Earlier intervention has greater effects in changing a child’s life course, but not always a higher benefit-cost ratio; both are important

Posted on September 25, 2014 by timbartik

In my new book, From Preschool to Prosperity, one issue I discuss is how the returns to high-quality early childhood programs vary with what child ages are targeted by the program. One popular notion is that the earlier the intervention, the better. Although this notion has a kernel of truth, the research evidence on the whole supports the notion that programs at pre-school ages tend to have the highest benefit-cost ratio.

The kernel of truth is that earlier interventions seem to have the potential for more dramatic effects on adult outcomes. For example, evidence from the Abecedarian experiment, which provided full-time child care and pre-K from birth to age 5, suggests that such services can boost adult earnings by over 25%. (The Abecedarian program is similar to today’s Educare program.)

But from a benefit-cost standpoint, earlier interventions also cost more per child. The younger the child, the more interventions have to have smaller class sizes or be one-on-one in order to be high quality. This lowers the benefit-cost ratio.

For example, the evidence suggests that one-year of full-day pre-K at age 4 can raise future earnings by 10%. This increases the present value of future earnings for former child participants by about $50,000, at a cost of around $10,000, for a benefit-cost ratio of over 5 to 1.  Extending this program to younger ages by adding in all the services of an Abecedarian-style program from birth to age 4 would increase the future earnings benefits to 26%, or two-and-a-half times as much.  The increase in the present value of future earnings is around $130,000, which is a much more dramatic increase than for one-year of preschool. But the cost of this program is almost 9 times as much, or almost $90,000 for the five years of full-time child care and pre-K.  The benefit-cost ratio in terms of benefits for children is around 1.5, which is considerably lower than for one year of pre-K.  An Abecedarian program has much larger child care benefits for parents, which would more than double the earnings effects of the program. But even with these parental earnings benefits, preschool at age 4 still has a somewhat higher benefit-cost ratio than full-time child care and pre-K from birth to age 5.

The reason that one-year of pre-K has a higher bang for the buck than the earlier child care interventions is in part that child care has to have smaller group sizes than preschool to be high-quality. High-quality pre-K can have class sizes of 17 children to 2 teachers, or even 20 children to 2 teachers, class sizes that would be infeasible for younger age programs that aspire to high quality. Preschool at ages 3 and 4 appears to be an age range in which children are still malleable enough to significantly affect their future life-course, but children are old enough that they can be dealt with in large enough groups to significantly lower program costs per child.

On the other hand, if one truly wants very large effects on a child’s future life course, these earlier interventions can do more than just one year of pre-K. “Bang for the buck” is not the only criteria for evaluating programs. We also want to have large impacts.

Both pre-K and earlier interventions make sense. Pre-K is an extremely cost-effective approach to affecting a child’s future. But if we want to have truly large effects on bringing children out of poverty, we need to supplement pre-K with earlier high-quality interventions.

Posted in Early childhood program design issues | Comments Off on Earlier intervention has greater effects in changing a child’s life course, but not always a higher benefit-cost ratio; both are important

The Head Start experiment versus prior research on Head Start: the curious case of the changing control group

Posted on September 24, 2014 by timbartik

In my new book, From Preschool to Prosperity,  one issue about Head Start that I discuss is the change over time in results from studies of the effectiveness of Head Start. Earlier studies of Head Start find evidence of Head Start’s effectiveness, in both the short-run and long-run. (For examples, see studies by Deming, Currie et al., and Ludwig et al.) But the most recent Head Start experiment finds that Head Start’s short-term benefits quickly fade. How can these conflicting results be explained?

As mentioned in yesterday’s blog post, one explanation is that the Head Start experiment’s fading test score results need not imply a lack of long-run effects on adult outcomes. However, this explanation seems incomplete, as previous Head Start studies, such as Deming’s study, although they find some fading of test score effects, do not find test score effects to decline so drastically so quickly.

