Investing in Preschool Programs

8/12/2019 Investing in Preschool Programs

1/24

Journal of Economic PerspectivesVolume 27, Number 2Spring 2013Pages 109132

At the beginning of kindergarten, the math and reading achievement gapst the beginning of kindergarten, the math and reading achievement gapsbetween children in the bottom and top income quintiles amount to morebetween children in the bottom and top income quintiles amount to morethan a full standard deviation. Early childhood education programs providethan a full standard deviation. Early childhood education programs providechild care services and may facilitate the labor market careers of parents, but theirchild care services and may facilitate the labor market careers of parents, but theirgreatest potential value is as a human capital investment in young children, particu-greatest potential value is as a human capital investment in young children, particu-larly children from economically disadvantaged families (Heckman 2006). After all,larly children from economically disadvantaged families (Heckman 2006). After all,both human and animal studies highlight the critical importance of experiences inboth human and animal studies highlight the critical importance of experiences inthe earliest years of life for establishing the brain architecture that will shape futurethe earliest years of life for establishing the brain architecture that will shape futurecognitive, social, and emotional development, as well as physical and mental healthcognitive, social, and emotional development, as well as physical and mental health(Sapolsky 2004; Knudsen, Heckman, Cameron, and Shonkoff 2006). Moreover,(Sapolsky 2004; Knudsen, Heckman, Cameron, and Shonkoff 2006). Moreover,research on the malleability (plasticity) of cognitive abilities nds these skills to beresearch on the malleability (plasticity) of cognitive abilities nds these skills to behighly responsive to environmental enrichment during the early childhood periodhighly responsive to environmental enrichment during the early childhood period(Nelson and Sheridan 2011). Perhaps early childhood education programs can be(Nelson and Sheridan 2011). Perhaps early childhood education programs can bedesigned to provide the kinds of enrichment that low-income children most needdesigned to provide the kinds of enrichment that low-income children most needto do well in school and succeed in the labor market.to do well in school and succeed in the labor market.We summarize the available evidence on the extent to which expenditures onWe summarize the available evidence on the extent to which expenditures onearly childhood education programs constitute worthy social investments in theearly childhood education programs constitute worthy social investments in thehuman capital of children. We begin with a short overview of existing early child-human capital of children. We begin with a short overview of existing early child-hood education programs, and then summarize results from a substantial body ofhood education programs, and then summarize results from a substantial body ofmethodologically sound evaluations of the impacts of early childhood education. Wemethodologically sound evaluations of the impacts of early childhood education. Wend that the evidence supports few unquali ed conclusions. Many early childhoodnd that the evidence supports few unqualied conclusions. Many early childhood

Investing in Preschool Programs

Greg J. Duncan is Distinguished Professor in the School of Education, University of California,Irvine. Katherine Magnuson is Associate Professor of Social Work, University of Wisconsin

Madison. Their email addresses are [email protected] [email protected]. To access the Appendix and disclosure statements, visithttp://dx.doi.org/10.1257/jep.27.2.109. doi=10.1257/jep.27.2.109

Greg J. Duncan and Katherine Magnuson
mailto:[email protected]:[email protected]://dx.doi.org/10.1257/jep.27.2.109http://dx.doi.org/10.1257/jep.27.2.109http://dx.doi.org/10.1257/jep.27.2.109http://dx.doi.org/10.1257/jep.27.2.109mailto:[email protected]:[email protected]


2/24

110 Journal of Economic Perspectives

education programs appear to boost cognitive ability and early school achievementeducation programs appear to boost cognitive ability and early school achievementin the short run. However, most of them show smaller impacts than those generatedin the short run. However, most of them show smaller impacts than those generatedby the best-known programs, and their cognitive impacts largely disappear within aby the best-known programs, and their cognitive impacts largely disappear within afew years. Despite this fade-out, long-run follow-ups from a handful of well-knownfew years. Despite this fade-out, long-run follow-ups from a handful of well-knownprograms show lasting positive effects on such outcomes as greater educationalprograms show lasting positive effects on such outcomes as greater educationalattainment, higher earnings, and lower rates of crime. Since ndings regardingattainment, higher earnings, and lower rates of crime. Since ndings regardingshort and longer-run impacts on noncognitive outcomes are mixed, it is uncer-short and longer-run impacts on noncognitive outcomes are mixed, it is uncer-tain what skills, behaviors, or developmental processes are particularly important intain what skills, behaviors, or developmental processes are particularly important inproducing these longer-run impacts.producing these longer-run impacts.Our review also describes different models of human development used by socialOur review also describes different models of human development used by socialscientists, examines heterogeneous results across groups, and tries to identify thescientists, examines heterogeneous results across groups, and tries to identify theingredients of early childhood education programs that are most likely to improveingredients of early childhood education programs that are most likely to improvethe performance of these programs. We use the terms early childhood educationthe performance of these programs. We use the terms early childhood educationand preschool interchangeably to denote the subset of programs that provide group-and preschool interchangeably to denote the subset of programs that provide group-based care in a center setting and offer some kind of developmental and educationalbased care in a center setting and offer some kind of developmental and educationalfocus. This de nition is intentionally broad, as historical distinctions between earlyfocus. This denition is intentionally broad, as historical distinctions between earlyeducation and other kinds of center-based child care programs have blurred. Manyeducation and other kinds of center-based child care programs have blurred. Manyearly education programs now claim the dual goals of supporting working families andearly education programs now claim the dual goals of supporting working families andproviding enriched learning environments to children, while many child care centersproviding enriched learning environments to children, while many child care centersalso foster early learning and development (Adams and Rohacek 2002).also foster early learning and development (Adams and Rohacek 2002).

Existing Preschool Programs

Most children enrolled in early childhood education attend private programs,Most children enrolled in early childhood education attend private programs,some nonpro t and others for-pro t. In 2011, the average cost of full-time, center-some nonprot and others for-prot. In 2011, the average cost of full-time, center-based care for a four-year old ranged from $3,900 in Mississippi to just over $14,000 inbased care for a four-year old ranged from $3,900 in Mississippi to just over $14,000 inthe District of Columbia (National Association of Child Care Resource and Referralthe District of Columbia (National Association of Child Care Resource and ReferralAgencies 2012). Given the high cost of care, it is unsurprising that enrollment rates ofAgencies 2012). Given the high cost of care, it is unsurprising that enrollment rates ofchildren residing in families with incomes in the bottom half of the income distribu-children residing in families with incomes in the bottom half of the income distribu-tion are persistently 10 20 percentage points lower than for children in the highesttion are persistently 10 20 percentage points lower than for children in the highestquarter. Figure 1, based on the data from the October Supplement to the Currentquarter. Figure 1, based on the data from the October Supplement to the CurrentPopulation Survey, shows this enrollment gap by income level. The gure also showsPopulation Survey, shows this enrollment gap by income level. The gure also showsa steady rise in enrollment in early childhood education programs among three - anda steady rise in enrollment in early childhood education programs among three - andfour year-olds over the past 40-some years. This increase is broad-based, across incomefour - year-olds over the past 40-some years. This increase is broad-based, across incomegroups and for the children of both employed and nonemployed mothers.groups and for the children of both employed and nonemployed mothers.States and the federal government have sought to increase the participationStates and the federal government have sought to increase the participationof low-income children in early childhood education programs in a number ofof low-income children in early childhood education programs in a number ofways: through Head Start, pre-kindergarten programs, and means-tested child careways: through Head Start, pre-kindergarten programs, and means-tested child careassistance programs that can be used to pay for center-based care.assistance programs that can be used to pay for center-based care.11Overall, bothOverall, bothfederal and state investments in these programs increased substantially in real termsfederal and state investments in these programs increased substantially in real terms1 The federal government also provides some nancial assistance to families seeking child care via theChild and Dependent Care Tax Credit as well as exclusions from income for benets under dependent


3/24

Greg J. Duncan and Katherine Magnuson 111

through the early 2000s, but in more recent years funding has not grown substan-through the early 2000s, but in more recent years funding has not grown substan-tially (Barnett, Carolan, Fitzgerald, and Squires 2011; Magnuson and Shager 2010;tially (Barnett, Carolan, Fitzgerald, and Squires 2011; Magnuson and Shager 2010;Schulman and Blank 2012).Schulman and Blank 2012).Head Start, the federal governments largest compensatory preschool program,, the federal governments largest compensatory preschool program,

is designed to enhance childrens social and cognitive development by providing ais designed to enhance childrens social and cognitive development by providing acomprehensive set of educational, health, nutritional, and other social services. Incomprehensive set of educational, health, nutritional, and other social services. In2005, virtually all Head Start programs were center-based and half offered full-day2005, virtually all Head Start programs were center-based and half offered full-day(six hours or more) services, ve days a week (Hamm 2006). Most children enrolled(six hours or more) services, ve days a week (Hamm 2006). Most children enrolledin Head Start in 2009 were three (36 percent) or four years old (51 percent). Inin Head Start in 2009 were three (36 percent) or four years old (51 percent). In2010, the federal Head Start appropriation of about $7.2 billion was distributed2010, the federal Head Start appropriation of about $7.2 billion was distributedto 1,591 local private and public nonpro t grantees serving 904,153 children.to 1,591 local private and public nonprot grantees serving 904,153 children.Some states supplement federal funds to increase access to Head Start programs;Some states supplement federal funds to increase access to Head Start programs;for details, see the Head Start website at http://eclkc.ohs.acf.hhs.gov/hslc/mrfor details, see the Head Start website at http://eclkc.ohs.acf.hhs.gov/hslc/mrcare assistance programs; however, few low-income families benet from these programs (Forry andSorenson 2006; Magnuson, Meyers, and Waldfogel 2007).

