Early Childhood Development
and Social Mobility
W. Steven Barnett and Clive R. Belfield
Summary
Steven Barnett and Clive Belfield examine the effects of preschool education on social mobility
in the United States. They note that under current policy three- and four-year-old children
from economically and educationally disadvantaged families have higher preschool attendance
rates than other children. But current programs fail to enroll even half of poor three- and four-
year-olds. Hispanics and children of mothers who drop out of school also participate at rela?
tively low rates. The programs also do little to improve learning and development.
The most effective programs, they explain, are intensive interventions such as the model
Abecedarian and Perry Preschool programs, which feature highly qualified teachers and small
group sizes. State preschool programs with the highest standards rank next, followed by Head
Start and the average state program, which produce effects ranging from one-tenth to one-
quarter of those of the best programs. Typical child care and family support programs rank last.
Barnett and Belfield point out that preschool programs raise academic skills on average, but do
not appear to have notably different effects for different groups of children, and so do not
strongly enhance social mobility. In such areas as crime, welfare, and teen parenting, however,
preschool seems more able to break links between parental behaviors and child outcomes.
Increased investment in preschool, conclude Barnett and Belfield, could raise social mobility.
Program expansions targeted to disadvantaged children would help them move up the ladder,
as would a more universal set of policies from which disadvantaged children gained dispropor?
tionately. Increasing the educational effectiveness of early childhood programs would provide
for greater gains in social mobility than increasing participation rates alone.
The authors observe that if future expansions of preschool programs end up serving all chil?
dren, not just the poorest, society as a whole would gain. Benefits would exceed costs and there
would be more economic growth, but relative gains for disadvantaged children would be
smaller than absolute gains because there would be some (smaller) benefits to other children.
www. futureofchildren.org
W. Steven Barnett is director of the National Institute for Early Education Research, Rutgers University. Clive R. Belfield is assistant profes?
sor in the Department of Economics, Queens College, City University of New York.
VOL. 16 / NO. 2 / FALL 2006 73

W. Steven Barnett and Clive R. Belfield
Investments
in the skills of a nation s
citizens can affect both the general
level of their productivity and income
and disparities in incomes and living
standards among them. In this article
we examine how current public investments
in preschool education for U.S. children are
affecting the skills of those children gener?
ally, as well as the extent to which those in?
vestments are reducing income-related dis?
parities among them?not only during
childhood but also when they are adults. We
also consider how new investments in those
programs might affect children's skills and in?
crease social mobility.
Much research on preschool education and
children's skills has been motivated by con?
cerns about income-related disparities in
young children's language and cognitive abili?
ties, as well as other measures of their devel?
opment, including socioemotional skills.
Such disparities become evident in children
as young as age three and appear to persist?
indeed, may even widen?through the school
years.1 Researchers have examined various
preschool education programs to learn which
might best prevent or reduce these early dis?
parities so that poor children can enter
school with skills more nearly equal to those
of more highly advantaged children. How?
ever, in recent years at least, researchers have
paid less attention to an important related
question: how preschool education can en?
hance social mobility by enabling disadvan?
taged children to achieve as adults greater so?
cioeconomic success than did their parents.
That poor children begin their lives with
lower skills than those of more privileged
children is clear. Figures 1 and 2 present esti?
mates of the link between preschool chil?
dren's skills, both cognitive and social, and
the income of their families; they suggest that
skills rise evenly with family income. At pres?
ent federal, state, and local governments in
the United States fund a wide variety of early
childhood education programs that serve
many but not all children. Parents of more
advantaged children often pay privately for
various preschool programs for their chil?
dren. Although existing publicly funded pro?
grams are demonstrably raising the skills of
the children who participate in them, they
clearly have not?as figures 1 and 2 show?
broken the link between children's skills and
family income. Would increased public in?
vestment in preschool education provide ad?
ditional benefits for children in poverty and
help to improve social mobility? If so, what
form of investment would be most effective?
Some observers argue in favor of limiting in?
creased spending to programs that serve only
poor children. Others favor creating a new,
universal preschool program that would serve
all children alike. A key empirical question
related to the latter proposal is whether a
quality preschool education program for all
children would shift the entire slopes of fig?
ures 1 and 2 upward or would rotate the bot?
tom of the slopes upward while the top re?
mained anchored.
The extent to which preschool policies im?
prove the abilities of all children or reduce
disparities in learning and development will
depend on the answers to several questions.
First, to what extent do such policies alter the
distribution of preschool education opportu?
nities? Do they increase the participation of
disadvantaged children from low-income and
low-education families in effective programs?
