Dallas ISD Is Looking for a Better Way to Measure Student Poverty

On paper, Edwin J. Kiest Elementary in East Dallas' Casa View neighborhood and Paul Laurence Dunbar Elementary in South Dallas/Fair Park seem about the same, poverty-wise. Dallas ISD counts 96.1 percent of students at Kiest as "economically disadvantaged," compared with 99.7 percent at Dunbar. Both are overwhelmingly, distressingly poor.

On the ground, the two schools' circumstances could hardly be more different. In the neighborhood around Kiest, the median household makes about $44,000, roughly four times the income of a Dunbar-area household. According to census data, kids zoned to attend Kiest are three times as likely to come from two-parent households (62 percent) as those zoned for Dunbar (19 percent); their families are twice as likely to own their own homes (73 percent to 34 percent) and five times as likely to have at least a bachelor's degree (16 percent to 2.9 percent). All of these factors — family income, family structure, homeownership, parents' educational attainment — are strongly correlated with student achievement and life outcomes. When it comes to gauging the collective disadvantage of their student body, Kiest and Dunbar might as well be on different continents.

Historically, however, there hasn't been a good statistical way to differentiate between the two. Dallas ISD — like just about every other school district in the country, plus education researchers and the federal government — has used the percentage of students eligible for free-and-reduced lunch as a stand-in for poverty despite some fairly obvious shortcomings, most of which have to do with the figure being designed to decide which kids to charge for lunch, not detailed analysis of educational policy. Free-and-reduced lunch is available to any student whose family makes 185 percent of the federal poverty level or less, which means that the only child of a stay-at-home mom and blue-collar dad making $37,000 a year counts the same as the children of an unemployed single mom living in an extended-stay motel.

For awhile now, researchers and policymakers have been pushing for the adoption of metrics that better capture students' socioeconomic status, which education wonks typically shorten to SES. SES is basically a bunch of non-school factors, like the income/family structure/educational attainment mentioned above, that have a documented impact on educational outcomes. Over the summer, the U.S. Department of Education-funded National Forum on Education Statistics published a white paper detailing some of the shortcomings of free-and-reduced lunch as a proxy for SES (in addition, the exclusion of factors other than income, federal educational privacy rules and the breadth of the statistic make it useless for tailoring services to individual students) and recommending an overhaul of how the students' socioeconomic status is collected and measured.

Doing so will be hard, given how thoroughly free-and-reduced lunch is baked into the modern education system, influencing everything from school funding to state accountability ratings. Free-and-reduced lunch also has the advantages of being uniform and readily available. There's no consensus at the moment on exactly which factors to include in an assessment of SES or how much weight to give them.

School districts aren't necessarily waiting. In Chicago, the school system has been divided up into four "socioeconomic tiers," with the first tier representing neighborhoods in the lowest quartile of the city in terms of income and adult education, and the fourth tier representing those with the highest. DISD has also begun experimenting with tiering, albeit in a more limited way.

When I met last week with Mike Koprowski and Mohammed Choudhury to discuss gentrification — the chief and director, respectively, of DISD's office of transformation and innovation — Koprowski showed me a color-coded map his office created using Census data to carve Dallas into four socioeconomic tiers.
Koprowski's team analyzed four factors for each of the 808 census blocks that feed into DISD: median income, adult educational attainment, homeownership rate and the percentage of single-parent households. They calculated a percentile for each census block for each of the four factors, then added them together to give census blocks a "socioeconomic score." The census blocks were then assigned to Blocks 1-4, with Block 1 representing the least disadvantaged 25 percent of neighborhoods in DISD and 4 representing the most disadvantaged 25 percent. Then they mapped them.

The color distribution is striking, deviating in significant ways from what would appear if the map were carved up simply by income. Preston Hollow is green, of course, but so are considerably less affluent pockets of Northwest Dallas, Oak Cliff and Seagoville. Big pockets of South and West Dallas, on the other hand, are deep red.

"Any educator will tell you that there is a big difference in the challenges faced by a student who is just under the FRL [free-and-reduced price lunch] line and a student who is living in abject poverty," Koprowski explained in an email this week. "There is no doubt that any student eligible for FRL faces challenges, but there are significant differences even within the FRL population that are not always captured by our binary system. The potential answer is to look at the district in more nuanced socioeconomic blocks."

Koprowski describes the map as an "exploratory exercise," but it's not entirely academic. When DISD opens the former Bonham Elementary as an all-girls STEAM Academy next fall, it will set aside half of the seats for lower-income students and half for higher-income students based on the traditional metric of free-and-reduced lunch eligibility. In the spring, once the school has been filled through a lottery, Koprowski's office will conduct an "equity audit" to ensure that at least 12 percent of the school's students come from Tier 3 census blocks and at least 12 percent come from Tier 4 census blocks.  "This way, we can ensure that at least a quarter of students are coming from the most disadvantaged neighborhoods in the city," Koprowski writes.

DISD's experiment is limited, and the U.S. education system doesn't seem likely to stop overusing free-and-reduced lunch eligibility data anytime soon, but it's a step in the right direction. Better, more nuanced data is never a bad thing.