To Fix Education, We Need a Different Lens for Test Scores

If we stop viewing racial and income groups as monolithic, we might learn something about what schools can do to help all kids succeed


The data we collect about a problem can shape how we try to solve it. That’s why we need to look beyond race and income when talking about test scores and “achievement gaps.”

For decades, standardized test scores for different groups of students have been reported primarily in terms of race and family income. The picture we get depicts whites, Asians, and the affluent on the high side of the test-score gap, with Blacks, Hispanics, and the poor on the other. But these broad strokes give us an oversimplified view, leading to reform efforts that miss the root of the problem.

And now, it seems, that picture is getting even more simplified. The standard measure of student poverty has been eligibility for free and reduced meals (FRM) at school. FRM eligibility has never been an accurate measure of poverty, but partly as a result of the pandemic—which prompted authorities to provide free meals to all students—it’s become less reliable than ever.

The expansion of FRM, which two states have already made permanent, is a welcome development. It’s far better to provide free food to some children who don’t need it than to risk letting others go hungry. But because the measure has even less to do with poverty than it used to, some are choosing to focus on other test-score categories when analyzing student achievement.

Take scores for the National Assessment of Educational Progress, or NAEP—reading and math tests periodically administered to a representative sample of American students. At a press event in May, officials announced score gaps between student subgroups in terms of race and ethnicity but not economic disadvantage, citing the unreliability of FRM as a measure of family income. Especially against the background of recent critiques holding that standardized tests are inherently racist, the effect was to elevate race above income as a factor in test-score gaps.

The intersection of race and income

Racism has clearly been a potent force throughout American history, with deep and lasting effects. But as education commentator Ian Rowe has pointed out, the number of white students who score below “proficient” on NAEP reading tests is far greater than the number of Black or Hispanic students who do so. It’s true that Black and Hispanic students are disproportionately likely to score at the “basic” or “below basic” level, but if racism were the fundamental reason for poor performance, why would so many white students also fall into those categories?

In fact, when you refine the data to include both race and income for the same set of students, a different picture emerges. One study found that what appear to be racial achievement gaps are entirely explained by poverty: all students at high-poverty schools tend to fare poorly on tests, regardless of race. Another found that racial disparities in special education largely disappear if you control for family income.

It’s clearly important to understand the role of poverty as well as race in education outcomes. But it’s even more important to understand how those factors intersect with others we rarely measure or emphasize. If we think test-score gaps are about race, we’ll attack racism. If we think they’re about poverty, we’ll attack that. While our society should definitely be fighting racism and poverty, that struggle is likely to be long and difficult. What if test-score gaps could be linked even more closely to other measurable student characteristics? And what if that information could point the way to changes that would have a more direct effect on academic achievement and be quicker and easier to bring about?

Parental education data could change the picture

The characteristics I’m talking about are parental occupation and level of education. Those two indicators, along with income, are considered the primary components of socioeconomic status, but they’re rarely measured in the context of academic testing. What if we had data showing, for example, that scores are significantly higher for Black students with highly educated parents, or students living in poverty whose mothers are doctors?

We do have a little of that kind of data, and it shows we need more. For example, a report from the Brookings Institution in 2012 found that poor children were, on average, significantly less ready for kindergarten than their wealthier peers—only 48% were prepared, versus 75% for those from moderate or high-income families. But there was one exception: poor children who had mothers with a BA or above were more likely to be kindergarten-ready than wealthier ones in the same category: 91% as opposed to 84%.

Armed with this kind of information, we could start asking why poor kids with highly educated parents, however few in number, do so much better than others of the same income level—or, perhaps, the same race. The likely answer, judging by what we know from cognitive science, is that they’re able to absorb more academic knowledge and vocabulary at home, which better equips them for further learning. And the solution—if we’re thinking clearly—would be to provide kids from less highly educated families with more and better exposure to that kind of knowledge and vocabulary at school.

Many assume that’s what schools have been doing, or trying to, and it just hasn’t worked, But for complicated reasons that’s not what’s happening, at least at the crucial elementary level. Elementary schools could do much more to boost learning for all children by adopting content-rich, coherent curricula beginning in kindergarten and implementing them well—as some schools across the country are now beginning to do.

True, we would have to rely on students to report their parents’ job titles, or whether they have a college degree, and experts have pointed out that kids may make mistakes. But the data couldn’t be any more flawed than the FRM data we’ve been using to measure poverty, and it could be a lot more valuable.

For decades, we’ve been framing disparities in academic outcomes in terms of poverty and—especially recently—race. We don’t have much to show for that, with only about a third of all students testing proficient or above on NAEP reading tests. If we start collecting different kinds of data—and slicing and dicing it so that racial and income categories are no longer viewed as monolithic—we may come to a different and more accurate understanding of the causes of the problem. And if that happens, we might finally begin to make some real educational progress.

This post originally appeared on Forbes.com.