Do they affect the participation of more ad?
vantaged children? Second, to what extent
are such programs educationally effective? A
subsidiary question is the extent to which
preschool programs may improve the abilities
of disadvantaged children relative to those of
74 THE FUTURE OF CHILDREN

Early Childhood Development and Social Mobility
Figure 1. Abilities of Entering Kindergartners, by Family Income, National Data,
Fall 1998 (reported by NIEER from ECLS-K)
Abilities cores
60
40
Second
lowest
Middle
Family income quintile
Second
highest
Reading
Math
General
knowledge
Highest
Source: W. Steven Barnett, Kirsty Brown, and Rima Shore, "The Universal vs. Targeted Debate: Should the United States Have Preschool for
All? Preschool Policy Matters 6 (New Brunswick, N.J.: NIEER, 2004).
Figure 2. Social Skills of Entering Kindergartners, by Family Income
(NIEER Analysis of ECLS-K)
Social scores
9.8
8.2
Second
lowest
Second
highest
Highest
Family income quintile
Source: See figure 1.
advantaged children. Third, to what extent do
these early effects on children's learning and
development contribute to their abilities as
they grow older, and what aspects of public
policy contribute to sustained effects? Is it
possible that these early effects may not only
be sustained throughout a lifetime but even
be passed on to later generations as they af?
fect parents' investments in children?
Participation in Early Childhood
Programs
Early childhood programs fall into three
broad types: early schooling for children from
ages three to five, interventions and child
care for children from birth to age two, and
parenting education. The coverage of the lat?
ter two is limited. Before age three, children
participate in interventions and center-based
VOL. 16 / NO. 2 / FALL 2006 75

W. Steven Barnett and Clive R. Belfield
care at quite low rates. The largest compre?
hensive child development program for chil?
dren under age three (other than early inter?
ventions for children with disabilities) is the
federal Early Head Start program, which
served fewer than 62,000 children in 2003.2
Programs for parents also have quite limited
participation. A few states?Minnesota, Mis?
souri, and Arkansas?invest in these pro-
Participation rates in parent
programs did not differ
significantly by family
characteristics such as income
and parental education.
grams more than others, but even their fund?
ing remains limited. Some programs target
economically disadvantaged families, others
do not. The Parents as Teachers program
served more than 325,000 children in
261,000 families in 2003-04, far more than
any other parent program.3 Our analyses of
data from the National Household Education
Survey (NHES) of 2001 found that just 12
percent of young children had parents who
reported participating in a parenting educa?
tion program or support group (9 percent for
parenting education alone). Participation
rates in parent programs did not differ signif?
icantly by family characteristics such as in?
come and parental education.
Among children nearing school age, on the
other hand, participation in preschool educa?
tion is increasing dramatically. In 1950 only
21 percent of five-year-olds were in school.
By 1965, 70 percent of five-year-olds at?
tended kindergarten, but only 16 percent of
four-year-olds and 5 percent of three-year-
olds were in school. Today kindergarten at?
tendance is nearly universal and 65 percent
of four-year-olds and 42 percent of three-
year-olds attend school.4 These figures, how?
ever, are based on parents' reports and thus
necessarily on parents' views about what con?
stitutes "school."
For parents of five-year-olds, "school" is al?
most entirely kindergarten, a preschool pro?
gram that has some uniformity and is prima?
rily provided in public schools. Three- and
four-year-old children, however, attend a
complex patchwork of public and private pro?
grams that go by a variety of names, including
preschool, pre-kindergarten (pre-K), four-
year-old kindergarten (4K), Head Start, child
care, day care, and nursery school. These
programs vary widely in educational intent.
Parents of three- and four-year-olds typically
report private child care provided in class?
rooms, but not child care in private homes, as
school.
Kindergarten
Some children still do not attend kinder?
garten, which is not compulsory in most
states. There is little research on why they do
not attend, though the fact that only half-day
programs are available in some communities
may be a factor for working families. Only in
the past twenty years has full-day kinder?
garten become common, with 63 percent of
children who attend kindergarten participat?
ing in a regular school day of about six hours.
The others attend half-day for two and a half
to three hours, frequently in double shifts,
some in the morning and some in the after?
noon. The distribution of full-day kinder?
garten is uneven. Of the nine states that re?
quire it, all are in the Southeast.5 Full-day
attendance is much more common for
African American children (76 percent) than
76 THE FUTURE OF CHILDREN

Early Childhood Development and Social Mobility
for white (56 percent), Hispanic (60 per?
cent), or Asian children (57 percent).6 It is
also more common among children in
poverty (63 percent) than among others (55
percent).7
Public Preschool Education
At ages three and four children attend a vari?
ety of public preschool programs. For chil?
dren in poverty, the federal government pro?
vides Head Start. State and local education
agencies also provide preschool education
programs. In addition, federal and state gov?
ernments subsidize child care, and many
children attend private child care centers
with and without public subsidies. These pro?
grams vary in their goals, resources, stan?
dards and regulation, and length of day and
year.
