Help boys, but first do no harm

In his recent book, Richard Reeves (2022) brought to the fore the important challenges faced by men and boys. The rapidly changing economy and evolving social norms have been particularly hard on men, resulting in too many of them—one in nine prime-age men in 2022—not in the labor force. This in turn influences a range of additional social maladies including fewer marriages (and fewer children living with their fathers) and increased deaths of despair. Boys have been falling behind girls across a range of measures of academic success from pre-kindergarten through college. Reeves is absolutely right to point out these ways in which the modern male is struggling. Men's success matters—to families, to communities, to our economy and to society.

To address these growing problems, Reeves suggests a series of reforms to policy and practice in education. Most of these I also endorse. More choices in educational options can help families find the best educational environment for their children to thrive. One size does not fit all in education. Some boys may be more likely to thrive in certain environments, such as the vocational high schools and apprenticeships endorsed by Reeves. They should also include a wider array of options such as single-sex public schools and charter schools.

Reeves also recognizes the importance of mentoring for boys’ educational success. He calls for policies to encourage more men to become teachers—a worthy goal, to be sure. I would add to this menu wider use of individual, in-school tutoring, which has been shown to be a cost-effective way to improve boys’ achievement in high school (Guryan et al., 2023).

Note that improving boys’ school performance does not imply harming girls’ performance. Skill development is not zero-sum, and when students learn more, they go on to add more to our shared economy. With this in mind, there is likely untapped potential both to improve educational outcomes and to close achievement gaps through improvements in curriculum, innovative use of educational technology, and in more traditional approaches such as reduced class size and increased school funding. We owe it to our nation's future to continuously improve education—which is especially needed as we work to offset the substantial learning losses during the COVID-19 pandemic (Hanushek & Woessman, 2020).

One of Reeves's most provocative recommendations is to change the default rule for school entry age, so that boys start kindergarten a year older than girls. This practice is commonly called “redshirting.” It is here that Reeves and I part company. I think one-size-fits-all redshirting will on net be socially and financially costly.

At first glance, Reeves's logic on redshirting is compelling. Boys are less mature than girls, and their brains develop differently. Boys have behavioral problems that improve with age. Girls might even benefit if they are in classes with more mature, and better behaved, boys. As a result, maybe redshirting is a low-cost way to improve outcomes for all students.

On deeper reflection, universal redshirting is a costly, one-size-fits-all policy for a problem that is relatively isolated. There are many cheaper, better solutions that can address the challenges more directly. And the boys and their families affected by the Reeves proposal would certainly incur substantial costs—a later start to their careers, an increased risk of high school dropout, as well as the upfront costs of an additional year of childcare prior to the delayed kindergarten entry. Moreover, to the extent that a delay in school entry may be helpful for some children, there are many boys who would not benefit from delay as well as some girls who would. A better approach would be to allow parents to make individualized and informed decisions about redshirting their own children.

These theoretical and cost-benefit considerations aside, the empirical evidence in general it does not support Reeves's policy proposal. We can help struggling boys in a less costly, more targeted, and more effective manner than delaying school starting for them all.

A “redshirted” child—where redshirting comes from a term borrowed from college athletics—starts kindergarten a year later than their initial age-eligibility. Parents tend to do this for children who they think need more time to develop skills before entering elementary school and who would be advantaged by being relatively older in their classroom throughout their school years. Empirically, boys who are redshirted are mostly summer-born boys from families with high levels of income.

But Reeves does not argue for redshirting with an eye on improving kindergarten readiness and early-grade success. Instead, he argues that the extra year of age will be most beneficial during high school when the differences between boys’ and girls’ brain development are largest. That additional year of development would potentially mean that more boys would succeed in high school math classes, get better grades across the board, score better on college entrance exams, and generally benefit more from the schooling investments that society makes in them.

In some communities redshirting is common, but nationally only a small minority of children—fewer than one in ten—are redshirted. Even among highly educated families with summer-born sons, the group most likely to be redshirted, the most recent national statistics suggest only 1 out of 5 children delay kindergarten entry. Furthermore, redshirting rates are not all that different for boys and girls (Schanzenbach & Larson, 2017). Redshirting all boys, as Reeves suggests, would be a radical change in norms.

