Universal Preschool?
Tis education forms the common mind / Just as the twig is bent the tree's inclined. (Alexander Pope, 1732.)
Should publicly-funded preschool be available to all children?
This isn't just a question for parents and educators. In part, it's a question about how to spend tax dollars – and about how to promote the well-being of future generations of Americans.
In this newsletter I'll be discussing some of the evidence bearing on this question, including new data from longitudinal studies in Oklahoma and Tennessee, one attesting to the long-term benefits of universal preschool, the other apparently revealing negative influences. I'll also be describing how political leaders and even scholars may misrepresent statistical data in the service of an ideological agenda.
I want to start by defining some terms, and, in the process, provide some context for the issues.
What is "preschool"?
Briefly, a preschool is an establishment that provides education and care prior to elementary school.
Typically there's no legal distinction between preschools and "day cares", "playcares", "early childhood learning centers", etc. They all have to meet the same minimum requirements to operate. But a business that calls itself a "preschool" is implying an educational emphasis, including a curriculum intended to prepare children for school.
The terms "preschool" and "pre-kindergarten" (or "pre-K") are often used interchangeably. In some cases, the term "pre-kindergarten" is reserved for programs that serve children during the year immediately prior to kindergarten.
What is "universal preschool"?
"Universal preschool" refers to publicly-funded preschool available to all children of a certain age range. In practice, this often amounts to universal pre-K, because funding is available for just one year.
Universal pre-K can be be contrasted with targeted pre-K – i.e., publicly funded programs restricted to specific groups (usually children from economically disadvantaged backgrounds). The mix of programs available to children varies from place to place.
For example, all 4-year-olds who live in Boston are eligible for one free year of high-quality preschool – specifically, up to 6.5 hours per day, for up to 180 days, in an accredited program where all teachers have degrees in early childhood education. Boston's universal pre-K program is considered one of the best in the nation, in part because studies have linked participation to cognitive and socio-emotional gains, as well as long-term benefits such as higher rates of college attendance. Happily, Boston is now piloting universal pre-K with 3-year-olds.
Across the Charles River from Boston is Cambridge, where I live. Here what you'll find are discussions by the City Council for possibly implementing some sort of universal preschool program by 2026, but no actual programs. Our city doesn't offer much targeted pre-K either. What we have are "scholarships" which cover part of the tuition costs for a small number of eligible families.
In between these extremes is the nearby city of Watertown, which offers free pre-K targeted to economically disadvantaged 3-through 5-year-olds, and is using money from its federal COVID school assistance grant to expand these services. The plan is for targeted pre-K in Watertown to gradually expand until it becomes universal within the city, although it's unclear whether these programs will be held to the same standards as those in Boston.
Why isn't universal pre-K universal?
In 2021, the proposed Build Back Better Act included roughly 200 billion dollars to fund universal pre-K for all American 3- and 4-year-olds. This part of the Act was one of the casualties of negotiations with Joe Manchin. The version signed into law this August, the Inflation Reduction Act, contains no funding related to education or child care. Thus, what America continues to have is a mix of states and cities that more or less resemble Boston, Cambridge, or Watertown. In other words, the kinds of universal and/or targeted pre-K programs available to families, if any, depend on where they live. It goes without saying that the quality of what's available varies widely from city to city.
In some instances, both supporters and opponents of President Biden's plan were guided more by ideology than by evidence. Here, for example, is the administration's official statement on the academic benefits of Build Back Better's universal pre-K funding:
"Preschool is critical to ensuring that children start kindergarten with the skills and supports that set them up for success in school. In fact, research shows that kids who attend universal pre-K are more likely to take honors classes and less likely to repeat a grade, and another study finds low-income children who attend universal programs do better in math and reading as late as eighth grade. Unfortunately, many children, but especially children of color and low-income children, do not have access to the full range of high-quality pre-school programs available to their more affluent peers..."
I'm a supporter of universal pre-K, but this statement doesn't do justice to the existing data.
1. In some ways, the statement isn't as strong as it could be. A number of studies have linked pre-K attendance to high school graduation rates, college enrollment, and other positive long-term outcomes.
2. In some ways, the statement is too strong. Not all studies indicate benefits of preschool attendance beyond kindergarten. In some cases, the results of favorable studies aren't generalizable, owing to small sample sizes and/or a focus on programs of unusually high quality.
3. The statement conflates universal and targeted preschool. Researchers may disagree about the benefits of preschool, but there's consensus that to the extent that it is beneficial, the benefits are greater for children from disadvantaged groups. In other words, the need for more targeted preschool funding seems clear, but this alone doesn't justify a need for universal preschool funding. To make a case for universal preschool, you need to show that middle-class kids benefit too. I do think one could make a case for that, but the White House statement doesn't try. Rather, it uses the successes of preschool among disadvantaged kids as a stand-in for likely successes among all kids.