Head Start serves about 900,000 children,
the vast majority at ages three and four. It
serves 12 percent of children at age four, and
serves just over half of those children for two
years starting at age three. Although Head
Start targets children in poverty, self-
reported household income data on program
participation indicate that by the second half
of the school year about half the children
served are not poor but "near poor."8 The
reasons why the targeting is less than exact
include allowable exceptions to poverty in
the eligibility rules, changes in families' eco?
nomic circumstances after enrollment, and
probably some children enrolling who do not
meet the eligibility criteria. It also seems
likely that some of this apparent difference is
due to Head Starts use of family income
rather than household income to determine
eligibility. Although the overwhelming major?
ity of Head Start children are from lower-
income families, it is incorrect and mislead?
ing to simply subtract Head Start enrollment
from the total number of three- and four-
year-olds whose household income falls
below the poverty line to determine how
many poor children are not served. The fact
that poverty is a moving target presents a se?
rious challenge for education programs that
aim to serve all poor children.9
State and local governments support two
types of preschool education programs. First,
every state serves young children with dis?
abilities in the public schools, though the
percentage served varies substantially.10
States can serve children with developmental
delays in these programs, as well as those
with identified disabilities. Second, the Dis?
trict of Columbia and forty-one states also
fund preschool education for other children
(though in a few cases this is only through
supplements to Head Start). Most of these
programs target children in poverty or other?
wise at elevated risk for poor achievement
later. Oklahoma and Georgia have for several
years sought to provide preschool education
to all four-year-olds, and Oklahoma has es?
sentially achieved that goal. Florida moved to
join them in 2005, and other states have
taken steps in that direction. While state pre?
school special education programs serve chil?
dren at ages three and four, most of the
states' general preschool education programs
focus primarily or entirely on four-year-olds.
These publicly funded preschool education
programs are sometimes based in the public
schools and sometimes in private programs.
In 2004-05 state preschool programs served
6 percent of four-year-olds in special educa?
tion and 17 percent of four-year-olds, most of
them disadvantaged, in general programs, al?
though precise demographic descriptions of
the children are not available. The corre?
sponding figures for three-year-olds are 4
percent in special education and 3 percent in
other state preschool programs. Additional
VOL. 16 / NO. 2 / FALL 2006 77

W. Steven Barnett and Clive R. Belfield
children attend preschool programs in local
public schools using local or federal funds,
but no one tracks their numbers nationally.11
Child Care and Private Preschool
Education
Children also attend preschool programs paid
for publicly through federal and state child
care funds and privately by parents. State ed?
ucational standards for these programs are
minimal. The only reliable data on the num?
ber of children enrolled in all public and pri?
vate programs are provided by parental re?
ports in national surveys and the decennial
census. These data do not allow reliable
breakdowns by type of program or funding
source, because parents report virtually any
classroom as educational regardless of teacher
qualifications and educational practices, and
many children attend multiple programs or
programs that blend funding streams. Pub?
licly funded child care programs generally do
not enroll children for an entire school year
because enrollment is contingent on family
income and parental employment, which fluc?
tuate over time. Thus, while an average of
1.73 million children receive services (57 per?
cent in centers) subsidized by the Child Care
Development Fund (CCDF) each month?
roughly 225,000 at age three and 225,000 at
age four in fiscal year 2004?this does not
mean that all of them receive services contin?
uously during the calendar or school year.12
At the national level one can roughly estimate
the number of children in child care and local
public or private preschool programs by sub?
tracting from parent-reported total enroll?
ment the number of children in major public
education programs (Head Start, special edu?
cation, and regular state preschool). At age
four, about 66 percent of children attend a
center-based program of some sort. The
major public education programs account for
34 percent, leaving 32 percent in private pro?
grams or locally funded public school pro?
grams. At age three, 39 percent attend a cen?
ter-based program, and subtracting the 14
percent in major public education programs
leaves about a quarter of the population (25
percent) in child care and local private or
public preschool. Thus, most three- and four-
year-old children in a classroom are not in one
of the major public preschool education pro?
grams and most of this residual group is not
receiving a direct child care subsidy (13 per?
cent of three- and four-year-olds receive a
CCDF subsidy, but not all are in centers).13
Program Participation by Family
Background
Data from the National Household Educa?
tion Survey can be used to estimate pre?
school program participation (public and pri?
vate combined) by various family background
characteristics and to explore the determi?
nants of program participation.14 There are
striking differences in participation by in?
come, parental education, ethnicity, and re?
gion. From figure 3, it is apparent that pre?
school participation declines as income falls
until a point just below median income.