Figure 1

Percent of Three- and Four-year-olds Enrolled in Preschool by Family Income

Quartile

Source:Authors using data from the October Current Population Survey.

Notes: Data represent three-year moving averages. Parents report on whether the child attends regularschool. The line break in 1994 corresponds to the addition of a question prompt, which dened regularschool as including nursery school, kindergarten or elementary school . . . See Magnuson, Meyers, and

Waldfogel (2007) for further discussion of how the Current Population Survey compares with other sourcesof data on preschool enrollment.

Percentat

tendingpreschool

0

1970 2010

20

40

60

80

1980 1990 2000

Income quartile 1

Income quartile 2

Income quartile 3

Income quartile 4
http://eclkc.ohs.acf.hhs.gov/hslc/mr/factsheets/fHeadStartProgr.htmhttp://eclkc.ohs.acf.hhs.gov/hslc/mr/factsheets/fHeadStartProgr.htmhttp://eclkc.ohs.acf.hhs.gov/hslc/mr/factsheets/fHeadStartProgr.htm


4/24


/factsheets/fHeadStartProgr.htm. Local grantees are required to provide at least/factsheets/fHeadStartProgr.htm. Local grantees are required to provide at least20 percent matching funds. All this brings program costs to around $9,000 per child20 percent matching funds. All this brings program costs to around $9,000 per childper year (Ludwig and Phillips 2007).per year (Ludwig and Phillips 2007).Pre-kindergarten programsare funded primarily by states or local school districts.are funded primarily by states or local school districts.In 2011, 39 states and the District of Columbia spent about $5.5 billion on pre-In 2011, 39 states and the District of Columbia spent about $5.5 billion on pre-kindergarten initiatives that collectively served approximately 28 percent of thekindergarten initiatives that collectively served approximately 28 percent of thenations four-year-olds and 4 percent of three-year-olds (for details, see Barnett,nations four-year-olds and 4 percent of three-year-olds (for details, see Barnett,Carolan, Fitzgerald, and Squires 2011). Most pre-kindergarten programs targetCarolan, Fitzgerald, and Squires 2011). Most pre-kindergarten programs targetlow-income children (31 state programs have income eligibility requirements), andlow-income children (31 state programs have income eligibility requirements), andmost offer health, vision, and hearing screenings as well as at least one other formmost offer health, vision, and hearing screenings as well as at least one other formof support service. One-half of state pre-kindergarten programs require teachers toof support service. One-half of state pre-kindergarten programs require teachers tohave training in early child development and nearly one-third require BA degrees.have training in early child development and nearly one-third require BA degrees.

Typically, states use a mixed service delivery system that provides programming inTypically, states use a mixed service delivery system that provides programming inlocal elementary schools as well as community-based settings.local elementary schools as well as community-based settings.With expenditures in 2010 amounting to approximately $9.5 billion, federalWith expenditures in 2010 amounting to approximately $9.5 billion, federaland state-fundedand state-funded means-tested child care subsidies can be used for various typescan be used for various typesof child care, including center-based care, family day care, and other forms ofof child care, including center-based care, family day care, and other forms ofinformal care, and they cover a wide age range of children (birth through age 12).informal care, and they cover a wide age range of children (birth through age 12).Their primary goal has continued to be supporting working families rather thanTheir primary goal has continued to be supporting working families rather thaneducating young children, although increased spending on subsidies has beeneducating young children, although increased spending on subsidies has beenlinked to higher rates of preschool attendance among young children (Magnuson,linked to higher rates of preschool attendance among young children (Magnuson,Meyers, and Waldfogel 2007). Because parents preferences and needs for childMeyers, and Waldfogel 2007). Because parents preferences and needs for childcare may not always align well with what is provided by preschool programs, andcare may not always align well with what is provided by preschool programs, andbecause child care subsidy spells are often quite short (Ha, Magnuson, and Ybarrabecause child care subsidy spells are often quite short (Ha, Magnuson, and Ybarra2012), these subsidies are best viewed as an indirect way to promote early child-2012), these subsidies are best viewed as an indirect way to promote early child-hood education for three- and four-year-olds.hood education for three- and four-year-olds.

Empirical Studies of the Effectiveness of Early Childhood Education

Empirical studies of the effects of investments in early childhood educationEmpirical studies of the effects of investments in early childhood educationon childrens human capital encompass a range of methodologies and a wideon childrens human capital encompass a range of methodologies and a widevariety of programs. We focus on evaluations of preschool programs conductedvariety of programs. We focus on evaluations of preschool programs conductedover the course of the last half-century that are based on strong experimentalover the course of the last half-century that are based on strong experimentalor quasi-experimental methods and provide impact estimates for cognitive oror quasi-experimental methods and provide impact estimates for cognitive orachievement-related outcomes.achievement-related outcomes.22Despite the hundreds of evaluation studies of earlyDespite the hundreds of evaluation studies of earlychildhood education programs that have been published over the past 50 years,childhood education programs that have been published over the past 50 years,2A full list of these studies appears in the online appendix available with this paper at http://ejep.org.

As described there, programs selected for our analysis had both treatment and control/comparisongroups, included at least 10 participants in each condition, incurred less than 50 percent attrition, andmeasured childrens cognitive development close to the end of their treatment programs. Studies had

to have used random assignment or one of the following quasi-experimental designs: change models,xed effects modes, regression discontinuity, difference in difference, propensity score matching,interrupted time series, instrumental variables, and some other types of matching. Studies that used
http://eclkc.ohs.acf.hhs.gov/hslc/mr/factsheets/fHeadStartProgr.htmhttp://eclkc.ohs.acf.hhs.gov/hslc/mr/factsheets/fHeadStartProgr.htmhttp://eclkc.ohs.acf.hhs.gov/hslc/mr/factsheets/fHeadStartProgr.htm


5/24

Investing in Preschool Programs 113

a handful of programs have gured especially prominently in policy discussions:a handful of programs have gured especially prominently in policy discussions:in particular, Perry Preschool, the Abecedarianin particular, Perry Preschool, the Abecedarian33program, Head Start, and moreprogram, Head Start, and morerecently some state and local pre-kindergarten programs.recently some state and local pre-kindergarten programs.Meta-AnalysisFigure 2shows the distribution of 84 program-average treatment effect sizes forFigure 2shows the distribution of 84 program-average treatment effect sizes forcognitive and achievement outcomes, measured at the end of each programs treat-cognitive and achievement outcomes, measured at the end of each programs treat-ment period, by the calendar year in which the program began. Reecting theirment period, by the calendar year in which the program began. Reecting theirapproximate contributions to weighted results, bubble sizes are proportional toapproximate contributions to weighted results, bubble sizes are proportional tothe inverse of the squared standard error of the estimated program impact. Thethe inverse of the squared standard error of the estimated program impact. Thegure differentiates between evaluations of Head Start and other early childhoodgure differentiates between evaluations of Head Start and other early childhood

quasi-experimental designs must have had pre- and post-test information on the outcome or establishedbaseline equivalence of groups on demographic characteristics determined by a joint test.3 Abecedarian can mean one who is learning the alphabet.

Figure 2

Average Impact of Early Child Care Programs at End of Treatment

(standard deviation units)

Source:Authors.Notes: Figure 2 shows the distribution of 84 program-average treatment effect sizes for cognitive andachievement outcomes, measured at the end of each programs treatment period, by the calendar year in

which the program began. Reecting their approximate contributions to weighted results, bubble sizesare proportional to the inverse of the squared standard error of the estimated program impact. There isa weighted regression line of effect size by calendar year.

