fast children learned both when school was in and when it was out (summer).

My team of researchers at Ohio State analyze a restricted-use version of the ECLS-K:1998 data, which allows us to link children’s scores on reading and math tests to their personal information, like their socioeconomic status. To protect the anonymity of the respondents we are required to analyze the data on a standalone computer, lacking any connection to the internet. On random occasions, officials from the Department of Education arrive at my office and inspect our data protection protocols to ensure that we are handling the data appropriately. This creates extra work for us, but it is important so that parents can feel comfortable that their answers to the questionnaires will be kept confidential.

A close-up

If you believe that widely unequal schools are responsible for achievement gaps, open my book to page 17 and look at Figure 2.1. That figure shows that high-SES children are ahead of low-SES children in reading skills by .64 standard deviations at kindergarten entry. At this age, we are really measuring mostly pre-reading skills, like knowing letters and the sounds they make, and recognizing a few simple words and short sentences. The standard deviation is a useful way to gauge the gap between two groups because it allows us to assess the magnitude of that gap, even when the kinds of reading skills assessed are different at later ages. If schools increase inequality, we should observe a growing gap (as measured by standard deviations) over time. Over the next nine years, however, this gap does not increase in the way it should if schools are the culprit. Instead, it declines to .55 standard deviations by the end of the eighth grade. The math results, Figure 2.3 on p. 19 in the book, show the same pattern.

If these patterns were unique to the ECLS-K:1998 data, we might be able to dismiss them, but the overall pattern replicates in a wide range of nationally representative datasets. It is really hard to develop a story about schools that is both consistent with this pattern and defines them as a source of achievement gaps. Instead, the patterns suggest that children arrive at kindergarten on highly unequal learning trajectories. Once in school, however, achievement gaps largely stop growing and even begin to narrow somewhat—the very opposite of what we would observe if schools were a source of achievement gaps.

Lastly

My hope is that this book prompts a more careful consideration of schools’ role in shaping achievement gaps. The assumption that schools are largely responsible is misplaced and can divert attention from larger social problems that likely are the source.

Consider how we stack up against Canada. Our 15-year-olds score .30 standard deviation units behind Canada’s on international reading tests. Most would blame our schools for this gap but it turns out that the same cohort of children were .31 standard deviation units behind Canadians at age 4-5, before schools had a chance to matter. This pattern highlights how school reform is not the likely explanation for why our teenagers’ skills are behind those in other countries. The problem is rooted in early childhood conditions where too many of our children experience stressful environments. Notably, high-performing American five-year-olds did about as well as high-performing Canadian children. The large gap across the two countries is due primarily to very poor-performing children. The U.S. has more very poor-performing students than does Canada.

So, are Canadian children just genetically superior to American children? That doesn’t seem likely (unless you’re Canadian). A more plausible explanation is that Canadian children, on average, experience better early childhood conditions. And that is likely due to a broad range of policy decisions Canadians have made differently than Americans, such as the provision of universal health care, which reduces stress for the disadvantaged. Of course, the battle in the U.S. over policies like universal health care is a more difficult battle than school reform. But just because it is easier to focus on schools does not mean that they are the appropriate policy lever. Reducing inequality requires us to think bigger than school reform.