Thereafter, participation levels off or even
rises as income falls. It seems reasonable to
infer from this graph that existing public pro?
grams are already substantially increasing
preschool program participation rates among
economically disadvantaged children. NHES
data on enrollment at age four in 1991 and
2001 indicate a substantial increase over time
for children whose mothers are high school
dropouts (36 percent to 49 percent), but
these children continue to participate in pre?
school programs at lower rates than do chil?
dren of high school graduates (65 percent)
and college graduates (70 percent).15 Clearly
there is room for further equalization of ac?
cess to preschool.
78 THE FUTURE OF CHILDREN

Early Childhood Development and Social Mobility
Figure 3. Preschool Participation, by Income, 2001
Number enrolled (percent)
80
70
60
50
40
30
20
10
0
<10 10-20 20-30 30-40 40-50 50-60 60-75 75-100 >100
Family Income (thousands of dollars)
Source: W. Steven Barnett and Donald Yarosz, "Who Goes to Preschool and Why Does It Matter?" Preschool Policy Matters 7 (New
Brunswick, N.J.: NIEER, 2004).
Preschool participation rates also vary by eth?
nicity. African American children have the
highest rates, with rates for white non-
Hispanic children and Asians only slightly
lower. Hispanic children have by far the low?
est rates. Rates vary by ethnicity partly be?
cause the South provides many public pro?
grams and the West provides few. Once
family background characteristics and re?
gions are taken into account, participation
rates for Hispanic children are not signifi?
cantly lower than for white non- Hispanic
children. Rates for African American chil?
dren remain somewhat higher even after
such adjustments.16
Overall, current U.S. public policy increases
preschool participation at ages three and four
for children from economically and educa?
tionally disadvantaged families relative to
others, largely through major public educa?
tion programs. But current programs fail to
enroll even half of the children in poverty at
ages three and four, or half of the children
whose mothers are high school dropouts,
even at age four. There is thus tremendous
room for public policies to increase enroll-
ment of the most disadvantaged children in
preschool education programs. Moreover,
the programs that do serve such children?
child care and even some public education
programs?do little to improve their learning
and development. Public policies could also
do much more to increase participation rates
of children from families up to about the me?
dian income. Smaller but still substantial in?
creases in enrollment are possible for all but
the wealthiest and best educated.
Influence of Early Childhood
Programs on Child Development
and Adult Outcomes
How do current programs affect children's
eventual educational attainment, earnings,
family formation, and propensity to commit
crime? And how much more effective might
other policies be? Many researchers have ex?
amined the effects of various early childhood
education programs, with the vast majority
focusing on short-term effects on learning
and development.
There are literally hundreds of studies of the
immediate and short-term effects of child
VOL. 16 / NO. 2 / FALL 2006 79

W. Steven Barnett and Clive R. Belfield
care and early interventions, and their find?
ings have been conveniently summarized in
both quantitative meta-analyses and tradi?
tional literature reviews.17 Across these stud?
ies, the average initial effect on cognitive
abilities is about 0.50 standard deviations,
roughly equivalent to 7 or 8 points on an IQ
test with a 100-point scale and a standard de?
viation of 15. Average effects on self-esteem,
motivation, and social behavior are also posi?
tive, though somewhat smaller. In what fol?
lows, we review the best evidence to summa?
rize what is known about how various
programs?family support, child care, Head
Start, public preschool, and several very in?
tensive educational interventions (which
have yet to be implemented on a large
scale)?affect children's skills.
Family Support Programs
Although some studies produce larger esti?
mates, the most reliable research?random?
ized experimental trials?estimates that fam?
ily support programs improve both cognitive
and social development by perhaps 0.10 stan?
dard deviations.18 Randomized trials of many
home-visiting programs have failed to find
consistent effects on child development,
probably because very few of these programs
are intensive enough to produce significant
cognitive benefits for children.19 Similarly,
randomized trials of comprehensive services
delivered in
"two-generation" models?so
called because they serve both children and
parents?have disappointing findings, again
because they do not provide substantial di?
rect services to children.20
These findings support two conclusions
about program effectiveness, both of which
are consistent with other reviews of the re?
search.21 First, influencing child develop?
ment indirectly through parents appears to
be relatively ineffective. Second, a program's
effect on child development varies with the
frequency and duration of the intervention
provided. Even the most intensive family
support program devotes far fewer hours to
parents than child-directed interventions de?
vote to children. In addition, the costs of
such programs, particularly those intensive
enough to produce even modest benefits, are
substantial, thus likely making them less cost-
effective than other preschool programs.22
Despite the modest effects of most home-
visiting programs on children's cognitive de?
velopment, one very intensive program has
substantially improved the home environ?
ment and development of young children.