0.50

0.00

0.50

1.00

1.50

2.00

1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010

Averageeffectsizeinsdunits

Head Start

NonHead Start

Perry

National Head Start

Abecedarian


6/24


education programs and also includes a weighted regression line of effect size byeducation programs and also includes a weighted regression line of effect size bycalendar year.calendar year.Taken as a whole, the simple average effects size for early childhood educationTaken as a whole, the simple average effects size for early childhood educationon cognitive and achievement scores was .35 standard deviations at the end of theon cognitive and achievement scores was .35 standard deviations at the end of thetreatment periods, an amount equal to nearly half of race differences in the kinder-treatment periods, an amount equal to nearly half of race differences in the kinder-garten achievement gap (Duncan and Magnuson 2011). However, as can be seengarten achievement gap (Duncan and Magnuson 2011). However, as can be seenfrom Figure 2, average effect sizes vary substantially and studies with the largest effectfrom Figure 2, average effect sizes vary substantially and studies with the largest effectsizes tended to have the fewest subjects. When weighted by the inverse of the squaredsizes tended to have the fewest subjects. When weighted by the inverse of the squaredstandard errors of the estimates, the average drops to .21 standard deviations.standard errors of the estimates, the average drops to .21 standard deviations.All of the 84 programs that generated the effect size data shown in Figure 2All of the 84 programs that generated the effect size data shown in Figure 2met minimum standards for quality of research methods. However, some of themet minimum standards for quality of research methods. However, some of theprograms lasted for only a couple of summer months, while others ran for as longprograms lasted for only a couple of summer months, while others ran for as longas ve years. Some of the evaluations used random assignment while others reliedas ve years. Some of the evaluations used random assignment while others reliedon less-rigorous quasi-experimental methods. Almost all focused on childrenon less-rigorous quasi-experimental methods. Almost all focused on childrenfrom low-income families, but they varied in the racial and ethnic composition offrom low-income families, but they varied in the racial and ethnic composition oftreatment groups.treatment groups.One might assume that these differences would account for much of theOne might assume that these differences would account for much of theeffect-size variability observed in Figure 2. However, that is not always the case.effect-size variability observed in Figure 2. However, that is not always the case.Weighted average effect sizes were insigni cantly different between evaluations thatWeighted average effect sizes were insignicantly different between evaluations thatdid (.25 standard deviations) and did not (.19 standard deviations) use randomdid (.25 standard deviations) and did not (.19 standard deviations) use randomassignment; and between those that were (.31 standard deviations) and were notassignment; and between those that were (.31 standard deviations) and were not(.18 standard deviations) published in peer-review journals. The effect sizes of(.18 standard deviations) published in peer-review journals. The effect sizes ofprograms designed by researchers (.39 standard deviations) were signi cantly largerprograms designed by researchers (.39 standard deviations) were signicantly largerthan programs not designed by researchers (.18 standard deviations).than programs not designed by researchers (.18 standard deviations).Programs beginning before 1980 produced signi cantly larger effect sizesPrograms beginning before 1980 produced signicantly larger effect sizes(.33 standard deviations) than those that began later (.16 standard deviations).(.33 standard deviations) than those that began later (.16 standard deviations).Declining effect sizes over time are disappointing, as we might hope that lessonsDeclining effect sizes over time are disappointing, as we might hope that lessonsfrom prior evaluations and advances in the science of child development would havefrom prior evaluations and advances in the science of child development would haveled to an increase in program effects over time. However, the likely reason for theled to an increase in program effects over time. However, the likely reason for thedecline is that counterfactual conditions for children in the control groups in thesedecline is that counterfactual conditions for children in the control groups in thesestudies have improved substantially. We have already seen in Figure 1 how much morestudies have improved substantially. We have already seen in Figure 1 how much morelikely low-income children are to be attending some form of center-based care nowlikely low-income children are to be attending some form of center-based care nowrelative to 40 years ago. This matters because, though center-based care programsrelative to 40 years ago. This matters because, though center-based care programshave varying degrees of educational focus, most research suggests that center-basedhave varying degrees of educational focus, most research suggests that center-basedcare is associated with better cognitive and achievement outcomes for preschool agecare is associated with better cognitive and achievement outcomes for preschool agechildren (NICHD Early Childcare Research Network and Duncan 2003).children (NICHD Early Childcare Research Network and Duncan 2003).Even more impressive are gains in the likely quality of the home environmentEven more impressive are gains in the likely quality of the home environmentprovided by low-income mothers, as indexed by their completed schooling. In 1970,provided by low-income mothers, as indexed by their completed schooling. In 1970,some 71 percent of preschool age children in the bottom 20 percent of the incomesome 71 percent of preschool age children in the bottom 20 percent of the incomedistribution had mothers who lacked a high school degree, while only 5 percentdistribution had mothers who lacked a high school degree, while only 5 percentof the mothers had attended at least some postsecondary schooling (based onof the mothers had attended at least some postsecondary schooling (based onauthors calculation of the October Current Population Survey data). By 2000,authors calculation of the October Current Population Survey data). By 2000,the corresponding percentage of children with mothers who did not have a highthe corresponding percentage of children with mothers who did not have a highschool degree had dropped by nearly half (to 37 percent), while the percentageschool degree had dropped by nearly half (to 37 percent), while the percentage


7/24


with mothers who had completed some postsecondary schooling increased ve-foldwith mothers who had completed some postsecondary schooling increased ve-fold(to just over 25 percent). Today, therefore, children from low-income households(to just over 25 percent). Today, therefore, children from low-income householdsare likely to be bene ting from much higher-quality home environments thanare likely to be beneting from much higher-quality home environments thantheir counterparts four decades ago. Both higher-quality home environments andtheir counterparts four decades ago. Both higher-quality home environments andincreases in other forms of center-based child care raise the bar for impact estimatesincreases in other forms of center-based child care raise the bar for impact estimatescoming from early childhood education programs.coming from early childhood education programs.Two particularly salient features of early childhood education programs areTwo particularly salient features of early childhood education programs areduration and starting age. Abundant literature suggests that the number of yearsduration and starting age. Abundant literature suggests that the number of yearsspent in K12 or postsecondary education is linked to labor market success (Cardspent in K12 or postsecondary education is linked to labor market success (Card1999). Thus, it seems plausible to expect that longer exposure to early childhood1999). Thus, it seems plausible to expect that longer exposure to early childhoodeducation environments before school entry should boost later academic achieve-education environments before school entry should boost later academic achieve-ment as well. But while simple associations indicate that longer participation inment as well. But while simple associations indicate that longer participation ina preschool program generates larger treatment effects, models with a full set ofa preschool program generates larger treatment effects, models with a full set ofcontrols for program and evaluation quality yield only small and statistically insignif-controls for program and evaluation quality yield only small and statistically insignif-icant associations (icant associations (++04 standard deviations per additional year) between program.04 standard deviations per additional year) between programduration and magnitudes of impacts (Leak, Duncan, Li, Magnuson, Schindler, andduration and magnitudes of impacts (Leak, Duncan, Li, Magnuson, Schindler, andYoshikawa 2012). The absence of larger effects for longer-duration programs mayYoshikawa 2012). The absence of larger effects for longer-duration programs maybe due to the failure of such programs to use curricula and activities that capitalizebe due to the failure of such programs to use curricula and activities that capitalizefully on the skills gained in the early years of program participation.fully on the skills gained in the early years of program participation.As for starting age, neuroscience evidence on the plasticity of cognitive andAs for starting age, neuroscience evidence on the plasticity of cognitive andlanguage abilities suggests that these skills are highly amenable to environmentallanguage abilities suggests that these skills are highly amenable to environmentalenrichment during the early childhood period. Starting in infancy, responsiveenrichment during the early childhood period. Starting in infancy, responsivecaregiving and language-rich interactions are associated with better developmentalcaregiving and language-rich interactions are associated with better developmentaloutcomes, and more speci cally stronger early language development (Tamis-outcomes, and more specically stronger early language development (Tamis-LeMonda, Bornstein, and Baumwell 2001). Based on such ndings, we mightLeMonda, Bornstein, and Baumwell 2001). Based on such ndings, we mightexpect to nd an earlier is better pattern of effects for early childhood educa-expect to nd an earlier is better pattern of effects for early childhood educa-tion programs that provide such high-quality interactions for children. Evidencetion programs that provide such high-quality interactions for children. Evidencefrom the best-known early-life preschool programs is mixed: programs such as Earlyfrom the best-known early-life preschool programs is mixed: programs such as EarlyHead Start produce very small impacts on cognitive development (Love et al. 2003),Head Start produce very small impacts on cognitive development (Love et al. 2003),whereas others, like the Abecedarian program, show much larger impacts (Rameywhereas others, like the Abecedarian program, show much larger impacts (Rameyand Campbell 1984). Analysis of the meta-analytic database shows that, takenand Campbell 1984). Analysis of the meta-analytic database shows that, takenas a whole, effect sizes were neither larger nor smaller for children who startedas a whole, effect sizes were neither larger nor smaller for children who startedprograms at younger ages (Leak, Duncan, Li, Magnuson, Schindler, and Yoshikawaprograms at younger ages (Leak, Duncan, Li, Magnuson, Schindler, and Yoshikawa2012). This suggests that other modes of early childhood investmentsfor example,2012). This suggests that other modes of early childhood investmentsfor example,home visitation for high-risk, rst-time mothers (Olds, Sadler, and Kitzman 2007)home visitation for high-risk, rst-time mothers (Olds, Sadler, and Kitzman 2007)or developmental screenings and interventions for children living in families withor developmental screenings and interventions for children living in families withdocumented domestic violencemay be more-effective ways of building childrensdocumented domestic violencemay be more-effective ways of building childrenscapacities during the very early years of life.capacities during the very early years of life.Model Program Impacts: Perry Preschool and AbecedarianAs shown in Figure 2, average end-of-treatment effect sizes for the PerryAs shown in Figure 2, average end-of-treatment effect sizes for the PerryPreschool and Abecedarian programs are several times larger than the weightedPreschool and Abecedarian programs are several times larger than the weightedmean effect size for all studies in the meta-analytic database that met our inclusionmean effect size for all studies in the meta-analytic database that met our inclusioncriteria. A key reason for the prominence of these two studies and a few others iscriteria. A key reason for the prominence of these two studies and a few others is


