Arthur Reynolds

Introduction

Arthur J. Reynolds is an American prevention scientist at the University of Minnesota's Institute of Child Development who has spent his career on a single question: whether a publicly funded preschool program for children in high-poverty neighborhoods produces effects that survive into adulthood. He directs the Chicago Longitudinal Study, one of the largest and longest-running investigations of early childhood intervention ever mounted, which has followed a single cohort of roughly 1,500 children from Chicago's poorest neighborhoods from the mid-1980s into their late thirties. His findings — higher rates of high school completion, lower rates of special education placement, grade retention and juvenile arrest, and measurable economic and health advantages in midlife — have been among the most frequently cited evidence in American debates about public investment in early education. They have also been among the most contested, because the study is quasi-experimental, and the critics are not mainly arguing about the results.

Early Life and Education

Little about Reynolds's early life is a matter of public record, and he has not written about it; what is documented is his training. He earned his Ph.D. in 1989 from the University of Illinois at Chicago, in public policy analysis rather than in psychology. The distinction matters for everything that followed. His graduate training was in the evaluation of social programs, not in developmental psychology, and the questions he asks are the questions of an evaluator: does this program work, for whom, by what pathway, and is it worth what it costs?

He was a professor at the University of Illinois at Chicago before moving to the University of Minnesota, where he is a professor in the Institute of Child Development and co-director of the Human Capital Research Collaborative.

Academic and Professional Career

Reynolds began the Chicago Longitudinal Study in the 1980s, before he had finished his doctorate. The Child-Parent Center program had been operating in Chicago public schools since 1967, funded under Title I, and provided a comprehensive preschool-to-third-grade model: small classes, activity-based instruction, deliberate alignment of instruction across ages, family services, home visiting, and health and nutrition workshops. It was one of the oldest federally funded preschool programs in the country and, at the point Reynolds began studying it, had never been evaluated over the long term.

The study assembled a treatment group of 989 children who enrolled in CPC preschools at ages three and four between 1983 and 1985, and a comparison group of 550 same-age children who attended the usual district programs — principally full-day kindergarten in schools participating in the Chicago Effective Schools Project. Nearly all participants were African American or Latino, growing up in neighborhoods with roughly double the city's poverty rate. Data have been drawn from birth records, K–12 school records, surveys, interviews, and administrative records on education, health, and earnings, and the cohort has been followed into its late thirties.

Reynolds's substantive contribution beyond the data collection is what he calls the Five Hypothesis Model, which specifies the pathways by which early education is supposed to produce adult outcomes: cognitive advantage, family support, motivational advantage, socio-emotional adjustment, and school and community support. The model is an attempt to explain the mechanism rather than merely to report the effect, and it has shaped how subsequent interventions in the CPC tradition are designed.

With the economist Judy Temple he has conducted a series of cost-benefit analyses, the first at age 26 and later work extending to health outcomes at age 35–37. These reported social returns of several dollars per dollar invested, and they are among the most politically consequential numbers in the field, cited regularly in legislative debates about preschool funding. In 2025 the Society for Prevention Research gave him its Prevention Science Award. He also directs the Midwest Child-Parent Center Expansion Project, launched in 2012, which is replicating the model in schools in Minnesota, Illinois, Wisconsin, and Iowa — and doing so, this time, as a randomized trial.

Major Contributions to Psychology

Evidence That Early Intervention Survives

The core finding of the CLS is durability. Most early childhood programs show impressive gains on school-entry readiness tests and then those gains fade. Reynolds's work has reported effects that do not fade: higher rates of high school completion, reduced grade retention and special education placement, lower rates of juvenile arrest and felony conviction, and — in the age-35 and age-37 waves — better adult health, including lower rates of smoking, drug use, diabetes, and depression, together with measurable economic benefits. The midlife health findings are unusual in the intervention literature and are among the first to connect a preschool program to physical health three decades later.

An Evaluator Independent of the Program

One methodological point in the study's favor is rarely mentioned in critiques of early childhood research. In the Perry Preschool and Abecedarian evaluations, the people who designed the programs were also the people who evaluated them. Reynolds did not develop the Child-Parent Center program; he came to it afterward, which makes the CPC evaluation more independent than the two most famous studies it is usually compared to. This is a genuine advantage and reviewers have said so.

