PPN Home > Expert Perspectives > Head Start

Expert Perspectives

Child Policy Experts Answer Your Questions about Head Start

Children in Head Start program

What does research tell us about Head Start?   Where do we go from here?

The Head Start program is perhaps the most well known early childhood program in the United States. Despite hundreds of research studies conducted on the Head Start program over nearly half a century, controversy remains regarding the effectiveness of Head Start in achieving its multidimensional goal of promoting school readiness by providing educational, health, nutritional, social and other services to enrolled children and families. Recently published first grade follow-up results of the Head Start Impact Study have refueled this debate with the finding that Head Start participants achieve cognitive gains that fade out by the end of the first grade year.

This Expert Perspectives feature offered PPN visitors the opportunity to ask leading child policy experts their questions about the evidence related to Head Start.

Be sure to visit the More About this Topic section at the bottom of the page. You will find links to PPN's issue brief on Head Start, PPN's Head Start program summary, an updated program summary for Early Head Start, and more.

About the Experts

Janet Currie John Love Jens Ludwig
Janet Currie, PhD
Professor of Economics
Columbia University
About this expert
John Love, PhD
Senior Fellow
Mathematica Policy Research, Inc.
About this expert
Jens Ludwig, PhD
McCormick Foundation Professor of Social Service Administration, Law, and Public Policy
University of Chicago
About this expert

Question Topics

Below is a list of the question topics that were submitted to our experts. Follow the links below to read a particular question and answer, or scroll down the page to read them all.

Did the latest evaluation of Head Start find any program characteristics for which there was less fade-out of cognitive gains by the end of first grade?

Do you think that forces outside the Head Start classroom have an impact on the cognitive gains made by children who complete Head Start programs?

How might a future study be designed to determine the impact of the higher qualifications of today's Head Start teachers?

Has Head Start research examined the quality characteristics of K-5 classrooms as a factor in the fade-out effect?

Are there any successful evidence-based practices or programs used in Head Start that could be scaled up so that programs could be more consistent and prove lasting growth?

Questions and Answers

It's my understanding that the latest evaluation of Head Start showed "mixed" results in outcomes for children, i.e., fade-out of cognitive gains by the end of first grade. Did the evaluation find any program characteristics for which the fade-out was less? I am particularly interested if there was less fade-out in programs where there were higher requirements for teacher education and/or a defined curriculum to be followed.

John Love: This is an excellent question because researchers, policymakers, and program designers have become increasingly aware that some of the most important policy questions about creating effective programs for young children and families are not found by focusing only on the overall, whole-group analysis of program impacts. The Head Start Impact Study authors noted that "Congress mandated that the study identify circumstances under which the program achieves its greatest impact, in terms of both child and family circumstances—what works best for which children?" (Final Report, p. 2-59). [1] Thus, in addition to analyzing the overall impacts you correctly refer to, the authors also examined the program's impacts on seven subgroups, three defined by child characteristics (special needs [yes/no], pre-academic skills as assessed at baseline by the Woodcock-Johnson test [low/not low], home language [English/dual-language learner]) and four defined by parent/home characteristics (biological mother's race/ethnicity [white, African-American, Hispanic], household risk index [low/no, medium, high], urbanicity [urban/nonurban], and parent-reported depressive symptoms [none, mild, moderate, severe]). See the section on "Estimating Variations in Impact" and Figure 2.12 in the Final Report. [1]

These subgroup analyses found a number of favorable Head Start impacts at first grade in the cognitive, social-emotional, and health domains for both age cohorts and a few unfavorable impacts in the 3-year-old cohort, results that were not found when the average for the total sample was examined. One example is that Head Start resulted in African-American children showing less inattentiveness; fewer problems with structured learning, peer interactions, or teacher interactions; and better relationships with teachers than their control group counterparts. Another example is that Head Start children who entered as 4-year-olds in the lowest quartile of pre-academic skills had greater positive cognitive impacts than the children who scored higher at baseline, again in contrast with the comparable group of control group children.