8/24


that long-term follow-ups show strikingly positive impacts in adulthood and impres-that long-term follow-ups show strikingly positive impacts in adulthood and impres-sive bene tcost ratios.sive benetcost ratios.Perry provided one or two years of part-day educational services and weeklyPerry provided one or two years of part-day educational services and weeklyhome visits to 58 low-income, low-IQ, African American children aged three andhome visits to 58 low-income, low-IQ, African American children aged three andfour in Ypsilanti, Michigan, during the 1960s. The curriculum was geared to thefour in Ypsilanti, Michigan, during the 1960s. The curriculum was geared to thechildrens age and capabilities, emphasizing child-initiated learning activities.childrens age and capabilities, emphasizing child-initiated learning activities.Staff encouraged children to engage in play activities that would promote theirStaff encouraged children to engage in play activities that would promote theirproblem-solving skills as well as their intellectual, social, and physical develop-problem-solving skills as well as their intellectual, social, and physical develop-ment. Program staff made weekly one- to two-hour afternoon visits to each family.ment. Program staff made weekly one- to two-hour afternoon visits to each family.The centers child-to-teacher ratio was low; each of four teachers served onlyThe centers child-to-teacher ratio was low; each of four teachers served only20 25 children every year. Per-pupil costs amounted to about $20,000 per child (in20 25 children every year. Per-pupil costs amounted to about $20,000 per child (in2011 dollars). While Perrys large impacts on IQ at the point of school entry had2011 dollars). While Perrys large impacts on IQ at the point of school entry hadall but disappeared by third grade (Schweinhart, Montie, Xiang, Barnett, Bel eld,all but disappeared by third grade (Schweinhart, Montie, Xiang, Barnett, Beleld,and Nores 2005), the program produced lasting improvements through age 40 onand Nores 2005), the program produced lasting improvements through age 40 onemployment rates and substantially reduced the likelihood that participants hademployment rates and substantially reduced the likelihood that participants hadbeen arrested. Heckman, Moon, Pinto, Savelyev, and Yavitz (2010) estimate thatbeen arrested. Heckman, Moon, Pinto, Savelyev, and Yavitz (2010) estimate thatthe program generated about $152,000 in bene ts over the life course, boostingthe program generated about $152,000 in benets over the life course, boostingindividuals earnings, reducing use of welfare programs, and, most importantlyindividuals earnings, reducing use of welfare programs, and, most importantlyfor the bene t calculation, reducing criminal activity. These nancial bene tsfor the benet calculation, reducing criminal activity. These nancial benetsproduced a social rate of return between 7 and 10 percent.produced a social rate of return between 7 and 10 percent.The Abecedarian program, which served 57 low-income, mostly AfricanThe Abecedarian program, which served 57 low-income, mostly AfricanAmerican families from Chapel Hill, North Carolina, provided even more-American families from Chapel Hill, North Carolina, provided even more-intensive services than Perry Preschool. Beginning in 1972, children assignedintensive services than Perry Preschool. Beginning in 1972, children assignedto the Abecedarian treatment received year-round, full-time center-based careto the Abecedarian treatment received year-round, full-time center-based carefor ve years, starting in the childs rst year of life. The Abecedarian preschoolfor ve years, starting in the childs rst year of life. The Abecedarian preschoolprogram included transportation, individualized educational activities thatprogram included transportation, individualized educational activities thatchanged as the children grew older, and low childteacher ratios of 3:1 for thechanged as the children grew older, and low childteacher ratios of 3:1 for theyoungest children and up to 6:1 for older children. Abecedarian teachers followedyoungest children and up to 6:1 for older children. Abecedarian teachers followeda curriculum that focused on language development and explained to teachersa curriculum that focused on language development and explained to teachersthe importance of each task as well as how to teach it. High-quality health care,the importance of each task as well as how to teach it. High-quality health care,additional social services, and nutritional supplements were also provided toadditional social services, and nutritional supplements were also provided toparticipating families (Ramey and Campbell 1979; Campbell, Ramey, Pungello,participating families (Ramey and Campbell 1979; Campbell, Ramey, Pungello,Sparkling, and Miller-Johnson 2002).Sparkling, and Miller-Johnson 2002).At two years of age, the control-group children in the Abecedarian program hadAt two years of age, the control-group children in the Abecedarian program hadIQ scores that averaged about one standard deviation below the mean, as would beIQ scores that averaged about one standard deviation below the mean, as would beexpected for children from very economically disadvantaged backgrounds (Ramey,expected for children from very economically disadvantaged backgrounds (Ramey,Campbell, Burchinal, Skinner, Gardner, and Ramey 2000). By the time the childrenCampbell, Burchinal, Skinner, Gardner, and Ramey 2000). By the time the childrenreached age ve, however, their IQ scores were close to the national average, andreached age ve, however, their IQ scores were close to the national average, and10 points higher than scores of comparable children who did not participate in the10 points higher than scores of comparable children who did not participate in theprogram. Similarly large effects were observed for achievement on verbal and quan-program. Similarly large effects were observed for achievement on verbal and quan-titative tests (Ramey and Campbell 1984). Nearly 15 years later, the programs effecttitative tests (Ramey and Campbell 1984). Nearly 15 years later, the programs effecton IQ scores at age 21 (.38 standard deviations) was still substantial but smaller thanon IQ scores at age 21 (.38 standard deviations) was still substantial but smaller thanat age ve. Children in the Abecedarian program entered college at 2.5 times theat age ve. Children in the Abecedarian program entered college at 2.5 times therate of children in the control group, and the intervention also reduced rates ofrate of children in the control group, and the intervention also reduced rates of