Objections to Build Back Better's universal preschool plan, mostly from Republicans, ranged from sensible to ridiculous.
Sensible concerns included the usual questions about who would pick up the tab. High-quality preschool is expensive, and it was never clear from President Biden's iterations of the plan how much cost states would bear at first, or what would happen when federal support expired. The White House presented data on the long-term economic benefits of universal preschool (which would accrue from better-prepared students, as well from parents who could work while their children are in preschool). However, bipartisan discussion was hampered by each party's focus on their own statistical truths. The Republicans, along with folks like Joe Manchin, cited statistics on the economic burden of universal preschool, while the Biden Administration cited statistics on long-term payoffs. Even if each camp's statistics were accurate, they were still talking past each other. The Republicans were saying: We can't afford this ship. The Biden Administration was saying: The ship will pay for itself, after it sets sail.
At the ridiculous end of the spectrum were Republicans who accused President Biden of pushing a socialist agenda and trying to brainwash American children via standardized preschool. Tennessee Senator Marsha Blackburn, for example, tweeted "You know who else liked universal day care[?]", answering her own question with a link to a story about subsidized programs in the Soviet Union, where, among other things, children were indoctrinated in communist ideology. Shame on Senator Blackburn for this unhelpful observation, and for using the term "day care" in her tweet. Both the Soviet programs, and what President Biden proposed, are preschools, not day cares.
In the next section I briefly describe the case against universal pre-K, followed by the case for it, discussing in each instance new data emerging from Tennessee and Oklahoma. My focus will be on the potential educational benefits for children. In other words, I won't have much to say about cost-benefit estimates, economic impacts on society, benefits for parents, health outcomes for children, and other issues of relevance that scholars discuss and debate.
The case against universal pre-K
Opponents of publicly-funded universal pre-K typically assert one or more of the following:
1. Pre-K doesn't benefit kids unless it's unusually high quality (and thus prohibitively expensive).
2. Pre-K only benefits disadvantaged kids (in which case, we only need targeted pre-K).
3. Whoever does benefit from pre-K won't enjoy those benefits for long – i.e., the effects fade out, and/or the other kids catch up.
The first argument is difficult to address given disagreements about how to define quality, and, in some cases, lack of details about quality in particular studies. People also disagree on what the benefits of preschool should be. However, I think it's safe to say that if cost were not an issue, preschool could be beneficial for most children, according to what most people want preschool programs to accomplish.
The second argument seems overstated. Literature reviews and meta-analyses only suggest that pre-K benefits disadvantaged kids more, or more consistently, than kids from middle-class backgrounds, not that kids from more advantaged backgrounds never benefit. Large-scale, rigorous studies have documented long-term academic benefits of preschool among middle-class children.
The third argument is contradicted by various studies linking pre-K participation to long-term benefits. In the case of Head Start, for example, as well as the Boston Universal Pre-K program, the data do reveal that initial benefits disappear, but later on, program participants show higher rates of high school graduation and college enrollment, for example.
Studies on the educational impact of preschool attendance tend to either show positive effects or no effects at all. Relatively few studies have documented negative effects. (In a moment I'll be discussing one of the prominent exceptions.) Thus, one might argue that universal pre-K is worthwhile, because it will benefit some children, and some of the benefits will be long-term, but negative outcomes will be rare.
Here's an analogy: The R21 vaccine, currently under development, looks to be safe and about 80% effective at preventing malaria. Should we withhold this vaccine from children simply because it won't be effective in every case? Of course not. Since we don't have a better alternative, we will administer this vaccine if it's approved, and, hopefully, discover why it's not effective in some cases. Should this vaccine be withheld if we later discovered that its effectiveness dwindled substantially after three years? Here again I think not. But we might try to improve mosquito control, so that after three years passed, the risk of exposure to malaria would be lessened.
In short, we don't withhold good things from children simply because not all children would benefit, because many of the benefits don't endure, or because not every study agrees on the benefits. If you oppose universal pre-K, you need to stick your neck all the way out and say: Here's how much it would cost to maintain quality preschool programs that have long-term benefits for many children, but we shouldn't fund these programs, because we don't have enough money, or because the percentage of children who will benefit just isn't worth the cost.
As I mentioned, most studies indicate that the educational impact of pre-K is either positive or neutral. This brings me to data that seems to reveal a different outcome – i.e., that children who attend high quality pre-K programs end up worse off than their peers.
The TN-VPK studies
Researchers at Vanderbilt are currently tracking the academic and behavioral progress of two cohorts of students who enrolled in the Tennessee Voluntary Prekindergarten Program (TN-VPK) during the 2009-2010 and 2010-2011 academic years. This is a natural experiment, in that the TN-VPK programs were oversubscribed, and so each year children were randomly chosen to either participate or be assigned to a waitlist (and then denied admission). Across the two cohorts, a total of 2,990 students are being tracked. Data analyses focus on comparisons between the two groups: children who participated in TN-VPK for one year prior to kindergarten vs. children who did not. In the researchers' view, the TN-VPK programs meet most definitions of high quality.