These analyses provide strong evidence that Head Start is differentially effective for different types of children and families. The evidence is strong because the groups were defined by baseline characteristics, and those characteristics did not change over the course of the study (for either the Head Start or the control group children and families). Thus, the study provides true experimental evidence that the impacts for certain subgroups are substantially different from the overall study impacts. Unfortunately, the Head Start Impact Study authors have not provided us with any analyses of subgroups defined by program characteristics (such as the curriculum offered, teacher characteristics, part-day vs. full-day programs, and so forth). Such analyses would be very useful for policymakers, as programmatic decisions could be made about implementing particular curriculum approaches or hiring teachers with particular characteristics, including their educational qualifications, as you asked about. One basic message that your question implies is that one can conclude that Head Start had no lasting effects through first grade only if one ignores the important findings for key policy-relevant subgroups.

Back to topTop  

Do you think that forces outside the Head Start classroom have an impact on the cognitive gains made by children who complete Head Start programs? Some research indicates that the education level of the mother is the best predictor of a child's success. Have you looked at this correlation? What about social-emotional issues?

Janet Currie: Forces outside Head Start definitely have an impact on the gains made by children who complete Head Start programs. One study that Duncan Thomas and I conducted of children who participated in the Head Start program several decades ago found that, while white and African-American children both experience test score gains with Head Start, gains were more likely to "fade out" for African-American children. This is likely due to the fact that African-American Head Start children tend to subsequently attend the worst schools. In contrast, white Head Start children tend to go on to attend schools of average quality. [2]

Maternal education is also a very important determinant of child outcomes. Mothers with more education have children of higher birth weight, are more likely to read to their children, and are better able to support them throughout their schooling. This may be one reason why Head Start does not fully close the gap between disadvantaged children and other children.

However, Head Start has been shown to have lasting positive effects on children, even in the absence of permanent test score gains. For example, Head Start children are more likely to go to college than siblings who did not attend, and they are less likely to have been booked or charged with a crime by young adulthood. It is possible that these gains reflect better emotional self-regulation that is a result of the Head Start intervention.

Jens Ludwig: It seems useful to distinguish between those baseline characteristics of families that predict better children's academic or social-emotional outcomes, versus those baseline characteristics that help predict which children will benefit the most from some social program (that is, be most responsive to the intervention). One reason that we target Head Start to the most disadvantaged children is because we know from past research that, on average, their academic and social-emotional outcomes tend to be somewhat lower than those of more affluent children, while at the same time we also hypothesize that disadvantaged children may benefit more from program participation.

Knowing more about which specific sub-populations are in fact most responsive to social program participation is extremely important for targeting scarce social policy resources, and can in some cases also provide insights into the underlying mechanisms or ways in which programs improve people's life chances. Although previous studies have not consistently collected background data on the schooling attainment or other socio-economic characteristics of mothers and families of Head Start children, many studies do report results separately for children from different racial/ethnic groups.

In this light, it is interesting that the recent Head Start experiment shows some signs that academic achievement impacts persist for African-American children but not for white or Hispanic children. To generate this result, I took the subgroup impact estimates for the years 2003 through 2006, available at http://www.acf.hhs.gov/programs/opre/hs/impact_study/. I focused on the children who were four years old at baseline; the results for children who were age three at baseline are harder to think about because such a large share of the control group winds up enrolling in Head Start during the second year of the study period. I then averaged together all the cognitive outcome estimates reported in the tables each year, focusing on the version of the impact estimates that are reported in standard deviation form (known in the evaluation literature as effect sizes). For African-American children, the estimated impact at the end of the first year of the study was equal to around 0.28 standard deviations, and this impact was still around 0.024 standard deviations at the end of first grade. In contrast, the first-year impact for Hispanic children was around 0.23 standard deviations but declined to around -0.01 standard deviations by the end of first grade, while the impact for white children was around 0.1 standard deviations the first year and down to around 0.01 standard deviations by the end of the first year. Similar findings of relatively larger test score impacts for African-American children come from the recent Early Head Start experiment, also sponsored by the Department of Health and Human Services, [3] as well as the widely cited Tennessee STAR experiment. [4]

While this pattern is less true among earlier cohorts of children, as noted by the Currie and Thomas study mentioned above, many recent studies of Head Start and other social programs suggest that children whose family background characteristics are predictive of lower achievement levels—that is, relatively more disadvantaged children, such as African-Americans—seem to benefit relatively more from these interventions.

Back to topTop  

Since the Impact Study relied on outcome data from several years ago (in the early 2000s), when fewer Head Start teachers held bachelor's degrees and specialized certificates, I wonder whether the outcome data might be different if collected today. How might a study be designed to determine the impact of the higher qualifications of today's Head Start teachers?