Prevention Science as a Discipline

Reynolds's broader contribution has been to put early childhood education inside the framework of prevention science, where the standard of evidence is the same as in clinical psychology and public health: defined programs, measured implementation, long-term follow-up, and economic evaluation. The Human Capital Research Collaborative he co-directs is an institutional expression of that approach, and the Midwest expansion project, by using random assignment, is an attempt to meet his critics on their own terms.

Major Publications

  • Reynolds, A. J. (2000). Success in Early Intervention: The Child-Parent Centers. Madison: University of Wisconsin Press.
  • Reynolds, A. J., & Temple, J. A. (2008). Cost-effective early childhood development programs from preschool to third grade. Annual Review of Clinical Psychology, 4, 109–139.
  • Ou, S.-R., & Reynolds, A. J. (2008). Predictors of educational attainment in the Chicago Longitudinal Study. School Psychology Quarterly, 23(2), 199–229.
  • Reynolds, A. J., Temple, J. A., Robertson, D. L., & Mann, E. A. (2011). Age 26 cost-benefit analysis of the Child-Parent Center early education program. Child Development, 82(1), 379–404.
  • Reynolds, A. J. (Ed.). (2003). Early Childhood Programs for a New Century. Washington, DC: Child Welfare League of America.
  • Varshney, N., Temple, J., & Reynolds, A. (2022). Early education and adult health: Age 37 impacts and economic benefits of the Child-Parent Center Preschool Program. Journal of Benefit-Cost Analysis, 13(1), 57–90.

Impact and Legacy

Reynolds's work has been cited in support of preschool expansion at every level of American government, and his cost-benefit figures have been among the most persuasive numbers available to advocates. The durability of the CPC findings, and the midlife health results in particular, have strengthened the case that early education is a public health investment and not merely an educational one.

The criticisms have focused almost entirely on design, and they are serious. The Chicago Longitudinal Study is not a randomized experiment. Families selected themselves into the program, or were selected by it, and the comparison group consisted of children who attended other programs — full-day kindergarten, and in about fifteen percent of cases Head Start — rather than children who received nothing. Every causal claim therefore rests on the assumption that propensity score models, inverse probability weighting, and Heckman-type selection models have removed the differences that matter. Thomas Cook and Vivian Wong, both specialists in quasi-experimentation rather than opponents of early childhood education, wrote that the CPC evaluation "depends on an opaque matching procedure and on data analyses that have routinely failed to recreate similar effect sizes to an experiment on the same topic," which implies the possibility of a selection confound that has not been fully controlled. Douglas Besharov and colleagues went further, arguing that high levels of missing data, the absence of a true randomized experiment, and various self-selection biases "raise considerable uncertainty about the findings," and noting that only unmeasured variables associated with measurable proxies can be modeled — so some uncertainty is irreducible.

Two further problems are specific to this study. The first is timing: the program began in 1983 but the CLS cohort's main background variables were collected retrospectively from 1986, whereas in the Perry and Abecedarian evaluations data collection began when the programs were implemented. The second is that the background variables themselves are treated as static over decades in which neighborhoods and families changed substantially, and Reynolds's own analyses have produced noticeably different effect estimates depending on which dataset, which completeness threshold, and which statistical method is used. Independent reviewers have also observed that the CPC team's reporting has a positive-outcome emphasis that makes the null and negative results harder to see.

Generalizability is the remaining constraint. The cohort is a single group of children born in 1979–1980, ninety-three percent African American, in the highest-poverty neighborhoods of one city, served by a Title I program with resources and class sizes that are not typical. Whatever the study shows, it shows it for that population in that period.

Reynolds's response has been methodological rather than rhetorical. The study team has adopted increasingly elaborate covariate balancing and attrition weighting; the comparison against children in other programs is, as they point out, a conservative test rather than a flattering one; and the Midwest Child-Parent Center Expansion Project was designed from the outset as a randomized controlled trial specifically to supply the causal evidence that a matched-group design cannot. That study is the proper test of the CPC claim, and it is the reason the argument about Chicago is unlikely to be settled by further analysis of the Chicago data.

Learn More

Last updated: 2026-10-20

Menu
Loading