One plausible additional explanation is that there has been some change in the quality of the alternatives to Head Start. Studies of Head Start or any program are always comparing the program to whatever happens to the comparison or control group in the study. Many of the studies that find long-run effects of Head Start are looking at Head Start in the environment it operated in many years ago, back in the 1960s, 1970s, and 1980s. In those time periods, the comparison groups would have in most cases not been enrolled in high-quality pre-K programs. In contrast, in more recent time periods, children that did not enroll in Head Start are more likely to have been enrolled in a high-quality pre-K programs, such as the many state pre-K programs that have grown rapidly over the past 15 years.

There is some evidence for this hypothesis from the recent Head Start experiment. In the recent Head Start experiment, the randomly-assigned treatment group ended up having 80% of the children enrolled in Head Start. The randomly-assigned control group ended up with about half of its children enrolled in some pre-K program, 14 percent in Head Start, and 35 percent in some other pre-K program.

If some of the control group pre-K enrollment was in state pre-K programs that yielded higher test score effects than Head Start, this would reduce the relative test score advantages of Head Start. The Head Start experiment may not show that Head Start has no effects relative to no preschool, but rather that its effects are not large compared to alternative pre-K programs.

If this interpretation is correct, the Head Start experiment should not be interpreted as evidence that preschool is ineffective. Rather, the experiment should be interpreted as meaning that Head Start needs to improve, or at least needed to improve as of 2002-03,  to catch up with the quality of its best pre-K alternatives. And there have been attempts to improve Head Start relative to the 2002-03 time period studied in the experiment. In recent years, more literacy instruction has been pushed in Head Start. In addition, the federal government is evaluating the quality of local Head Start programs, and requiring recompetition for Head Start grants in cases where there are issues with local quality.

A 2002-2003 study that compares one preschool program, Head Start, with other preschool programs, does not show that all preschool at all times is ineffective compared to zero preschool. The case for preschool is that quality preschool programs can help children relative to zero preschool or low-quality preschool. The Head Start experiment does not provide strong evidence against the overall case for quality preschool.

Posted in Early childhood programs | Comments Off on The Head Start experiment versus prior research on Head Start: the curious case of the changing control group

Eye on Early Education reviews my new book, “From Preschool to Prosperity”

Posted on September 23, 2014 by timbartik

Alyssa Haywoode, of  “Eye on Early Education”, a blog of  the organization Strategies for Children, has a review and interview about my new book, From Preschool to Prosperity.

One accomplishment of  Ms. Haywoode in the interview was to push me a bit on some of the broader vision about early childhood education that I’m trying to outline in the book. As I said in the interview, “People [advocating for early childhood education] should see themselves as part of a historic movement”.  This expansion of access to high-quality early childhood education for all, if it is successful, will be seen in the future as one of the major historical advances in our educational system. These educational advances in turn will help our overall economy and society. Early childhood education does not solve all problems, but it can make a big dent in some of them.

Check out the review and interview here.

Posted in Uncategorized | Comments Off on Eye on Early Education reviews my new book, “From Preschool to Prosperity”

Preschool often has large adult effects even if test score effects fade somewhat.

Posted on September 23, 2014 by timbartik

My just-published book, From Preschool to Prosperity, includes a chapter discussing common criticisms of the research evidence for early childhood programs. One frequent criticism of the evidence for early childhood programs is based on results from the recent Head Start experiment. These results show that most test score effects for the Head Start treatment group, compared to the control group, have faded to statistical insignificance by the end of 3rd grade. Critics see this evidence from a random assignment experiment, the “gold standard” of research rigor, as suggesting that early childhood programs do not work today, at least as currently designed and run on a large scale.

But test score fading is common for many early childhood programs. It is often the case that early test score effects, at the end of preschool or the beginning of kindergarten, have faded quite a bit by 3rd grade. But in many of these cases, the program still has large effects on adult outcomes.

For example, for the Perry Preschool program, the end-of-program effects on test scores are large enough to predict that the program would increase adult earnings by 12%. But the 3rd grade test score effects of Perry would only predict adult earnings increases of 2 to 5%. The actual adult earnings impact of Perry is estimated to be 19%.