To better understand how redshirting would impact boys’ progression through school, it is useful to separate the impact into two factors that would both change. First is the biological age at which a boy will face a given curriculum. Universal redshirting would push that age back one year for boys, so for example the age at which a boy would face 9th grade algebra, from now typical 14 to a redshirted age 15. To the extent that some children's brains may be more ready for algebra at the older age, we may observe improved achievement due to the shift in curricular timing. In the research studies I discuss below, this mechanism is often referred to as the “absolute age” effect. But any shift in boys’ absolute ages also shifts their relative age in the classroom. To think clearly about the impacts of redshirting, we have to disentangle these two effects.

On the side of social interactions, the question is whether education be improved if older boys and younger girls share the same classroom. If so, there are other, less costly, ways to address each of these factors that would not involve delaying school entry for half of the population. For example, if we were to want to change the social composition of classrooms by putting older boys into classrooms with younger girls, we could accomplish this by mixing grades or making other changes to classroom assignments that could generate some of the benefits but with lower costs.

It is likely that the optimal age to introduce a child to a particular curriculum is different for every child. Instead of limiting these options to two (i.e., on the current schedule or delayed by an entire year), there are more targeted and effective strategies to better tailor the curriculum to children's individual needs. This can be done through a variety of approaches, including classroom groupings, tutors, and computerized instruction.

Another flaw with the Reeves proposal for a universal redshirting policy is its sizeable economic costs. While the decision to redshirt is often framed by school administrators as the “gift of extra time,” it is more appropriate to frame it as the “cost of extra time.” And these costs should be carefully weighed against potential benefits.

Starting school at age 6 instead of age 5 means heading off to college or entering the full-time workforce at age 19 instead of age 18. If the number of completed years of education is unchanged, the redshirted child will have a career that is 1 year shorter than he would have had if he started kindergarten on time. The implied present discounted cost of losing that year in the labor force for a college-educated male who retires at age 67 is over $80,000. To offset this loss, the implied increase in test scores would have to be quite large.

Universal redshirting would also impose sizeable economic costs to the extent that it changes the composition of high school dropouts. To be sure, dropping out of high school is one of the worst economic decisions that a student can make; earnings trajectories are miserable for high-school dropouts. The concerns here are twofold. First, students who drop out at the minimum schooling age (which ranges from 16 to 18 depending on the state) will have completed one fewer year of schooling and accumulated fewer skills to take to the market. Second, students who otherwise would have stuck out school through age 18 and graduated would then have an extra year to decide to drop out. In other words, instead of leaving school at age 18 with a high school diploma, more will be leaving at age 18 with an 11th grade education. It is difficult to predict the magnitude of this response but given what we know about the social costs of high school dropout in terms of increased crime, reduced wages, and less stable families (Rumberger, 2020), even a small increase in the dropout rate will impose large costs.

Furthermore, universal redshirting will also impose additional costs on families with boys to remain out of the workforce or provide childcare for an additional year. Currently low-income families are less likely to delay their child's school starting age, which in part likely reflects these burdensome costs.

Notice who bears each of these costs: the children themselves or their families. A better approach would be to shift the costs of better educating boys to schools and communities.

As I argue above, the costs of universal redshirting are large. Further, the research evidence suggests that the benefits are likely to be small.

Research studies have attempted to separate the effects of redshirting (or school staring age more generally) into absolute age vs. relative age effects. This is a daunting empirical challenge for many reasons, not the least of which is that students who are old for their grade—due to reasons including redshirting or retention—differ along many dimensions from those who progress through school at the typical ages. Since there has not been a true randomized experiment on redshirting, researchers have use quasi-experimental methods to isolate the impacts of absolute and relative age on student outcomes.

In an excellent example of such an approach, Elder and Lubtosky (2009) took advantage of differences across states in the birth dates used for school-entry cutoffs. In some states, to be eligible for kindergarten, a child must turn 5 by December 2, others require a child's fifth birthday to fall before September 1, and so on. In states with earlier cutoffs, children who adhere to the guidelines (and do not enter a year late or early) are on average older than they are in states with later cutoffs. This type of variation allows them to measure the impact of the child's age at kindergarten entry.