Janet Currie: When considering the Impact Study, it is important to keep in mind that most children in the control group participated in some other type of preschool. This is in contrast to previous evaluations of Head Start, in which Head Start children were compared against children who did not attend preschool.

Model preschool programs that have been shown to have large effects on children's outcomes typically employ teachers with a high degree of training (e.g., a master's degree). But they also generally have a particular curriculum, small group sizes, a high degree of oversight, and other interventions (such as home-visitation components). The state of public schools in many jurisdictions should make us cautious about viewing higher formal teacher qualifications by themselves as a panacea for school programs.

One way to evaluate the impact of more-highly-qualified teachers would be to run a randomized trial. This could be done by asking programs in need of additional staff to submit applications for funds that would allow them to hire people at higher wages, then choosing the best applications and randomly picking some of them to receive the funding. Alternatively, it would be helpful to be able to identify high-performing and low-performing Head Start centers and study differences between them. This would, however, require tracking of child outcomes in order to be able to rank the performance of different centers.

John Love: You are correct that the results of the Head Start Impact Study apply to the program as it existed at the time the sample children were enrolled—that is, 2002-2003 for 4-year-old children and 2002-2003 and 2003-2004 for 3-year-olds. If Head Start programs have changed substantially in the past seven or eight years in ways that might make a difference in the impacts they achieve, then the findings reported in the Head Start Impact Study may no longer hold up. Teacher qualification is one program characteristic worth examining for possible change, because Congress has mandated increasing the percentage of teachers with bachelor's degrees: As the report notes, "The current Head Start reauthorization act ([5]) requires that by 2013, at least 50 percent of Head Start teachers nationwide in center-based programs have at least (1) a bachelor's degree in early childhood education (ECE), or (2) a bachelor's degree and coursework equivalent to an ECE major plus experience in teaching preschool children," Final Report, p. 2-42. [1]

In the representative sample of programs included in the Head Start Impact Study, 30.8 percent of teachers held a bachelor's or higher degree (Final Report, Exhibit 3.13, p. 3-18). [1] Head Start has also been conducting descriptive studies (that is, studies of the Head Start program itself without any control or comparison groups). Known as Head Start FACES (the Family and Child Experiences Survey), these studies have been conducted every three years since 1997. FACES data are based on nationally representative samples, although the samples are of course different from the sample of programs that participated in the impact study. According to a FACES 2003 report, 37.8 percent of Head Start teachers had a bachelor's or higher degree. [6] FACES 2006 reported that the percentage had increased to 39.7. [7]

It's hard to say whether an increase of 9 percentage points would make a big difference in Head Start's impacts, but the changes reported by the FACES studies suggest that, in the past seven or eight years, Head Start has experienced about a 30 percent increase in the percentage of teachers meeting the new standard. One reason that it is difficult to know what effect this has is that the research on teacher qualifications does not yield clear results. Over the years, considerable research has been devoted to trying to learn the importance of teachers' credentials for children's developmental progress. Some studies have shown that when teachers have a bachelor's degree, the children perform better. However, recent studies suggest that we should be cautious about that conclusion. Margaret Burchinal, Marilou Hyson, and Marty Zaslow presented a comprehensive review of this literature at Head Start's 2008 National Research Conference. [8] Among other research they discussed, they cited a study that analyzed data from seven large-scale preschool studies and concluded that there was "no consistent pattern of association between any index of teacher education and either classroom quality or child outcomes." [9] Thus, even though the percentage of teachers with a bachelor's degree has increased since fall 2002, it is not certain that the program impacts have changed.

A number of research designs could be implemented to investigate the impact of Head Start having a higher percentage of teachers with bachelor's degrees, short of repeating the very expensive impact study. Perhaps the most useful design would be one that selected a sample of Head Start programs and experimentally provided incentives for a random half of them to hire teachers with bachelor's degrees (or support current teachers to obtain additional education). Comparing the gains children make from fall to spring during a year subsequent to the program enhancement would tell us whether the increase in credentialing made a difference for children's school readiness.

Jens Ludwig: It's natural to think that the impacts of Head Start must be increasing over time, since the quality of the program seems to be getting better. But keep in mind that the effect of the program on children is the difference between the outcomes they experience from enrolling in Head Start versus what outcomes they would have experienced if they had not been in Head Start. So, while Head Start has gotten better over time, the alternatives to Head Start for poor children have also changed over time. For example just over the past few years, more and more states have been supporting universal pre-K programs. [10] Whether Head Start's impacts on poor children are growing or shrinking over time depends on this "horse race" between the improvements in Head Start versus the changes or improvements in the developmental quality of the environments that poor children experience if they don't attend Head Start.