9/24


teen parenthood and marijuana use by nearly half, although it did not lead to statis-teen parenthood and marijuana use by nearly half, although it did not lead to statis-tically signi cant reductions in criminal activity. Expressed in 2011 dollars, the coststically signicant reductions in criminal activity. Expressed in 2011 dollars, the costsassociated with Abecedarians ve-year duration totaled about $80,000 per child,associated with Abecedarians ve-year duration totaled about $80,000 per child,and the program is estimated to have produced $160,000 in net present bene ts forand the program is estimated to have produced $160,000 in net present benets forits participants and their parents (Barnett and Masse 2007; Currie 2001).its participants and their parents (Barnett and Masse 2007; Currie 2001).It is dif cult to extract policy lessons from these two initiatives for early child-It is difcult to extract policy lessons from these two initiatives for early child-hood education programs that states or the federal government might offer today.hood education programs that states or the federal government might offer today.Both programs were designed and evaluated by researchers and each served onlyBoth programs were designed and evaluated by researchers and each served onlyseveral dozen children conditions that scaled-up programs cannot match. More-several dozen children conditions that scaled-up programs cannot match. More-over, as we have pointed out above, counterfactual conditions three decades agoover, as we have pointed out above, counterfactual conditions three decades agowere likely of a comparatively low quality. The average number of years of maternalwere likely of a comparatively low quality. The average number of years of maternaleducation completed was about 10 years for both the Perry and Abecedarianeducation completed was about 10 years for both the Perry and Abecedarianpreschool treatment groups, reecting the low levels of parental education amongpreschool treatment groups, reecting the low levels of parental education amonglow-income families at that time.low-income families at that time.Head Start ImpactsLarge-scale policy lessons might be gleaned more reliably from studies ofLarge-scale policy lessons might be gleaned more reliably from studies ofHead Start, since that program now provides services to almost a million three- andHead Start, since that program now provides services to almost a million three- andfour-year-olds. Early quasi-experimental evaluations of Head Start found signi cantfour-year-olds. Early quasi-experimental evaluations of Head Start found signicantshort-term gains in participants achievement test scores, but as with Perry andshort-term gains in participants achievement test scores, but as with Perry andAbecedarian, these achievement gains appeared to fade over time (Cicirelli 1969;Abecedarian, these achievement gains appeared to fade over time (Cicirelli 1969;McKey, Condelli, Ganson, Barrett, McConkey, and Plantz 1985). Despite method-McKey, Condelli, Ganson, Barrett, McConkey, and Plantz 1985). Despite method-ological critiques of these early studies (McGroder 1990), a random-assignmentological critiques of these early studies (McGroder 1990), a random-assignmentnational study of Head Start was not undertaken for another 30 years.national study of Head Start was not undertaken for another 30 years.Begun in 2002, the Head Start Impact Study (HSIS) used wait-list lotteriesBegun in 2002, the Head Start Impact Study (HSIS) used wait-list lotteriesto assign children to the opportunity to enroll in a Head Start program. Resultsto assign children to the opportunity to enroll in a Head Start program. Resultsindicated that after one academic year in the program, four-year-olds who had theindicated that after one academic year in the program, four-year-olds who had theopportunity to enroll in Head Start gained signi cantly more in six language andopportunity to enroll in Head Start gained signicantly more in six language andliteracy areas than control-group children who lost the enrollment lotteries, withliteracy areas than control-group children who lost the enrollment lotteries, withthese intent-to-treat effects (effects for the group of children who had the oppor-these intent-to-treat effects (effects for the group of children who had the oppor-tunity to enroll) ranging from .09 to .31 standard deviations (US Department oftunity to enroll) ranging from .09 to .31 standard deviations (US Department ofHealth and Human Services 2005). In contrast, there were few program impactsHealth and Human Services 2005). In contrast, there were few program impactson math skills or on childrens attention, anti-social, or mental health problems.on math skills or on childrens attention, anti-social, or mental health problems.The of cial report of the Head Start Impact Study (US Department of Health andThe ofcial report of the Head Start Impact Study (US Department of Health andHuman Services 2005) provides estimates of differences between (parents of) chil-Human Services 2005) provides estimates of differences between (parents of) chil-dren offered and children not offered a chance to get into the Head Start centerdren offered and children not offered a chance to get into the Head Start centerwith the waitlist lottery. Some children offered the chance didnt take it, and somewith the waitlist lottery. Some children offered the chance didnt take it, and somechildren not offered a slot ended up in other Head Start centers. Ludwig andchildren not offered a slot ended up in other Head Start centers. Ludwig andPhillips (2007) make the proper treatment on the treated estimate in light of thisPhillips (2007) make the proper treatment on the treated estimate in light of thisnoncompliance, and the resulting effect sizes were roughly 50 percent larger thannoncompliance, and the resulting effect sizes were roughly 50 percent larger thanintent-to-treat effect sizes. By the end of rst grade, both achievement levels andintent-to-treat effect sizes. By the end of rst grade, both achievement levels andbehavioral ratings of treatment group children were essentially similar to achieve-behavioral ratings of treatment group children were essentially similar to achieve-ment levels of control-group children (US Department of Health and Humanment levels of control-group children (US Department of Health and HumanServices 2010).Services 2010).


10/24


Why might Head Starts initial achievement impacts disappear so quickly? AllWhy might Head Starts initial achievement impacts disappear so quickly? Allchildren learn, but they learn at different rates. If the test scores of Head Start andchildren learn, but they learn at different rates. If the test scores of Head Start andcomparison-group children converge during elementary school, then the treatmentcomparison-group children converge during elementary school, then the treatmentgroups preschool gains must be offset later by larger gains in the control group.groups preschool gains must be offset later by larger gains in the control group.Why this happens is not entirely clear; most arguments focus on the quality of subse-Why this happens is not entirely clear; most arguments focus on the quality of subse-quent schools that children attend. If little learning occurs in low-quality schools,quent schools that children attend. If little learning occurs in low-quality schools,then early advantages imparted by programs such as Head Start might be lost. Inthen early advantages imparted by programs such as Head Start might be lost. Inthis case, preschool does not immunize against the adverse effects of subsequentthis case, preschool does not immunize against the adverse effects of subsequentlow-quality schooling (Currie and Thomas 2000; Lee and Loeb 1995).low-quality schooling (Currie and Thomas 2000; Lee and Loeb 1995).Currie and Thomas (2000) showed that Head Start impacts fade out moreCurrie and Thomas (2000) showed that Head Start impacts fade out morerapidly for African-American children than for white children; in examining why,rapidly for African-American children than for white children; in examining why,they show that African-American children in Head Start attend lower-quality schools,they show that African-American children in Head Start attend lower-quality schools,as measured by students average test scores, relative to the schools attended byas measured by students average test scores, relative to the schools attended byAfrican-American children who did not attend Head Start. In contrast, for whiteAfrican-American children who did not attend Head Start. In contrast, for whitechildren, average school quality did not differ by Head Start participation status.children, average school quality did not differ by Head Start participation status.Similarly, Zhai, Raver, and Jones (2012) nd that the bene ts to children of an inter-Similarly, Zhai, Raver, and Jones (2012) nd that the benets to children of an inter-vention designed to enhance the developmental quality of Head Start programsvention designed to enhance the developmental quality of Head Start programspersisted into kindergarten only for those children who attended relatively higher-persisted into kindergarten only for those children who attended relatively higher-quality elementary schools, again measured by student test scores.quality elementary schools, again measured by student test scores.An alternative explanation of achievement-impact fadeout is that kindergartenAn alternative explanation of achievement-impact fadeout is that kindergartenteachers might be particularly effective at teaching children with low levels of skills.teachers might be particularly effective at teaching children with low levels of skills.In this case, it may be that the classroom is not of generally low quality, but instruc-In this case, it may be that the classroom is not of generally low quality, but instruc-tional efforts may favor children at the lower end of the skill distribution, whichtional efforts may favor children at the lower end of the skill distribution, whichwould include larger concentrations of children who had not participated in earlywould include larger concentrations of children who had not participated in earlychildhood education. Indirect evidence supporting this hypothesis is provided inchildhood education. Indirect evidence supporting this hypothesis is provided inthe work of Engel, Claessens, and Finch (forthcoming), who nd that kindergartenthe work of Engel, Claessens, and Finch (forthcoming), who nd that kindergartenteachers spend the most time on very basic math instruction (like learning numbers)teachers spend the most time on very basic math instruction (like learning numbers)despite the fact that the vast majority of kindergarteners have already acquired suchdespite the fact that the vast majority of kindergarteners have already acquired suchskills. If this explanation holds, the effects of early childhood education programsskills. If this explanation holds, the effects of early childhood education programsare most likely to persist in subsequent schooling environments in which learningare most likely to persist in subsequent schooling environments in which learninggains are equally distributed across children with high and low levels of initial skills.gains are equally distributed across children with high and low levels of initial skills.44As with Perry and Abecedarian program ndings, quickly declining test scoreAs with Perry and Abecedarian program ndings, quickly declining test scoreimpacts for recent cohorts of Head Start children appear to be at odds with theimpacts for recent cohorts of Head Start children appear to be at odds with thelong-term impacts on important young adult outcomes found in analyses of olderlong-term impacts on important young adult outcomes found in analyses of olderHead Start cohorts. Some of the older-cohort studies use strong quasi-experimentalHead Start cohorts. Some of the older-cohort studies use strong quasi-experimentalmethods and nd quite striking long-run program impacts. One of the mostmethods and nd quite striking long-run program impacts. One of the mostrecent and comprehensive is Demings (2009) sibling-based xed-effect analysis,recent and comprehensive is Demings (2009) sibling-based xed-effect analysis,which found that, compared with siblings who did not attend Head Start or otherwhich found that, compared with siblings who did not attend Head Start or otherpreschool programs, children who attended Head Start in the 1980s and early 1990spreschool programs, children who attended Head Start in the 1980s and early 1990s4A third explanation would be that program impacts do not persist because early elementary instruc-

tion is most benecial to children who enter school with high levels of initial skills and that Head Startprogram impacts are not sufciently large to get children to a point at which they will benet from suchinstruction. There does not seem to be good evidence to support this conjecture.


