Should we use genetics to understand high-potential, low-income students?
Low-income children face many disadvantages that impede academic success. But how much do those disadvantages matter, and how early do their effects become visible? Test scores—the standard tool for answering such questions—can take us only so far because, by the time children sit for them, family background, socioeconomic status, and early educational experiences have already shaped much of what they measure. The scores reflect not only students’ capacity to learn, but also the language, instruction, stability, and opportunities they’ve encountered—or not—along the way. The data therefore cannot cleanly distinguish underlying potential from the accumulated effects of disadvantage.
A recent working paper from the U.K. by John Jerrim, Maria Palma Carvajal, and Tim Morris takes an unusual, and surely controversial, approach to this problem. Rather than identifying high-potential children based on early test scores, the authors use a measure derived from children’s DNA—entering already-contested terrain, given the long and fraught history of efforts to distinguish innate ability from environmental advantage.
Their measure, called an education polygenic index (PGI), combines thousands of small genetic differences associated with spending more years in education. (See my colleague Mike Petrilli’s review of a book about education PGI for background.) It’s important to note that the measure is not a genetic IQ test, does not determine any child’s future, and is far too noisy and entangled with family and social circumstances to justify decisions about individual students.
Instead, its value here is narrower: Because a child’s DNA is fixed before birth, the measure lets researchers compare children with similarly high scores on the researchers’ education-linked genetic index, then examine how sharply their early skills and later academic outcomes diverge depending on family income. (For readers worried that the index is too imprecise to support much, the authors test a range of assumptions about its reliability, and the basic findings hold.)
The authors define low-income and high-income families as those in the bottom and top quarters, respectively, of a measure of average family income collected across multiple years on the United Kingdom’s Millennium Cohort Study, which has followed children born between 2000 and 2002 from infancy through adolescence. Their analysis includes 6,118 children with available genetic data, more than 99 percent of whom are white—a major limitation reflecting both the difficulty of using current polygenic indices across ancestry groups and the researchers’ effort to avoid mistaking ancestry-related social differences for genetic effects.
Of particular interest are the 143 low-income and 674 high-income children whose PGI scores fall in the top quartile of the measure. The researchers conduct comparisons in four areas: early cognitive skills, including vocabulary; literacy experiences at home; attitudes toward school; and self-reported grades on national exams taken around age 16.
The first notable result, on early cognitive skills, is how soon the gaps appear. The authors estimate that low-income three-year-olds with similarly high PGI scores perform about 0.4 standard deviations below their high-income peers on a vocabulary assessment. By age five, the estimated gap is roughly 0.7 standard deviations. Both gaps are statistically significant, though the evidence that the gap itself widened is less definitive.
Beyond vocabulary, the findings on early cognitive skills are less uniform. Income gaps also appear on tests of nonverbal pattern recognition, but they do not widen between ages five and seven, and the authors find no robust difference in math scores at age seven.
Family-income gaps also appear in children’s early literacy experiences at home. Parents in the low-income group report reading to their children and taking them to libraries less often. Those gaps are not enormous, but they suggest that differences in early literacy experiences may help explain part of the vocabulary divide.
The results on attitudes toward school are mostly inconclusive. One curious exception is that low-income children with high PGI scores are more likely to say they enjoy math at ages seven and 11.
By age 16, the academic divide associated with family income is hard to miss. Low-income children in the high-index group are roughly 20–30 percentage points less likely to earn a top grade in English or math on national exams than their high-income peers. But the gaps are smaller and less consistent when the outcome is merely clearing a basic passing threshold. Most low-income students in the high-index group manage to meet that standard, even as far fewer reach the highest levels.
These are sobering results, though the usual cautions apply with heightened force here, given the use of DNA. Because the group at the heart of the analysis includes just 143 low-income children, several estimates are imprecise. And the genetic measure remains noisy and partly entangled with the very social and environmental influences that the study is trying to distinguish. The paper therefore does not isolate the causal effect of income, nor does it tell us which particular features of disadvantage matter most. It also does not show that low-income children at the top of the researchers’ polygenic index inevitably fall behind. Indeed, the uneven findings across subjects and ages caution against that sort of deterministic reading.
What it does show is that children with similarly high scores on the authors’ genetic measure can have sharply different academic trajectories associated with family income. The divergence is visible in language skills by age three and in top-level exam grades by age 16.
For other researchers, the study demonstrates both the promise and the limits of using genetic information to study academic prospects before test scores have absorbed years of environmental and school influences. Larger and more diverse samples—and better measures—are plainly needed.
For educators and policymakers, the lessons are more immediate. Schools should help all students fulfill their potential, including advanced learners. Yet that is not happening consistently for low-income students. These U.K. findings echo what Fordham has documented about high-achieving, low-income students in the United States: Too many lose academic momentum along the path from K–12. If schools could better identify and cultivate their talents, many more would be prepared to succeed in college, including selective colleges, and beyond.
What this study reinforces is that present achievement is an imperfect guide to future academic success, especially for children whose early opportunities have been constrained by family income. Schools should therefore identify advanced learners broadly and repeatedly rather than screening only once or twice at fixed ages. By then, part of what educators hope to discover may already have been obscured by unequal opportunities.
SOURCE: John Jerrim, Maria Palma Carvajal, and Tim Morris, “Academic outcomes amongst children from rich and poor backgrounds with a genetic disposition for educational attainment: Evidence from the Millennium Cohort Study,” University College London working paper (2026).
* Brandon Wright is the Editorial Director of the Thomas B. Fordham Institute. He is the coauthor or coeditor of three books: Failing our Brightest Kids: The Global Challenge of Educating High-Ability Students (with Chester E. Finn, Jr.), Charter Schools at the Crossroads: Predicaments, Paradoxes, Possibilities (with Chester E. Finn, Jr. and Bruno V. Manno), and Getting the Most Bang for the Education Buck (edited with Frederick M. Hess). His writing has appeared in places like the Wall Street Journal, the Washington Post, U.S. News, the New York Daily News, the New York Post, National Review, Newsweek, Education Next, Education Week, Phi Delta Kappan, the Journal of School Choice, and dozens of state newspapers. He holds a J.D. from American University Washington College of Law and a B.A. from the University of Michigan. He lives with his wife and two children in Southeast Michigan.