Elder and Lubotsky (2009) found that an additional year in absolute age has a large effect on kindergarten math and reading test scores. The effects decline steadily as children age but remain statistically significant through 8th grade. When they analyze the absolute age effects separately by the child's family socio-economic status (SES), they find that effects are larger and more persistent for children from higher-SES families. Among students in the lowest SES quartile, the positive effect of an additional year of absolute age is no longer statistically significant in math by fifth grade and in reading by third grade. In unpublished tables provided to me, they find no differences in impact by gender. They also find that students get a boost from being in classrooms with older peers. Unfortunately, though, the laws of math do not allow us to group all students with older peers.

In a study I coauthored with Elizabeth Cascio, we separated the impacts of age and relative age using data from Project STAR, a large experiment that randomly assigned children to different-sized classrooms (Cascio & Schanzenbach, 2016). As a result of the randomization, by the luck of the draw, two children with the same absolute age will fall into different positions in their classroom's relative-age distribution. Similar to the prior study, we found that the benefits of absolute age declines as students get older, and that students’ performance improves when they are in class with older classmates. In contrast to the Elder and Lubotsky findings, we found that both absolute and relative age effects are stronger in 8th grade for children from low-income families (as measured by receipt of free or reduced-price lunch). Among children from higher-income families, neither absolute nor relative age effects are statistically significant by 8th grade. In support of Reeves's hypothesis that age effects are more important for boys, we did find that boys are more impacted by both age and relative age than girls are.

A few big caveats are worth mentioning here. First, both studies (as well as other studies such as Dhuey et al., 2019) found that there may be a positive impact of age when it comes to reasonably rare outcomes such as grade retention or diagnoses of learning disabilities. This finding suggests that targeted redshirting may help students at risk of such outcomes. Second, both studies used relatively modest variation in absolute and relative age to measure their impacts. If we were to adopt Reeves's policy of universal redshirting of boys, the shifts in both absolute and relative ages that would induce would be much larger than the variation used in current studies. Third, neither study directly investigated the high school performance that Reeves is attempting to impact through his universal redshirting proposal. It is possible—though I would not bet on it—that after declining across elementary and middle school years, age effects get larger in high school.

Both studies described above clearly find that students are helped, not harmed, when they are in classrooms with slightly older peers. Note that studies finding positive spillovers from older peers are based on the range of variation usually seen in classrooms—usually a couple of months—and should not be extrapolated to what would happen if a child was in a classroom with much older students. While the girls may be somewhat helped academically being grouped with older boys, at the same time the boys’ achievement would be somewhat harmed by the grouping. There may be other non-academic impacts to consider as well, such as potential effects on risky behaviors such as teen pregnancy and substance use.

The studies differ in their findings on the impact of absolute age. How do we reconcile these differences? Recall that Elder and Lubotsky's (2009) study compared students across states with differing school-entry cutoff ages, while Cascio and Schanzenbach's (2016) study compared students born in different months who attend the same school. In the former, states may have adopted different curricula and other practices based on the age distribution of their students. In other words, the kindergarten curriculum and what kindergarten teachers may cover could be more advanced in a state with a late school-entry cutoff (and thus with on average older students) than it is in a state with an early one (and thus with on average younger students). On the other hand, Cascio and Schanzenbach compared students who vary in absolute age but attend the same schools and classrooms, and thus are largely exposed to the same curriculum and practices. For the thought experiment of redshirting all boys, the Elder and Lubotsky results are likely more relevant because schools will certainly adapt their practices as a result of the change.

In summary, we certainly need to pay attention to improving boy's achievement levels. There are many evidence-based reforms that are likely to be useful to this end.

But we should not change the default school entry age for boys. The costs of this policy are likely to vastly outweigh the potential benefits. We can and should do better by addressing the unique developmental challenges faced by boys in more appropriate, targeted manners through changes in schools’ policies and practices. For the small share of children who would benefit from a delayed school starting age, that decision should be made by parents with input from their child's school, not by a one size fits all policy.