11/24


were over 8 percentage points more likely to graduate from high school. Demingswere over 8 percentage points more likely to graduate from high school. Demingsmore-general composite of positive early adult outcomesincluding high schoolmore-general composite of positive early adult outcomesincluding high schoolgraduation, college attendance, idleness, crime, teen parenthood, and healthgraduation, college attendance, idleness, crime, teen parenthood, and healthstatus shows an estimated impact of .23 standard deviations.status shows an estimated impact of .23 standard deviations.Ludwig and Millers (2007) regression discontinuity study of Head StartLudwig and Millers (2007) regression discontinuity study of Head Startattendees in the late 1960s found that successful efforts to increase the likelihoodattendees in the late 1960s found that successful efforts to increase the likelihoodthat poor counties would establish Head Start programs by providing federal grant-that poor counties would establish Head Start programs by providing federal grant-writing assistance led to gains of 3 4 percentage points in high school graduationwriting assistance led to gains of 3 4 percentage points in high school graduationrates and postsecondary schooling in the 1990 census data relative to counties withrates and postsecondary schooling in the 1990 census data relative to counties withvery similar levels of poverty that were not offered such assistance, although suchvery similar levels of poverty that were not offered such assistance, although sucheffects were attenuated by 2000. Taken together, these studies suggest that despiteeffects were attenuated by 2000. Taken together, these studies suggest that despitethe decline in program impacts on achievement test scores as children progressthe decline in program impacts on achievement test scores as children progressthrough elementary school, there may be measurable and important effects ofthrough elementary school, there may be measurable and important effects ofHead Start on childrens life chances.Head Start on childrens life chances.Pre-Kindergarten ProgramsSome rigorous evaluations of pre-kindergarten programs were completed tooSome rigorous evaluations of pre-kindergarten programs were completed toorecently to have been included in the database used to produce Figure 2. Mostrecently to have been included in the database used to produce Figure 2. Mostof these studies use regression discontinuity designs based on strict birthdayof these studies use regression discontinuity designs based on strict birthdaycutoffs. Test-based assessments are given to children who just started attendingcutoffs. Test-based assessments are given to children who just started attendingpre-kindergarten and those who just completed it. The tests of children who justpre-kindergarten and those who just completed it. The tests of children who justcompleted the program are compared with those about to attend. Children whosecompleted the program are compared with those about to attend. Children whoseparents are not interested in enrolling them in the program are not part of eitherparents are not interested in enrolling them in the program are not part of eithergroup. For this reason (and a few others), these designs are not directly comparablegroup. For this reason (and a few others), these designs are not directly comparableto either intent-to-treat or treatment-on-the-treated estimates from experimentalto either intent-to-treat or treatment-on-the-treated estimates from experimentalstudies (Lipsey, Weiland, Yoshikawa, Wilson, and Hofer 2011; Gibbs, Ludwig, andstudies (Lipsey, Weiland, Yoshikawa, Wilson, and Hofer 2011; Gibbs, Ludwig, andMiller 2011). The most comprehensive overview is Wong, Cook, Barnett, and JungMiller 2011). The most comprehensive overview is Wong, Cook, Barnett, and Jung(2008), which examines ve state pre-kindergarten programs and nds short-run(2008), which examines ve state pre-kindergarten programs and nds short-runeffects on achievement test scores that are somewhat larger than those estimatedeffects on achievement test scores that are somewhat larger than those estimatedin the National Head Start Impact Study, although the size of the impacts variesin the National Head Start Impact Study, although the size of the impacts variesconsiderably across states and types of test (weighted average intent-to-treat impactsconsiderably across states and types of test (weighted average intent-to-treat impactsrange from .17 for vocabulary to .68 for print awareness).range from .17 for vocabulary to .68 for print awareness).

The highly regarded Tulsa pre-kindergarten program has also been carefullyThe highly regarded Tulsa pre-kindergarten program has also been carefullyevaluated. A birthday cutoff-based regression discontinuity evaluation of the programevaluated. A birthday cutoff-based regression discontinuity evaluation of the programfound large and signi cant effects on childrens achievement, with effect sizesfound large and signicant effects on childrens achievement, with effect sizesranging from .38 to .79 (Gormley, Gayer, Phillips, and Dawson 2005). Adjusting forranging from .38 to .79 (Gormley, Gayer, Phillips, and Dawson 2005). Adjusting fordifferences in childrens backgrounds (using propensity score matching methods),differences in childrens backgrounds (using propensity score matching methods),the researchers found that the Tulsa pre-kindergarten program reduced attendeesthe researchers found that the Tulsa pre-kindergarten program reduced attendeestimidity and improved their attentiveness. The program did not appear to affecttimidity and improved their attentiveness. The program did not appear to affectdisobedience, apathy, aggression, learning task problems, or problems interactingdisobedience, apathy, aggression, learning task problems, or problems interactingwith peers or teachers (Gormley, Phillips, Newmark, Welti, and Adelstein 2011).with peers or teachers (Gormley, Phillips, Newmark, Welti, and Adelstein 2011).The only longer-run follow-up study conducted to date of pre-kindergartenThe only longer-run follow-up study conducted to date of pre-kindergartenprogram uses propensity matching and administrative data on third grade testprogram uses propensity matching and administrative data on third grade testscores. Hill, Gormley, and Adelstein (2012) estimating program impacts forscores. Hill, Gormley, and Adelstein (2012) estimating program impacts for


12/24


two cohorts. They nd no lasting discernible achievement impacts for the rsttwo cohorts. They nd no lasting discernible achievement impacts for the rstcohort by third grade. For the second cohort there is evidence of persisting mathcohort by third grade. For the second cohort there is evidence of persisting mathimpacts (.18 standard deviations), perhaps reecting an increased emphasis onimpacts (.18 standard deviations), perhaps reecting an increased emphasis onmath instruction, including the introduction of new curricula, during elementarymath instruction, including the introduction of new curricula, during elementaryschool. The lack of longer-run evaluations of pre-kindergarten programs suggestsschool. The lack of longer-run evaluations of pre-kindergarten programs suggeststhat drawing strong policy conclusions about their effectiveness is unwarranted, asthat drawing strong policy conclusions about their effectiveness is unwarranted, asother programs have likewise demonstrated early promising results that faded overother programs have likewise demonstrated early promising results that faded overthe rst few years of school.the rst few years of school.

The Puzzle: Academic Fade-Out, but Long-Term Benets

Most early childhood education studies that have tracked children beyondMost early childhood education studies that have tracked children beyondthe end of the program treatment nd that effects on test scores fade over time.the end of the program treatment nd that effects on test scores fade over time.An analysis of cognitive and achievement outcomes in our meta-analytic database,An analysis of cognitive and achievement outcomes in our meta-analytic database,which includes model programs such as Perry Preschool as well as Head Start andwhich includes model programs such as Perry Preschool as well as Head Start andmany other programs, shows an estimated decrease in program impact effect sizesmany other programs, shows an estimated decrease in program impact effect sizesof about .03 standard deviations per year. With end-of-treatment effect sizes aver-of about .03 standard deviations per year. With end-of-treatment effect sizes aver-aging around .30 standard deviations, this implies that positive effects persist foraging around .30 standard deviations, this implies that positive effects persist forroughly 10 years (Leak et al. 2011; see also Aos, Lieb, May eld, Miller, and Pennucciroughly 10 years (Leak et al. 2011; see also Aos, Lieb, Mayeld, Miller, and Pennucci2004; Camilli, Vargas, Ryan, and Barnett 2010). This nding raises a puzzle: How2004; Camilli, Vargas, Ryan, and Barnett 2010). This nding raises a puzzle: Howdo we reconcile the fade-out of preschool program impacts on test scores duringdo we reconcile the fade-out of preschool program impacts on test scores duringelementary school with the evidence showing that such programs nonetheless haveelementary school with the evidence showing that such programs nonetheless havebene cial impacts on a broad set of later-life outcomes like high school graduationbenecial impacts on a broad set of later-life outcomes like high school graduationrates, teen parenthood, and criminality?rates, teen parenthood, and criminality?One obvious possible explanation is that preschool programs may affect some-One obvious possible explanation is that preschool programs may affect some-thing other than basic achievement and cognitive test scores, and perhaps thesething other than basic achievement and cognitive test scores, and perhaps theseother program impacts, unlike achievement and cognitive impacts, persist overother program impacts, unlike achievement and cognitive impacts, persist overtime. In turn, this raises the question of exactly how early childhood educationtime. In turn, this raises the question of exactly how early childhood educationprograms affect various aspects of development, including cognitive skills, person-programs affect various aspects of development, including cognitive skills, person-ality traits like conscientiousness, and the behavior categories like attentiveness orality traits like conscientiousness, and the behavior categories like attentiveness orantisocial behavior that are often emphasized by development psychologists. Theantisocial behavior that are often emphasized by development psychologists. Theliterature on the effects of preschool has drawn on several different models ofliterature on the effects of preschool has drawn on several different models ofhuman development.human development.In one prominent example, Cunha and Heckman (2007) posit a cumula-In one prominent example, Cunha and Heckman (2007) posit a cumula-tive model of the production of human capital that allows for the possibility oftive model of the production of human capital that allows for the possibility ofdiffering childhood investment stages as well as roles for the past effects and futurediffering childhood investment stages as well as roles for the past effects and futuredevelopment of both cognitive and socio-emotional skills. In this model, childrendevelopment of both cognitive and socio-emotional skills. In this model, childrenhave endowments at birth of cognitive potential and temperament that reect ahave endowments at birth of cognitive potential and temperament that reect acombination of genetic and prenatal environmental inuences. The Cunha andcombination of genetic and prenatal environmental inuences. The Cunha andHeckman model highlights the interactive nature of skill building and investmentsHeckman model highlights the interactive nature of skill building and investmentsfrom families, preschools and schools, and other agents. It suggests that humanfrom families, preschools and schools, and other agents. It suggests that humancapital accumulation results from self-productivityskills developed in earliercapital accumulation results from self-productivityskills developed in earlierstages bolster the development of skills in later stagesas well as the dynamicstages bolster the development of skills in later stagesas well as the dynamic