David Olds and colleagues found in a series
of randomized trials that a program of home
visits by nurses to economically disadvan?
taged new mothers reduced the number and
improved the timing of pregnancies and
births after the first child and also reduced
the children's need for medical care for in?
juries and ingestions.23 Other popular med?
ically oriented programs with similar goals,
however, have not been similarly effective in
randomized trials.24 Olds's nurse home-
visiting program has also been found to im?
prove modestly both the children's cognitive
development (effect size of 0.15 using the
population standard deviations for the tests)
and parents' report of behavior problems.25
Child Care and Early Education
Of all the preschool programs available to di?
rectly serve children, only center-based pro?
grams in which children attend classrooms or
individual tutoring sessions improve cognitive
development. The type and quality of activi?
ties in these programs vary tremendously. In
the best programs children are systematically,
regularly, and frequently engaged in a mix of
teacher-led and child-initiated activities that
enhance the development of language, knowl-
80 THE FUTURE OF CHILDREN

Early Childhood Development and Social Mobility
edge of concepts and skills, problem-solving
abilities, self-regulation and other socio-
emotional skills, attitudes, values, and disposi?
tions. In the worst programs, where little is
planned, children wander aimlessly, with few
interesting and thought-provoking inter?
actions, activities, or materials, and teachers
are unresponsive to their interests or needs.
To the surprise of no one, the better programs
have the better outcomes.
Studies find that typical center-based child
care (as opposed to home or other types of
care) improves cognitive abilities by about
0.10-0.33 standard deviations. Most esti?
mates are in the 0.10-0.15 range for cogni?
tive and language development.26 Evidence
is mixed on whether effects are larger when
care begins before age three.27 Some nonex-
perimental studies have found that child care
can increase antisocial behavior at school
entry, with effect sizes of about 0.08-0.20.
The evidence is mixed with respect to
whether effects are larger for disadvantaged
children than for those from more advan?
taged homes.28 Some studies have found that
higher program quality, measured in various
ways, may lead to small improvements
(0.04-0.08) in cognitive and language ability
and in behavior.29 Most child care programs,
however, facing minimal government re?
quirements and poor funding, are not de?
signed to improve child development.
By contrast, Head Start, the federal govern?
ment's largest comprehensive child develop?
ment intervention, is specifically designed to
improve children's cognitive, social, emo?
tional, and physical development, as well as
to support their parents in a variety of ways.
An excellent recent randomized trial esti?
mates, however, that one year of Head Start
has fairly small effects, from less than 0.10 to
0.24 for standardized measures of language
and cognitive abilities.30 This finding echoes
that of an Early Head Start randomized trial
in which cognitive and language effects were
about 0.10 or smaller.31 Randomized trials of
both Head Start and Early Head Start find
small decreases (about 0.10) in antisocial be?
havior. They find no evidence of negative ef?
fects on social and emotional development.
(The Head Start study did not estimate the
Of all the preschool programs
available to directly serve
children, only center-based
programs in which children
attend classrooms or individ?
ual tutoring sessions improve
cognitive development.
effect of Head Start relative to no program,
but over and above whatever experiences
children received otherwise.)
The best short-term evidence on the effects
of preschool programs sponsored by public
schools comes from relatively rigorous studies
of the Chicago Child-Parent Centers and the
universal preschool program in Tulsa, Okla?
homa. These studies have found initial effects
on standardized tests of cognitive and lan?
guage abilities ranging from 0.38 to 0.79, de?
pending on the measure. The Chicago Child-
Parent Center study found a positive effect on
social adjustment in school; the Tulsa study
did not look at social development.
The Tulsa study can be directly compared
with the Head Start randomized trial on
three tests; in each case, the Tulsa effects are
VOL. 16 / NO. 2 / FALL 2006 81

W. Steven Barnett and Clive R. Belfield
several times as large.32 As with the Head
Start study, the Tulsa study estimates effects
over and above the experiences that children
can get outside the state program, here in?
cluding Head Start and child care. And the
Tulsa study, like the Head Start study, lasted
only one year; effects might be larger if the
program lasted longer. But clearly, caution is
warranted in comparing these two studies.
There is
relatively little basis
for estimating the effects of
intensive educational
interventions on children
from middle-income or highly
advantaged families.