I was very surprised to see that the estimated effects of Head Start on children in the recent Head Start Impact Study—which, as noted above, are on the order of about one-quarter of a standard deviation at the end of the first year for those who were four years old at baseline—are of about the same magnitude as what researchers such as Janet Currie and David Deming have estimated for cohorts of poor children who participated in Head Start in the 1970s and 1980s.

I definitely agree with the reader's question that it would be really useful to know more about the role of the different "active ingredients" in Head Start, in order to better understand what aspects of the program we should devote extra funding to and prioritize for improvement and strengthening. Janet Currie and Matthew Neidell have done some very nice work along these lines, published in the Economics of Education Review. [11] I think they would agree that better still would be for the federal government to explicitly fund some prospective randomized experiments to test this question, which could be done by doing something like the equivalent of the Tennessee STAR experiment for Head Start. Two of the key cost drivers for any educational program are class size (or, more accurately, student-staff ratios), and average teacher salaries, which tend to be related to things such as teacher educational attainment. To learn more about the value of pouring money into these different cost drivers, why not randomly assign Head Start children to large vs. small classrooms, and randomly assign teachers with different levels of experience and educational attainment to large vs. small classrooms as well?

Back to topTop  

Has Head Start research examined the quality characteristics of K-5 classrooms as a factor in the fade-out effect? Could you address the expectation that Head Start outcomes should sustain even if the child receives poor quality instruction in grades K-5?

Janet Currie: This is a question that should receive more attention. One study examined the quality of schools attended by Head Start children as of 8th grade. It found that African-American Head Start children went on to attend schools of very poor quality, as measured by average test scores, but that this was not true among white Head Start children. [2] It seems unreasonable to expect Head Start to inoculate children against future problems if they are subsequently exposed to the worst-quality schools.

John Love: The Impact Study examined both school- and classroom-level characteristics one might associate with quality of instruction in the kindergarten and first grade classes that the sample children attended after Head Start. For the 4-year-old cohort, although the schools the study children attended had higher percentages of students living in poverty than the average of schools across the country, the schools attended by children from the Head Start group were not different from the schools attended by control group children. This was true for three demographic characteristics: public vs. private school, percentage of students eligible for free- or reduced-price lunch, and percentage of various racial/ethic groups. It was also true for average student achievement in the schools, as measured by performance on state math and reading assessments (Final Report, p. 3-37 to 3-39). [1]

At the teacher and classroom level, however, the study compared about 40 teacher and classroom characteristics of the schools attended by children from the Head Start and control group (Exhibit 3.22). Only two were significantly different: In kindergarten, former Head Start children were less likely to be in classes with an adult volunteer in the classroom than the control group, and, in first grade, the Head Start group was less likely to be in classes with a paid assistant, co-teacher, or volunteer. There were somewhat more treatment-control differences for the 3-year-old cohort, but still not many. The authors acknowledge that their attempt to compare the quality of the classrooms and schools of the two study groups was imperfect: "Thus, the measures collected as part of this study provide an incomplete picture of what constitutes the total early childhood experiences of the study children. A host of factors could not be measured, and some of the variables that are discussed in this chapter are only proxies for children's educational and developmental experiences" (Final Report, p. 3-53). [1] I think we have to conclude that the Head Start impact study itself did not give sufficient information about the quality of the schools children attended after leaving Head Start for us to know—in this case—whether children's post-Head Start experiences might be responsible for "fade-out." For that, we need to turn to other studies, as Currie and Ludwig suggest.

Jens Ludwig: No one would deny that a child's elementary and secondary school experiences surely need to be of some basic minimal quality in order to preserve the benefits of even the best early childhood intervention. Because Janet Currie has done the empirical work on this question I will let her take the lead in answering this one.