13/24


complementary that results when skills acquired prior to a given investmentcomplementary that results when skills acquired prior to a given investmentincrease the productivity of that investment. These two principles are combined inincrease the productivity of that investment. These two principles are combined inthe hypothesis that skill begets skill.the hypothesis that skill begets skill.Several aspects of this model are relevant for preschool investment policy. IfSeveral aspects of this model are relevant for preschool investment policy. Iffocused on the preschool period, the Cunha and Heckman (2007) model impliesfocused on the preschool period, the Cunha and Heckman (2007) model impliesthat school readiness is a product of the childs cognitive and socio-emotional skillsthat school readiness is a product of the childs cognitive and socio-emotional skillsupon entry into the preschool period, plus preschool-period investments fromupon entry into the preschool period, plus preschool-period investments fromparents and possibly from an early childhood education program. The hypothesisparents and possibly from an early childhood education program. The hypothesisof dynamic complementarity implies that the effects of parental and early childhoodof dynamic complementarity implies that the effects of parental and early childhoodeducation investments on child outcomes will be largest for children who enter theeducation investments on child outcomes will be largest for children who enter thepreschool period with the highest levels of cognitive and socio-emotional skills.preschool period with the highest levels of cognitive and socio-emotional skills.Predictions emerging from the models of human capital development proposedPredictions emerging from the models of human capital development proposedin the developmental psychology literature are different. These models, too, focusin the developmental psychology literature are different. These models, too, focuson how individuals endowments interact with environmental experiences, andon how individuals endowments interact with environmental experiences, andsuggest that both individual capacities and experience shape development (Blairsuggest that both individual capacities and experience shape development (Blairand Raver 2012). However, they diverge from the Cunha and Heckman (2007)and Raver 2012). However, they diverge from the Cunha and Heckman (2007)model by distinguishing how environments and different types of investments (formodel by distinguishing how environments and different types of investments (forexample, parent and early-childhood-education investments) interact to shapeexample, parent and early-childhood-education investments) interact to shapedevelopment. Developmental models say that certain kinds of programs may bedevelopment. Developmental models say that certain kinds of programs may bemost productive for higher-skilled children while others are geared towards helpingmost productive for higher-skilled children while others are geared towards helpingbring up the skills of low-skill children and dont match well to the needs of higher-bring up the skills of low-skill children and dont match well to the needs of higher-skill children. For example, Ramey and Rameys compensatory model (1998)skill children. For example, Ramey and Rameys compensatory model (1998)posits that preschool investments can function as a substitute for enriched homeposits that preschool investments can function as a substitute for enriched homeenvironments. Thus, children whose skill development may be compromised byenvironments. Thus, children whose skill development may be compromised byeconomic disadvantage or low-quality home environments are predicted to bene teconomic disadvantage or low-quality home environments are predicted to benetmore from early childhood education programs than more-advantaged children.more from early childhood education programs than more-advantaged children.This hypothesis provided the rationale for the initial and continued funding forThis hypothesis provided the rationale for the initial and continued funding forprograms such as Head Start and Early Head Start, which target children fromprograms such as Head Start and Early Head Start, which target children fromdisadvantaged backgrounds.disadvantaged backgrounds.If early childhood education programs seek to build childrens early skills toIf early childhood education programs seek to build childrens early skills togenerate lasting changes in adults human capital, which skills should they target?generate lasting changes in adults human capital, which skills should they target?Economists tend to lump IQ and achievement into a cognitive category and every-Economists tend to lump IQ and achievement into a cognitive category and every-thing else into a noncognitive category, but this distinction is unhelpful for a varietything else into a noncognitive category, but this distinction is unhelpful for a varietyof reasons. First, cognitive skills are a heterogeneous mixture of achievementof reasons. First, cognitive skills are a heterogeneous mixture of achievementand more-basic cognitive capacities. Although scores on tests of cognitive ability andand more-basic cognitive capacities. Although scores on tests of cognitive ability andachievement tend to be highly correlated, there is an important conceptual differ-achievement tend to be highly correlated, there is an important conceptual differ-ence between them. Achievement commonly refers to concrete academic skillsence between them. Achievement commonly refers to concrete academic skillssuch as literacy and numeracy that develop in response to parenting, schooling, andsuch as literacy and numeracy that develop in response to parenting, schooling, andother human capital investments, including early childhood education, whereasother human capital investments, including early childhood education, whereasIQ or general cognitive ability is considered to be a relatively more-stable trait.IQ or general cognitive ability is considered to be a relatively more-stable trait.Second, learning skills such as the ability to sustain attention when performingSecond, learning skills such as the ability to sustain attention when performingtasks, plan ahead, and control emotions in the face of provocation involve many oftasks, plan ahead, and control emotions in the face of provocation involve many ofthe same elements of brain circuitry as learning concrete skills, and are thereforethe same elements of brain circuitry as learning concrete skills, and are thereforeinherently cognitive. Third and most important, different branches of psychologyinherently cognitive. Third and most important, different branches of psychology