Similar fading of test score effects, but recovery of effects on adult outcomes, is found for the Chicago Child-Parent Center program, and the Abecedarian program. For all three of these programs (Perry, CPC, Abecedarian), the end of the program test score effects were a better predictor of adult earnings outcomes than the 3rd grade test scores.

What is going on here? The most plausible explanation is that adult earnings outcomes are affected by more than what is measured by standardized test scores, which mostly measure cognitive skills. But social skills and character skills also matter. It is these harder-to-measure skill effects that persist and contribute to important adult effects on educational attainment and earnings.

Cognitive effects of preschool matter. But they matter as part of a process at which higher cognitive skills at kindergarten entrance, along with social and character skills, lead to more learning in kindergarten, and so on in first grade and at later ages. Educational achievement and attainment, and adult success in the labor market, can go up by a lot even if long-run IQ test scores don’t change much. The initial cognitive skill effects of preschool are useful to monitor as a signal of broader effects of preschool on both cognitive and non-cognitive skills.

We certainly know from conversations with employers that when employers complain about worker skills, they are not just referring to cognitive skills, but also to social and character skills. A big part of worker skills is showing up consistently on time. Another big part of worker skills is getting along with supervisors, co-workers, and customers.

Just as there’s more to life than test scores, there’s more to preschool’s benefits than is captured by long-run cognitive test scores.

Posted in Uncategorized | Comments Off on Preschool often has large adult effects even if test score effects fade somewhat.

Early childhood programs work at a large scale

Posted on September 22, 2014 by timbartik

As I discuss in my just-published book, From Preschool to Prosperity, critics often argue that the research evidence for preschool only comes from small-scale programs run by researchers. Therefore, the argument goes, we don’t know if preschool can work if run at a large-scale by real-world government agencies.

But we actually do have good evidence that preschool can work at a large scale. As mentioned in a previous post, we have good evidence for the effectiveness of preschool in programs run at a large scale in Chicago, Boston, and Tulsa, and in many states. This evidence is not random assignment evidence – it would be quite difficult and expensive to run a random assignment experiment in a large-scale ongoing program run by a public agency. But this evidence comes from studies that have good comparison groups, with access to the program varying due to “natural experiments”, which argues that the estimated effects on adult outcomes and test scores represent true effects of the program, not pre-existing differences between program participants and non-participants.

The programs with good evidence are also programs run by a variety of public agencies. For example, in Chicago, Boston, and Tulsa, these preschool programs are run by the public schools in these cities. These preschools operate in diverse political, cultural, and economic situations, which suggests that preschool does not require special circumstances to be successful.

Preschool works. And we know this not just from small-scale “hothouse” experiments, but from large-scale implementation of preschool in many places across the U.S.

Posted in Early childhood programs | Comments Off on Early childhood programs work at a large scale

Early childhood programs today have credible evidence not only from random assignment experiments, but from “silver standard” studies with good comparison groups

Posted on September 19, 2014 by timbartik

My new book, From Preschool to Prosperity, argues that the research evidence for the effectiveness of early childhood programs is broad. The random assignment experiments, such as Perry Preschool, are often cited. Random assignment evidence is regarded as the “gold standard” for testing whether a program truly has causal effects on desired outcomes. But early childhood programs also have good evidence for effectiveness from many “silver standard” research studies.

The challenge in determining the true cause-and-effect relationships between programs and outcomes is unobserved characteristics of program participants. Program participants and non-participants will in general differ in both observed and unobserved characteristics. We can statistically control for the observed characteristics, but cannot do so for the unobserved characteristics. Therefore, it is hard to tell whether differences between program participants and non-participants in later outcomes are due to the program, or due to the pre-existing differences in unobserved characteristics.

For example, more ambitious families may be more likely to enroll their child in preschool. This will bias the study towards finding positive effects of preschool. On the other hand, perhaps families will choose to enroll children with more problems in preschool, which will bias studies towards finding negative effects of preschool.