The Head Start study involves many more
children in more diverse circumstances, and
the comparison addresses only one program
goal (children's cognitive development). It is
plausible that the Tulsa and Chicago pro?
grams produced larger gains because their
teachers were far more highly qualified than
Head Start teachers and were also paid much
higher salaries. Whereas Head Start requires
only that half of its teachers have a two-year
degree, Tulsa and Chicago required certified
teachers with a four-year-college degree. The
Tulsa findings were recently replicated in an
evaluation of state-funded preschool pro?
grams for four-year-olds in five states, all of
which require certified teachers (Oklahoma,
New Jersey, South Carolina, Michigan, and
West Virginia).33
Researchers using data from the Early Child?
hood Longitudinal Study-Kindergarten Co?
hort (ECLS-K), a national sample of children
entering kindergarten in 1998, have found
smaller effects for prekindergarten for disad?
vantaged children?0.16 to 0.28?perhaps
reflecting the poorer performance of state-
funded preschool programs overall (many
have weaker standards than the Chicago or
Tulsa program). The ECLS-K data suggest
even smaller effects for child care, where
regulations typically require little more than
a high school diploma for teachers.34
Randomized trials of North Carolina's
Abecedarian preschool program and Michi?
gan's Perry Preschool program find that these
more intensive interventions involving disad?
vantaged children up to the age of school
entry improve cognitive and language abili?
ties from 0.75 to 1.50 standard deviations?
twice the effect of the better state preschool
programs and eight times to ten times the ef?
fect of Early Head Start and Head Start.35
These effects suggest a dose-response rela?
tionship, in which high teacher quality, small
class sizes and high teacher-pupil ratios, and
the amount of education given are all
implicated.36
The Perry Preschool study found positive ef?
fects on social behaviors similar to most stud?
ies of such effects in the first years of school.
In contrast, the Abecedarian study, which ex?
amined intensive education through full-day
child care over five years, found negative,
though transitory, effects on social behavior
at school entry. Across studies of early educa?
tion intervention, intensive research pro?
grams, and large-scale public programs, in?
cluding Head Start, short-term effects
average 0.25 to 0.40 for self-esteem, problem
behavior, and other social behaviors.37
There is relatively little basis for estimating
the effects of intensive educational interven?
tions on children from middle-income or
82 THE FUTURE OF CHILDREN

Early Childhood Development and Social Mobility
highly advantaged families. Few researchers
have addressed the topic at all, and even
fewer have done so in a rigorous way. The
only randomized trial of a preschool program
for a highly advantaged population (average
IQ was 2 standard deviations above the
mean) had a very small sample, limiting its
ability to detect effects. Nevertheless it found
modest improvements in early academic abil?
ities, at least for boys.38 The Tulsa study and
the later five-state evaluation of preschool
education provide some insights, as Okla?
homa and West Virginia both serve the gen?
eral population, not just disadvantaged chil?
dren, and the other three states also serve
populations with some socioeconomic varia?
tion. Both studies find that effects are some?
what larger for disadvantaged children.
Long-Term Effects
Though early child care and education have
positive initial effects on cognitive abilities,
those effects tend to decline over time and in
many studies are negligible several years
after children leave the programs.39 The
fade-out is most salient for general cognitive
abilities, or aptitude, as measured by IQ and
similar measures. Only the longest-lasting,
most intensive educational interventions
(year round, full day over many years), like
the Abecedarian program, seem able to pro?
duce permanent gains in general cognitive
abilities, and these appear considerably
smaller than initial gains.40 Gains on subject-
specific cognitive abilities (reading, math,
and so forth) seem to be longer lasting, and
while these more enduring gains are smaller
than the initial gains, they do not appear to
fade as often or as much as IQ gains.
Although there is essentially no research on
the very long-term effects of typical U.S.
child care on educational achievement and
attainment, there are many studies of the
long-term effects of large-scale public pre?
school education programs and intensive ed?
ucational interventions on educational
achievement and school progress.41 Esti?
mated effects on achievement have been
highly variable because of differences in re?
search methods and procedures.42 In the
more rigorous studies, which tend to examine
the more intensive educational programs, ef?
fects on achievement ranged from 0.50 to
0.75 into the high school years. The Chicago
Child-Parent Centers study suggests smaller
than average long-term achievement gains
from large-scale public programs. For Head
Start, the initial gains would suggest even
smaller long-term achievement gains. Al?
though some studies have found long-term
educational gains from Head Start, the ef?
fects tend to vary by ethnicity. The lack of
such variation in other studies raises ques?
tions about these estimates.43
Full evidence on long-term effects is re?
ported in tables 1 and 2. Studies that use cu?
mulative school records data to look at grade
repetition, special education placements, and
high school graduation provide perhaps the
strongest basis for comparing the long-term
effects of different programs. They find uni?
formly positive and statistically significant ef?
fects on school progress and placement?ef?
fects that have been causally linked to
program effects on knowledge and skills.44
In an earlier review, Steven Barnett com?
bined data from long-term studies of pre?