14/24


typically categorized noncognitive skills in very different ways. Conceptualizing andtypically categorized noncognitive skills in very different ways. Conceptualizing andmeasuring distinct components of noncognitive skills is a vital rst step in under-measuring distinct components of noncognitive skills is a vital rst step in under-standing why early childhood education and other human capital inventions havestanding why early childhood education and other human capital inventions havean effect.an effect.Most personality psychologists have centered their work on the big ve person-Most personality psychologists have centered their work on the big ve person-ality traits, which are derived from factor analyses of observer- and self-reports ofality traits, which are derived from factor analyses of observer- and self-reports ofbehaviors and include conscientiousness, openness, agreeableness, emotional stability,behaviors and include conscientiousness, openness, agreeableness, emotional stability,and extraversionplus general cognitive ability. Education research consistentlyand extraversionplus general cognitive ability. Education research consistentlyshows that conscientiousness best correlates with overall attainment and achievementshows that conscientiousness best correlates with overall attainment and achievement(Almlund, Duckworth, Heckman, and Kautz 2011). Although these traits have tradi-(Almlund, Duckworth, Heckman, and Kautz 2011). Although these traits have tradi-tionally been viewed as relatively stable across the lifespan, some evidence indicatestionally been viewed as relatively stable across the lifespan, some evidence indicatesthat they can change in response to life experiences and interventions (for example,that they can change in response to life experiences and interventions (for example,Roberts, Walton, and Viechtbauer 2006; Almlund et al. 2011).Roberts, Walton, and Viechtbauer 2006; Almlund et al. 2011).Developmental psychologists view childrens skills and behaviors as determinedDevelopmental psychologists view childrens skills and behaviors as determinedby the interplay between their innate abilities, their dispositions, and the quality ofby the interplay between their innate abilities, their dispositions, and the quality oftheir early experiences which may include early childhood education (Committeetheir early experiences which may include early childhood education (Committeeon Integrating the Science of Early Childhood Development, 2000). They classifyon Integrating the Science of Early Childhood Development, 2000). They classifyskills and behaviors in a number of ways, and some of their categories correspondskills and behaviors in a number of ways, and some of their categories correspondto the big ve personality traits. For example, our own recent review classi edto the big ve personality traits. For example, our own recent review classiedimportant competencies into four groups: achievement, attention, externalizingimportant competencies into four groups: achievement, attention, externalizingbehavior problems, and mental health (Duncan and Magnuson 2011). Attentionbehavior problems, and mental health (Duncan and Magnuson 2011). Attentionrefers to the ability to control impulses and focus on tasks (for example, Raver 2004).refers to the ability to control impulses and focus on tasks (for example, Raver 2004).Externalizing behavior refers to a cluster of related behaviors including antisocialExternalizing behavior refers to a cluster of related behaviors including antisocialbehavior, conduct disorders, and more-general aggression (Campbell, Shaw, andbehavior, conduct disorders, and more-general aggression (Campbell, Shaw, andGilliom 2000). Mental health constructs include anxiety and depression as well asGilliom 2000). Mental health constructs include anxiety and depression as well assomatic complaints and withdrawn behavior (Bongers, Koot, van der Ende, andsomatic complaints and withdrawn behavior (Bongers, Koot, van der Ende, andVerhulst 2003). All of these skills and behaviors might respond to investments inVerhulst 2003). All of these skills and behaviors might respond to investments inearly childhood education.early childhood education.Testing and comparing how these theories of human development apply inTesting and comparing how these theories of human development apply inthe context of early childhood education is dif cult, because despite argumentsthe context of early childhood education is difcult, because despite argumentsthat early childhood education programs are likely to generate broad impactsthat early childhood education programs are likely to generate broad impactson childrens behavior and social competence (Zigler and Trickett 1978), moston childrens behavior and social competence (Zigler and Trickett 1978), mostpreschool studies do not measure many of these kinds of outcomes at programpreschool studies do not measure many of these kinds of outcomes at programcompletion. Some studies have included measures of problem behavior, typicallycompletion. Some studies have included measures of problem behavior, typicallyratings of childrens antisocial or aggressive behaviors, with mixed results. Perryratings of childrens antisocial or aggressive behaviors, with mixed results. Perrysigni cantly reduced problem behavior, especially among boys, and the examina-signicantly reduced problem behavior, especially among boys, and the examina-tion by Heckman, Moon, Pinto, Savelyev, and Yavitz (2010) of Perrys long-runtion by Heckman, Moon, Pinto, Savelyev, and Yavitz (2010) of Perrys long-runeffects nds that these behavior impacts explain a substantial proportion of theeffects nds that these behavior impacts explain a substantial proportion of theprograms effects on boys crime and employment outcomes. However, both earlyprograms effects on boys crime and employment outcomes. However, both earlycognitive and behavioral impacts explain program impacts on girls later outcomes.cognitive and behavioral impacts explain program impacts on girls later outcomes.Moreover, for both genders a substantial share of the program impacts on adultMoreover, for both genders a substantial share of the program impacts on adultoutcomes is not explained by any of the observed early program impacts.outcomes is not explained by any of the observed early program impacts.Other programs provide little evidence of program impacts on childrensOther programs provide little evidence of program impacts on childrensbehavior. Demings (2009) analysis of Head Start found no short-run effects of Headbehavior. Demings (2009) analysis of Head Start found no short-run effects of Head


15/24


Start on parental reports of childrens behavior problems. Haskins (1985) reportedStart on parental reports of childrens behavior problems. Haskins (1985) reportedthat the Abecedarian program had the unexpected effect of increasing teacherthat the Abecedarian program had the unexpected effect of increasing teacherreports of childrens aggressiveness in the early school years, although these effectsreports of childrens aggressiveness in the early school years, although these effectsappeared to fade with time. Of course, these studies are vulnerable to the criticismappeared to fade with time. Of course, these studies are vulnerable to the criticismthat they did not measure a broader set of relevant skills, including students atten-that they did not measure a broader set of relevant skills, including students atten-tion or other aspects of their behavior and mental health.tion or other aspects of their behavior and mental health.Overall, reconciling disparate patterns of impacts in the short and longer termOverall, reconciling disparate patterns of impacts in the short and longer termis a key challenge for anyone hoping to extract policy lessons about the effective-is a key challenge for anyone hoping to extract policy lessons about the effective-ness of early childhood education programs. Accomplishing this task will require aness of early childhood education programs. Accomplishing this task will require aproven model of human development that incorporates various cognitive, person-proven model of human development that incorporates various cognitive, person-ality, and behavioral dimensions and can predict what kinds of children stand toality, and behavioral dimensions and can predict what kinds of children stand tobene t most from early childhood education investments.benet most from early childhood education investments.

Within-Program Heterogeneity

Although policymakers appropriately care most about the average impactsAlthough policymakers appropriately care most about the average impactsof early childhood education programs, a number of lessons can be learned fromof early childhood education programs, a number of lessons can be learned fromlooking at the distribution of treatment effects of given programs. For example,looking at the distribution of treatment effects of given programs. For example,such heterogeneity might make it possible to identify groups that could particularlysuch heterogeneity might make it possible to identify groups that could particularlybene t from preschool programs. Data on treatment-effect heterogeneity may alsobenet from preschool programs. Data on treatment-effect heterogeneity may alsoboost our understanding of human capital development processes if they identifyboost our understanding of human capital development processes if they identifygroups that particularly bene t from the preschool setting.groups that particularly benet from the preschool setting.Consider evidence from the Infant Health and Development Program (IHDP),Consider evidence from the Infant Health and Development Program (IHDP),shown in Figure 3. Beginning shortly after a childs birth, the IHDP offered ashown in Figure 3. Beginning shortly after a childs birth, the IHDP offered apackage of services that included a full-day, cognitively enriching curriculum forpackage of services that included a full-day, cognitively enriching curriculum forchildren between ages one and three, modeled after the Abecedarian program.children between ages one and three, modeled after the Abecedarian program.Nearly 1,000 children in eight sites across the country were randomly assigned to theNearly 1,000 children in eight sites across the country were randomly assigned to theIHDP treatment or to a control group that received no early childhood educationIHDP treatment or to a control group that received no early childhood educationservices but some health services (Gross, Spiker, and Haynes 1997). To be eligibleservices but some health services (Gross, Spiker, and Haynes 1997). To be eligiblefor the program, infants had to have weighed less than 2,500 grams (5.5 pounds) atfor the program, infants had to have weighed less than 2,500 grams (5.5 pounds) atbirth, but eligibility was not restricted by family income, race, or ethnicity.birth, but eligibility was not restricted by family income, race, or ethnicity.

For the economically disadvantaged children in the samplethose with familyFor the economically disadvantaged children in the samplethose with familyincome below 180 percent of the poverty line in their rst year of lifeparticipationincome below 180 percent of the poverty line in their rst year of lifeparticipationin the Infant Health and Development Program (IHDP) produced large impactsin the Infant Health and Development Program (IHDP) produced large impactson cognitive development. Speci cally, children in the treatment group outscoredon cognitive development. Specically, children in the treatment group outscoredtheir control-group counterparts by .82 standard deviations on the StanfordBinettheir control-group counterparts by .82 standard deviations on the StanfordBinetIQ mental subscale by age three.IQ mental subscale by age three.55 For children in higher-income families, theFor children in higher-income families, theIHDPs program impact was much smaller, only .18 standard deviations. Thus, ifIHDPs program impact was much smaller, only .18 standard deviations. Thus, ifdisadvantage is de ned by family income, IHDP treatment impacts heavily favoreddisadvantage is dened by family income, IHDP treatment impacts heavily favored5 This estimate comes from Duncan and Sojourner (forthcoming) and is based on weights designed tomatch the demographic characteristics of the Infant Health and Development Program sample to thoseof all US births.


16/24


disadvantaged infants. However, an alternative de nition of disadvantage can leaddisadvantaged infants. However, an alternative denition of disadvantage can leadto a different conclusion. Children disadvantaged by being born with a very lowto a different conclusion. Children disadvantaged by being born with a very lowbirth weight (less than 1500 grams or 3.3 pounds) bene ted signi cantly less from thebirth weight (less than 1500 grams or 3.3 pounds) beneted signicantly less from theIHDP intervention than advantaged heavier babies in this low-birth-weight sample.IHDP intervention than advantaged heavier babies in this low-birth-weight sample.It is not dif cult to generate possible explanations for these patterns. ForIt is not difcult to generate possible explanations for these patterns. Forexample, the income results are consistent with theories positing that the focus ofexample, the income results are consistent with theories positing that the focus ofthe Infant Health and Development Program on enriched early learning compen-the Infant Health and Development Program on enriched early learning compen-sates or substitutes for lower levels of parental investment and academic stimulationsates or substitutes for lower levels of parental investment and academic stimulationin low-income families. The differences

Documents

Investing in Preschool Programs