Random assignment experiments solve this problem of unobserved characteristics by using some randomization procedure to select participants. This assures that on average, program participants and non-participants will have the same average unobserved characteristics. Differences in later outcomes are then more likely to be due to the program.

But in many cases randomization is not possible, or is too costly or cumbersome, or is ethically dubious. In these cases, we may find “natural experiments”, in which some feature of how program access is determined makes it likely that participants and non-participants are otherwise similar. These natural experiments provide “silver standard evidence” for the effects of early childhood programs on outcomes.

For early childhood programs, we have many studies that provide such “silver standard” evidence for the effectiveness of early childhood programs.  Studies of both Head Start and North Carolina’s “More at Four” preschool program provide evidence from geographic variation by county in access to the program that is arguably close to random.  Studies of the Chicago Child-Parent Center program use evidence from geographic variation in program access by neighborhood. Other studies of Head Start use evidence from variation in program usage within a family, comparing one sibling participating in Head Start versus another sibling who did not, which holds constant family factors affecting later outcomes.  Finally, many state pre-K studies compare test scores of children just entering pre-K, and children who have completed a year of pre-K and are entering kindergarten, who differ just slightly in age, with program participation varying because the children’s birthdates come just before or just after the age cut-off for entering kindergarten.

All these studies find evidence that early childhood programs have important effects on child outcomes. Because the participant and non-participant groups seem similar, this evidence is credible.

This evidence is “silver standard” because we cannot be utterly certain, for any particular study, that in fact there may not be some subtle differences between participants and non-participants that might bias results in one direction or another. A perfectly run random assignment experiment would provide better evidence, if such an experiment was readily available. But such random assignment evidence is not always available. And random assignment experiments are not always perfect – for example, in almost any random assignment experiment, there is some sample attrition, so that we lack data from some individuals in the participant and non-participant groups, and this sample attrition might also bias the evidence in one direction or another.

In the real world of social science or natural science, one study of one program rarely trumps all other studies. Each program is somewhat different, and each study has some imperfections. Rather, in determining the likely effects of a program, we need to see what pattern we find in a variety of more-or-less sound studies. For early childhood programs, the weight of the evidence points to these programs’ effectiveness in improving future adult outcomes for former child participants.

From Preschool to Prosperity is available for free as a pdf, for $0.99 on various e-book platforms, and is also available in hard-copy form.

Posted in Uncategorized | Comments Off on Early childhood programs today have credible evidence not only from random assignment experiments, but from “silver standard” studies with good comparison groups

High-quality random assignment experiments continue to provide part of the evidence for the economic payoff to early childhood education

Posted on September 18, 2014 by timbartik

My new book, From Preschool to Prosperity (available free as a pdf, for $0.99 on various e-book platforms, and also in hard copy form), explains why the research evidence for the effectiveness of early childhood education is credible.

Among this credible evidence is the oft-cited evidence from several well-done random assignment experiments. Despite arguments from some critics, such evidence remains relevant to today’s early childhood programs.

The most relevant random assignment experiments to today’s early childhood programs are Perry Preschool, the Abecedarian Program, and several experiments with the Nurse Family Partnership Program. Perry Preschool, run in Ypsilanti Michigan in the early 1960s, found evidence that half-day preschool at ages 3 and 4 increases adult earnings of former participants by 19%. The Abecedarian program was full-time child care and pre-K from birth to age 5, conducted in North Carolina in the 1970s. This program’s effects on adult outcomes predict an improvement in adult earnings of former participants of 26%.  Abecedarian also causes increases in earnings of parents of the child participants. The Nurse Family Partnership program, which provides one-on-one home visits by nurses to disadvantaged first-time mothers from the pre-natal period to age 2,  has been experimented with in various locations over the past 30 years. NFP’s  effects on outcomes for former participants predict an increase in adult earnings of 3%, and the program also produces significant earnings effects for the mothers.

Critics have argued that this research evidence is of limited relevance to today’s early childhood programs. The argument is that these experiments look at small-scale programs designed and run by researchers, and that it would be difficult or at least too expensive to duplicate these programs’ success today at a large-scale.