However, I would like to make a different point that might complicate people's intuition about the role of school quality, which is that right now we're not really sure whether what we are seeing is "fade-out" of Head Start's impact on the experimental group versus "catch-up" by the Head Start control group. From observing the experiences of my wife, who was a preschool special education teacher in Northern Virginia, and my six-year-old daughter, who is currently in kindergarten, it's clear that good teachers often try to identify the specific areas that some children are struggling with and then provide them with some extra attention and assistance to help them catch up with the rest of the class. It might be the case that low-income children who don't have a chance to enroll in Head Start show up for kindergarten or first grade slightly behind the Head Start alumni and other children in the classroom, and that the better teachers are the ones who are able to identify these skill lags and remedy them. Put differently, in principle, it could be the case that school quality above some minimum threshold level is positively rather than negatively correlated with the attenuation of differences in average outcomes between Head Start and other children, which in this case could be "catch-up," not "fade-out." I don't know whether this is actually true or not, but I think this would be a very important hypothesis to test once the Department of Health and Human Services makes the data from the Head Start experiment publicly available.

Back to topTop  

Are there any evidence-based practices or programs that are being used in Head Start or the Early On Program that are successful and could be scaled up so that programs could be more consistent and prove lasting growth?

Janet Currie: The Head Start Bureau sponsors the Head Start Quality Research Center (QRC) consortium, which is charged with developing and evaluating curricular and other improvements. The consortium currently is made up of eight research institutions and the Head Start Quality Research Center. However, the QRC does not currently evaluate new interventions in the most rigorous way possible, i.e., using a randomized experiment in which "treatment" centers that have implemented the innovation are compared with "control" centers that have not. Moreover, it does not systematically collect data on child outcomes that would allow one to compare outcomes across centers and identify high-performing vs. low-performing centers. This makes it more difficult to identify promising practices for dissemination.

Back to topTop  

More About This Topic

Listed below are resources with more information on Head Start.

Head Start:  What Do We Know? — Mar. 2010

Head Start Program Summary

Early Head Start Program Summary

Head Start Impact Study and Follow-Up — Jan. 2010

Head Start Program Fact Sheet — 2008

A National Survey of Obesity Prevention Practices in Head Start — Dec. 2009

Reaching Out to Kith and Kin Caregivers in Early Head Start — Apr. 2006

Head Start Impact Study: First Year Findings — Jun. 2005

Still Going Strong: Head Start Children, Families, Staff, and Programs in 2004 — Nov. 2005


Listed below are sources that are cited in the experts' answers above.

1.   U.S. Department of Health and Human Services, Administration for Children and Families, Office of Planning, Research and Evaluation, Head Start Impact Study: Final Report, Washington, D.C., 2010.

2.   Currie, J., and D. Thomas, "School Quality and the Longer-Term Effects of Head Start," Journal of Human Resources, Vol. 35, No. 4, 2000, pp. 755-774.

3.   U.S. Department of Health and Human Services, Administration for Children and Families, Office of Planning, Research, and Evaluation, Making a Difference in the Lives of Infants and Toddlers and Their Families: The Impacts of Early Head Start, Washington, D.C., 2002.

4.   Krueger, A. B., Experimental Estimates of Education Production Functions, Cambridge, Mass.: National Bureau of Economic Research, Inc., 1997.

5.   Public Law 110-134, Improving Head Start for School Readiness Act of 2007, 2007.

6.   U.S. Department of Health and Human Services, Administration for Children and Families, FACES Findings: New Research on Head Start Outcomes and Program Quality, Washington, D.C., 2006.

7.   West, Jerry, Louisa Tarullo, Nikki Aikens, and Lara Hulsey, Study Design and Data Tables for FACES 2006 Baseline Report, Washington, D.C.: U.S. Department of Health and Human Services, Office of Planning, Research, and Evaluation, 2008.

8.   Burchinal, M., M. Hyson, and M. Zaslow, "Competencies and Credentials for Early Childhood Educators: What Do We Know and What Do We Need to Know?" paper presented at Head Start's Ninth National Research Conference, Washington, D.C., June 23-25, 2008.

9.   Early, D. M., et al., "Teachers' Education, Classroom Quality, and Young Children's Academic Skills: Results From Seven Studies of Preschool Programs," Child Development, Vol. 78, 2007, pp. 558-580.

10.   National Institute for Early Education Research, State Preschool Yearbooks, various years. As of May 5, 2010: http://nieer.org/docs/index.php?DocID=131.

11.   Currie, Janet, and Matthew Neidell, "Getting Inside the 'Black Box' of Head Start Quality: What Matters and What Doesn't?" Economics of Education Review, Vol. 26, pp. 83-89, February 2007.