school program effects on grade repetition
and special education to compare the effects
of intensive interventions, Head Start, and
public school programs.45 Intensive interven?
tions had twice the effect in reducing grade
repetition (twenty-four studies) and four
times the effect in reducing special education
placement (twenty studies) of Head Start and
VOL. 16 / NO. 2 / FALL 2006 83

W. Steven Barnett and Clive R. Belfield
Table 1. Effects of Early Childhood
Interventions on Education
Change (percent except as indicated)
Intervention and educational outcome Effect
Special education placement
Abecedarian (ABC)
Perry Preschool
Chicago Child-Parent Centers
Head Start
Public School and Head Start8
Retained in grade
Abecedarian
Perry Preschool
Chicago Child-Parent Centers
Early Childhood Longitudinal
Study-Kindergarten Cohort
Public School and Head Startb
High school dropout likelihood
Abecedarian
Perry Preschool
Chicago Child-Parent Centers
High school completion
Head Start: white children
Head Start: African American children
College progression
Abecedarian enrollment in four-year college
Perry Preschool
Head Start: white children
Head Start: African American children
-48
-43
-32
-28
-29
-47
-13
-33
Negative effect
(reduces)
-30
-32
-25
-24
20 percentage
point increase
No clear effect
3 times as likely
No clear effect
28 percentage
point increase
No clear effect
Sources: Clive Belfield and others, "Cost-Benefit Analysis of the
High/Scope Perry Preschool Program Using Age 40 Follow-Up
Data," Journal of Human Resources 41 (2006): 162-91; W.
Steven Barnett, "Does Head Start Have Lasting Cognitive Effects?
The Myth of Fade Out," in The Head Start Debates, edited by Ed?
ward Zigler and Sally Syfco (Baltimore, Md.: Brookes Publishing
Co., 2004); W. Steven Barnett and Leonard Masse, "Comparative
Benefit-Cost Analysis of the Abecedarian Program and Its Policy
Implications," Economics of Education Review (in press); Arthur
Reynolds and others, "Age 21 Cost-Benefit Analysis of the Title I
Chicago Child-Parent Centers," Educational Evaluation and Policy
Analysis 24, no. 4 (2002): 267-303; Eliana Garces, Duncan
Thomas, and Janet Currie, "Longer-Term Effects of Head Start,"
American Economic Review 92 (2002): 999-1012; Janet Currie,
"Early Childhood Programs," Journal of Economic Perspectives 15
(2001): 213-38; Centers for Disease Control and Prevention,
"Community Interventions to Promote Healthy Social Environ?
ments. Early Childhood Development and Family Housing," Mor?
bidity and Mortality Weekly Report 51 (2002); Judy Temple, Arthur
Reynolds, and Wendy Miedel, "Can Early Intervention Prevent High
School Drop-Out? Evidence from the Chicago Child-Parent Cen?
ters," Urban Education 35 (2000): 31-56.
a. Nine-study average.
b. Ten-study average.
public school programs. Notably, many stud?
ies that have looked at these strong indicators
of school failure have very similar findings.
Given the small size of several of the experi?
mental studies, including Perry Preschool
and Abecedarian, the frequent replication of
their findings in these other studies strength?
ens confidence in their results.
Although fewer studies have looked at effects
on high school graduation, researchers con?
sistently find positive effects for Head Start,
public school programs, and more intensive
interventions. It is difficult to feel comfort?
able with generalizations from so few studies,
though grade repetition and special educa?
tion placement (which have been studied
much more often) are strong predictors of
dropping out of school. However, the esti?
mated effects of the three intensive programs
are quite consistent: a 15 to 20 percentage
point increase in high school graduations (not
GEDs or other substitutes), from around 50
percent to around 67 percent. The estimated
effect on high school graduation in the
Chicago study was about 10 percentage
points, roughly half that of the Perry
Preschool and Abecedarian programs. A few
studies have focused on Head Start, with in?
consistent results: one finds high school grad?
uation rates increased for girls by 15 percent?
age points, another finds a 20 percentage
point increase for whites only. Such gains
seem improbable, given the very small initial
effects found in the national impact study.