The Vanderbilt team previously reported that although TN-VPK participants entered kindergarten with better academic skills than their peers, these effects mostly disappeared by the end of kindergarten, and by 3rd grade, TN-VPK participants' mean achievement test performance was actually lower than that of their peers. The TN-VPK participants also showed slightly more documented violations of school rules, and a slightly higher incidence of special education placements.
The most recent report from the Vanderbilt team was published earlier this year in the prominent journal Developmental Psychology. Data were now available through 6th grade, and the negative impact of TN-VPK participation appeared to be stronger. That is, TN-VPK participants now lagged even further behind non-participants than they did on 3rd grade achievement test scores, while showing higher rates of absences, expulsions, suspensions, etc.
So, high quality pre-K programs can be bad for children...?
Bias in TN-VPK reports
Although the methodology is rigorous, there are strong signs of ideological bias in published reports of TN-VPK data. I'm not referring to political ideology necessarily, but rather to a bias for viewing data, past and present, through an especially negative lens. Here are some red flags:
1. In their 2022 article, the researchers' justification for tracking long-term outcomes is that classic studies in the 1960s and 1970s (Perry and Abecadarian) demonstrated a range of positive effects, but those studies focused on intensive programs for small numbers of disadvantaged youths, and thus "[w]hether similar results can be produced by less intensive scaled-up contemporary statewide programs for more diverse populations is a critical question for both policy and practice." In other words, the researchers ignore decades of more recent studies showing long-term benefits of participation in programs such as Head Start. Even though that research is somewhat mixed in quality and outcomes, there are numerous studies attesting to the benefits of pre-K that the researchers somehow fail to mention.
2. The researchers repeatedly cite non-significant effects that are negative, but not positive ones. This is one of the biggest red flags that I noticed. For example:
[In reference to their own 3rd-grade data] "By the end of kindergarten, most of the effects on achievement were no longer statistically significant and, in later years, nearly all had turned at least slightly negative, although generally short of statistical significance"
[In reference to a study by a different team] "It is notable, however, that the effect estimates for the third grade math scores and kindergarten special education placements were negative, though short of statistical significance."
You can't play the statistical significance game this way! Results are either significant or not, according to whatever criterion you set. If you want to call attention to trends that are "nearly" significant, or "marginally" significant, or "short of" significance, that's fine, many journal editors will allow that. But if you do that, you can't just focus on the negative trends. You have to also mention the positive ones that were "short of" significance. If your criterion for significance is a p value of .05, you can't treat a p value of .08 as suggestively close to significance only for those results that fit your agenda.
I find it particularly disturbing that the Vanderbilt researchers cited negative trends in other researchers' studies that aren't significant, but didn't mention the many positive ones. I started to count positive trends (p < .18) but quickly gave up, because there are probably dozens, if not hundreds of examples in the literature. Suffice to say that if you skim the pertinent Results sections, you can find lots of examples of pre-K participants outscoring non-participants years down the road – and yet the researchers don't treat these differences as "notable" or otherwise supportive of the idea that pre-K is beneficial, because they weren't significant. (Instead, the researchers routinely conclude that the groups don't differ on those particular comparisons.)
3. Although they claim to have been surprised and disappointed by their findings, the researchers seemed so committed to delivering a negative message that they downplayed or overlooked substantial interpretive issues. Given that some studies do attest to the long-term benefits of pre-K participation, the researchers make little sustained effort to consider qualifications and/or alternative explanations for their findings. Here are three examples of what I have in mind:
(a) The researchers fail to stress that they only examined a subset of important outcomes (scores on state achievement tests, disciplinary violations, and special education referrals), and that sleeper effects are sometimes observed. In other words, even if we trusted their negative findings, a variety of other important outcomes weren't measured, and there's evidence that pre-K programs have benefits that don't appear until high school or beyond. (It's fine to question the strength of that evidence, but the researchers barely acknowledge its existence.)
(b) The researchers paid little attention to the possibility of selection bias. Perhaps the reason TN-VPK participants performed more poorly than non-participants in 3rd and 6th grade was not their preschool experience, but simply that they tended to be less cognitively advanced in some respect. This is purely speculative, but the researchers barely mentioned the possibility, noting only that the groups were comparable in age, gender, and racial/ethnic composition.
(c) It's conceivable that the TN-VPK programs tended to be less cognitively stimulating than the activities pursued by non-participants during their pre-kindergarten year. Here again this is speculative, but the researchers only noted that the TN-VPK programs seem to be of high quality (which I'm willing to accept, although this doesn't tell us what the other kids were doing). This is a critical issue: We don't know much about the control group.