However, the Abecedarian program and the NFP program are both being run today, in some form. The current Educare program is designed similarly to the Abecedarian program. The NFP program is being sponsored around the U.S.  These programs are expensive, but they can apparently be run at a large scale in many locations. Because these programs have a high benefit-cost ratios, these programs’ high costs are outweighed by even higher benefits.

Furthermore, even Perry Preschool is more similar to today’s preschool programs than is sometimes realized. Perry used certified teachers paid public school wages. So do many successful preschool programs today, for example the programs in Tulsa and Boston. Perry was a half-day program at ages 3 and 4. Some programs today, such as the programs in Tulsa and Boston, are full-day programs at age 4, which has both advantages and disadvantages compared to Perry’s design. Among the advantages is that a full-day program finds it easier to elicit participation from a wide variety of families, because it does more to solve the child care problem for working parents. Perry had a well-designed curriculum, as do many of today’s preschool programs, for example the Boston Pre-K program has had success with the Building Blocks curriculum.

One advantage Perry Preschool had over current preschool programs was smaller class-sizes. Perry had class sizes that averaged 13 students to 2 teachers, whereas most preschool programs today would have larger class sizes, on the order of  15 to 20 students to 2 teachers.  Class size should make some difference to both benefits and costs of any educational program. Based on evidence from studies of class size in early elementary school, while lower class-size may pass a benefit-cost test, it seems unlikely that there is some dramatic differences in the relative cost-effectiveness of pre-K at a class size of 13 students versus a class-size of 15-20 students.

Early childhood education has more evidence for its effectiveness from random assignment experiments than does virtually any other social or educational program. We have more evidence for the effectiveness of early childhood programs than we do for the effectiveness of, for example, 3rd grade. With 3rd grade we can’t run a random assignment experiment that varies program access, because this intervention already has universal access. But we’ve run such experiments for early childhood programs, and found evidence of success for programs that are similar enough to today’s programs to be relevant to the current policy debate.

Posted in Early childhood programs | Tagged | Comments Off on High-quality random assignment experiments continue to provide part of the evidence for the economic payoff to early childhood education

Parental earnings benefits of early childhood programs shouldn’t be forgotten

Posted on September 17, 2014 by timbartik

One point made in my recently-published book From Preschool to Prosperity (available free as a pdf, and for $0.99 in various ebook formats, and also in paperback) is that parental earnings benefits are an important economic benefit for many early childhood programs.

For programs such as the Abecedarian program and Educare, which provide high-quality child care and pre-K from birth to age 5 for children from low-income families, the present value of earnings benefits for parents are actually about 50% greater than the earnings benefits for former child participants. The average present value of parental earnings benefits per family are over $190,000. These earnings benefits are in part short-term benefits from free child care enabling parents to work. But the free child care also enables parents to go back to school or obtain job training. The skills developed by the parent’s increased education or training, along with the skills developed by the parent’s increased work experience in the short-run, will help increase parental earnings in the long-term.

For the Nurse Family Partnership program and other home-visiting programs to improve parenting, in the real world an improvement in parenting is often linked to parents being empowered to improve their own lives. The available evidence suggests that this results in increased parent earnings, whose present value is similar to the positive effects of parenting programs on the future earnings of former child participants. The present value of increased maternal earnings per assisted first-time mother is over $15,000.

Parental earnings benefits are less important for programs providing only one year of preschool. This simply doesn’t provide enough free child care to significantly affect parental earnings. However, as my book details, one year of preschool is a very cost-effective way of increasing the future adult earnings of former child participants.

While early childhood programs should remain focused on the important goal of providing quality services that enhance the life prospects of the child, programs should not neglect a possible dual purpose of helping parents. From a standpoint of social benefits and costs, for many programs the benefits for parents can more than double the benefit-cost ratio.

