A Thousand Words

The graph above shows a group of schools in Denver, Colorado, courtesy of the State Department of Education. The numbers have some interesting implications for how to think about poverty, achievement, growth, and accountability in K-12 education.

Each of the circles represents an individual school. The size of the circle is a function of enrollment--the bigger the circle, the more students enrolled. The color is tied to poverty, with blue schools having less than 20 percent of students eligible for free or reduced price lunch, red schools more than 80 percent, and green, yellow and orange in between.  

The Y-axis represents a snapshot of achievement, based on the percent of students who scored as proficient or advanced on the state test. Schools closer to the top scored the highest. The X-axis represents yearly growth in achievement for the students in the school, in percentile terms, compared to other students who started the year in the same place academically. Schools farther to the right on the X-axis saw more achievement growth than most of their peers. 

What does the graph suggest? Several things. First, a clear relationship between poverty and overall achievement. All of the lowest-poverty schools (which presumably have a disproportionate number of high-income students, although this isn't always the case) have above-average achievement, and most have above the median growth. The high-poverty schools, by contrast, are closer to the bottom.  This shouldn't surprise anyone; most such education graphs look like this in some way, which leads many people to see poverty as determinative. 

But look closer. What this really suggests is the income is indeed pretty determinative--but only at the top of the income distribution. In other words, if you put a bunch of well-off students together in a school with few or no poor students, it's extremely likely that they'll do well. They have a lot of advantages outside of school, and their schools tend to be better-funded and staffed. One would be hard-pressed to find the inverse of the classic "beating the odds" school--someplace where, despite high income and low poverty, achievement is terrible. They just don't exist. 

Among high poverty schools, by contrast, there's a lot more variation. In terms of absolute performance, some are at the bottom, many are in the middle, and a few are closer to the top. And in terms of variation in growth, they're all over the map, from below the 20th percentile to above the 90th. 

In other words, it'd be tough to blunt or reverse the impact of socioeconomic status on educational achievement at the top of the income spectrum. But that's okay--why would you want to? On the other hand, there's a lot of potential to help high-poverty schools, particularly if we look at overall achievement and growth simultaneously. 

Graphs like this also point to the need to react to achievement data in a more sophisticated way. Under NCLB, all of the schools near the bottom of the distribution get treated the same, regardless of growth. That doesn't make much sense. Instead what we need to do is take the schools on the bottom left, the low-achieving / low-growth schools, and accelerate the process by which they either improve or are shut down. There's no point in waiting six years to close a school where achievement is low and students are learning less during the year than most other students who also started the year with low achievement. 

At the same time, we should be pouring resources into the high-poverty, high-growth schools, so we can support success and build on it.  Labeling is important, too: NCLB doesn't actually label schools as "failing" despite what you read in the newspaper, but it ought to for low-status/low-growth institutions. On the other hand, high-growth schools should be complimented and described in very different ways.