Here's a suggestive detail: The researchers mentioned in passing that "positive TN-VPK effects were found on the third grade state achievement tests for the small proportion (12%) of children who attended higher quality schools and were exposed to higher quality teachers." This finding could be viewed as consistent with the possibility that many TN-VPK programs were of lesser quality than what the non-participants were exposed to during their pre-K year, in which case one might conclude that once minimum standards of quality are attained, pre-K tends to boost achievement test scores. In other words, one could start with the same data and interpretive challenges that the researchers started with, but build a case for a completely different conclusion.
(For more information on the TN-VPK studies, see here.)
The case for universal preschool
I've described some of the educational benefits of preschool elsewhere in this newsletter; here I'll just add some comments pertinent to the rationale for universal programming.
1. Some of the skills that preschoolers acquire, like saying the alphabet, counting, or raising one's hand before asking a question, are genuinely beneficial, but not for long. Other kids catch up, because these are "constrained" skills – i.e., the kinds that get fully mastered and then serve as foundations for more advanced ones. The case for universal pre-K depends on also showing benefits that endure, or emerge later. A variety of studies do show such benefits, including better "unconstrained" skills, such as problem solving, which directly support learning across the lifespan.
2. Preschool participation has been linked to enhanced problem solving, critical thinking, and confidence in learning, as well as stronger academic performance and positive outcomes such as high school graduation and college enrollment. I'll discuss new data on this topic in a moment.
3. The case for universal pre-K depends on showing that all programs meeting minimum standards of quality tend to have long-term benefits for all types of students. I think the available evidence is consistent with this view, although the benefits are greater for children from disadvantaged backgrounds, and, as I've mentioned, there's no consensus yet on what those minimum standards might be.
The Tulsa study
Georgetown professor William Gormley and colleagues have been tracking the progress of 4,033 children in Tulsa who, in 2005, either participated or did not participate in a pre-K program (either the universal pre-K provided by the Tulsa Public Schools or Head Start). The latest findings from this study were presented at a conference hosted by the Georgetown's College of Business this September 20 and are currently available online as a working paper.
Briefly, the researchers' main finding was that children who had attended a pre-K program showed higher rates of 2- and 4-year college enrollment than children who hadn't attended a program. The effects were stronger and more consistent across student subgroups for the Tulsa universal pre-K graduates than for Head Start attendees.
As with the TN-VPK study, we might ask whether there were differences between the two groups (pre-K vs. no pre-K) that contributed to the later differences in college enrollment. For example, are parents who are more invested in their children's education, both attitudinally and behaviorally, more likely to enroll their children in subsidized preschool? If so, then the fact that their children are more likely to end up in college may not be due solely, or at all, to the preschool experience, but rather to everything else these parents did to support their children's academic success (e.g., homework guidance; enriching extracurriculars; conversations about the importance of education; etc.).
Although in practice it's difficult to rule out every version of selection bias I've mentioned here, the researchers ensured that the two groups were as comparable as possible on a range of variables. For example, they controlled for maternal education, internet access in the home, and neighborhood median income, among other things, all of which are known to be related to educational attainment. In addition, they used a technique called propensity score matching to ensure that each child in the pre-K group was compared to a child in the non-pre-K group who was most similar on numerous variables, including child care history reported by the parents.
So, if the two groups were truly comparable, as the researchers' data suggests, why would pre-K participation lead to higher rates of college enrollment? The researchers floated two possibilities: (a) The relatively high rate of pre-K participation in Tulsa may have spurred more rigorous instruction on the part of K-3 teachers, which continued to pave the way to future academic success, and (b) the pre-K experience may have readied children for the relatively strong magnet schools available to Tulsa middle school and high school students. Although it's hard to bridge the 14+ year gap between pre-kindergarten and college, I think these are reasonable possibilities. Importantly, these findings are consistent with those of many other studies pointing to the long-term benefits of pre-K.
(For more information on the Tulsa study, see here or here. For evidence that the program is highly cost-effective, see here.)
Closing remarks
A common argument against universal pre-K is that families from at least middle-class backgrounds don't need the support, because they can afford high quality programs or other educationally stimulating alternatives for their young children. I question where to draw the line between families who do and don't need this support. For example, in Cambridge the average cost of pre-K is now over $2,500 per month, which is to say that it would be cheaper to enroll your 4-year-old in any of our community colleges, or even some of the 4-year-colleges in the Boston area. I would rather see pre-K become a universally available option than a targeted one. High-quality universal pre-K would be advantageous for many of the children, rich or poor, who participate. The evidence seems clear that although some children would benefit more than others, and not all benefits would persist over time, long-term benefits would be observed among many who attend.
Thanks for reading!