Posted in Early childhood programs | Comments Off on Parental earnings benefits of early childhood programs shouldn’t be forgotten

Even under conservative estimates, early childhood education can do quite a bit to address income inequality

Posted on September 16, 2014 by timbartik

On September 16, the latest Census Bureau statistics on poverty and income distribution were released. Among other things, these latest stats showed a stagnant trend in relative income of different groups, based upon the Current Population Survey, which omits some incomes of very high-income groups. But this recent stagnant trend is in the context for a longer-term trend towards a much more unequal income distribution, with top incomes growing faster than income for the lowest income quintile or the middle-income quintile. (The lowest income quintile consists of households whose income puts them in the lowest fifth of all households; the middle-income quintile consists of households whose income puts them in the middle fifth of all households.)

As Conor Williams highlighted in his recent review, my new book, From Preschool to Prosperity, highlights that a full-scale commitment to a package of early childhood education proposals could make a major difference to the U.S. income distribution. Specifically, I estimate that a package of universal pre-K plus high-quality early child care for all low-income families could offset 5/6ths of the adverse income distribution trends of the last 30 years for the lowest income quintile, and 1/6ths of the adverse income distribution trends of the last 30 years for the middle-income quintile.

What I want to emphasize in this blog post is that my book’s estimates are quite conservative estimates. I am only counting estimates of the direct earnings effects of early childhood education on the earnings levels of former child participants. By raising skills, early childhood education can help both low-income children and middle-income children to increase their earnings as adults by large percentages, over 25 percent for the lowest income group, and about 5 percent for the middle-class.

But these estimates are deliberately quite conservative estimates that omit some possible additional effects. These estimates are conservative because it is hard to reliably quantify some of these additional effects based on the most rigorous possible research.

Some possible additional effects of early childhood education on the income distribution include:

  •  High-quality child care and parenting programs for low-income groups will boost parental earnings (I’ll have more to say about this effect tomorrow);
  • The increased worker skills for both former child participants, and parents, will have spillover effects on the earnings of others (I’ll have a blog post on this in the next few weeks, and my TED talk deals with this topic);
  • The increased earnings of former child participants will have second-generation effects on the earnings of their children;
  • My estimates assume that higher overall skills only affects productivity levels in the economy, but higher skills may also boost the overall rate of innovation and productivity growth in the economy.

It is hard to exactly quantify how all these effects will alter the American income distribution. But it seems likely that the boost to parental earnings will help the lowest income quintile the most, as will the second generation effects. In addition, the available evidence indicates that skill boosts tend to help lower-income workers in the economy the most. How changes in innovation will alter the economy is hard to say, but if all workers have more skills, one might think that technology and the economy might adapt to more fully utilize the skills of all workers.

In general, it is hard to completely model in any rigorous fashion all the social consequences of a full-scale commitment to early childhood education. What will happen in various neighborhoods if this investment leads to a drop in overall crime? How will political participation change if a higher percentage of all workers are more educated and better informed? One could go on.

What we can say is that even without some of these more profound effects of a full-scale investment in early childhood education, the direct effects we can quantify will have large benefits in helping make the income distribution more progressive. Early childhood education cannot by itself solve the income distribution problem. But a full-scale investment in high-quality versions of these early childhood programs can make a major difference in both growing our economy and more broadly distributing that growing economy’s wealth.

Posted in Distribution of benefits, Early childhood programs | Comments Off on Even under conservative estimates, early childhood education can do quite a bit to address income inequality

Conor Williams’s review of “From Preschool to Prosperity” in The Washington Monthly

Posted on September 16, 2014 by timbartik

Review is here, cross-posted at New America’s Ed Central. Conor Williams is a Senior Researcher in the Early Education Inititiative at the New America Foundation.

“Bartik bridges the distance between accessibility and expertise. From Preschool to Prosperity is supremely organized and readable, but it’s also jammed full of useful information. It’s the book for anyone who’s been following New York City’s pre-K expansion…and wants to know why it’s such a big deal. It’s the book for anyone who’s heard the … stats on the return on investing in pre-K…and finds themselves wondering just how researchers come up with those numbers.”

More here.

Posted in Early childhood programs | Comments Off on Conor Williams’s review of “From Preschool to Prosperity” in The Washington Monthly