The Abecedarian study, but not the Perry
Preschool study, found gains in college en?
rollment. It is difficult to know how to inter?
pret this finding. The Perry Preschool sample
was much more educationally disadvantaged
than the Abecedarian sample. It may have
been that college was just too far beyond
their reach, given their starting abilities,
84 THE FUTURE OF CHILDREN

Early Childhood Development and Social Mobility
Table 2. Effects of Early Childhood Interventions on Adolescent and Adult Behaviors
Percent except as indicated
Intervention and behaviors Control or comparison group Group receiving early childhood program
Teenage parenting rates
Abecedarian
Perry Preschool
Chicago Child-Parent Centers
Well-being
Health problem (Perry Preschool)
Drug user (Abecedarian)
Needed treatment for addiction (Perry Preschool)
Abortion (Perry Preschool)
Abuse/neglect by age 17 (Chicago Child-Parent Centers)
Criminal ctivity
Number of felony violent assaults (Perry Preschool)
Juvenile court petitions (Chicago Child-Parent Centers)
Booked or charged with a crime (Head Start)
Net earnings gain from participating in early childhood programs
Abecedarian
Perry Preschool Program
Chicago Child-Parent Centers
Head Start
45
37
27
29
39
34
38
9
0.37
25
$35,531
$38,892
$30,638
No effect
26
26
20
20
18
22
16
6
0.17
16
12 percentage points lower
Sources: Belfield and others (see table 1); Masse and Barnett (see table 1); Arthur Reynolds and others (see table 1); Garces and others
(see table 1); Currie (see table 1); Centers for Disease Control and Prevention (see table 1).
whereas the Abecedarian children were close
enough that the boost they received made
college possible for a significant share.
Although relatively few in number, most
studies that assessed long-term effects on so?
cial behavior found positive (though not al?
ways statistically significant) effects, and no
study reported increased aggression beyond
first grade.46 Five studies of educational
interventions that investigated long-term
effects on social behavior found beneficial ef?
fects on classroom behavior, social adjust?
ment, and crime.47 These include two of the
three studies that linked elevated aggression
with full-time child care that began in in?
fancy.48 The third, the Abecedarian study,
found no long-term effect on crime and
delinquency, though rates were relatively low
for both groups.49 The strongest effects on
crime were found in the Perry Preschool
study, where baseline rates for the control
group were quite high: the number of arrests
was cut by 50 percent. In the Chicago study,
the number of arrests by age eighteen was
cut by 40 percent, while the share of people
ever arrested was cut by a third (or 8 percent?
age points), from 25 percent to 17 percent.
Data on Head Start are limited to two stud?
ies: one finds a 12 percentage point reduction
in crime for African Americans only; the
other, a 10 percentage point reduction for
girls only.50
There is little research on the effects of pre?
school programs on later fertility behavior.
VOL. 16 / NO. 2 / FALL 2006 85

W. Steven Barnett and Clive R. Belfield
The model programs show strong effects, and
family support interventions have reported
direct effects on fertility behavior of the
mothers. Effects are reported in table 2.
Finally, direct effects have been found on
employment and earnings. One study found
that Head Start raised earnings, but only for
white children whose parents were high
school dropouts. The model program effects,
A reasonable conclusion is
that
auspices per se have
little to do with program
effectiveness, once goals,
standards, and resources are
taken into account.
shown in table 2, may be considered upper
bounds on the earnings gain from state-
funded preschool.
Program Design and Effectiveness
From the evidence reviewed so far, it should
be clear that some preschool programs are
more effective than others. A rough ranking
from least to most educationally effective
under current policies is typical child care
and family support programs, Head Start and
many state preschool programs, state pre?
school programs with high standards (far
from all of them), and intensive educational
interventions. On average, state preschool
programs differ little from Head Start in
their effects on child development, but states
with lower standards likely have worse out?
comes and those with higher standards, bet?
ter outcomes. A reasonable conclusion is that
auspices per se have little to do with program
effectiveness, once goals, standards, and re?
sources are taken into account. The pattern is
clearest for short-term outcomes, where the
most data are available. It is less clear for
long-term effects on educational attainment
and adult social and economic outcomes,
where fewer data are available. It does not
seem plausible that programs with very weak
initial effects would have proportionately
larger effects on adult outcomes than on
short- and medium-term outcomes.
Given the limits of the data, it appears best to
produce a range of estimates of the programs'
effects on cognitive and social-emotional de?
velopment. An upper bound would be effects
of the size produced by the Perry Preschool
and Abecedarian programs. One then might
expect high-quality interventions in public
preschool programs to produce effects of half
that size. Less educationally intensive public
programs, including Head Start under cur?
rent policies, would be likely to produce ef?
fects of one-quarter or less, and possibly only
one-tenth, of those of Perry and Abecedarian.
Regarding effects on children who are not
disadvantaged, based on the meager evidence
we consider two different scenarios. One is
that effects are half those estimated for disad?
vantaged children. The other is that effects on
the educational attainment of advantaged
children are essentially zero. Given the small
effects of child care, if programs effectively
target disadvantaged children, then effects on
other children are irrelevant.
In designing policy proposals to improve pre?
school programs, it should be kept in mind
that the most effective educational interven?
tions were more intensive in two senses.
First, they had highly qualified, well-paid
teachers and high ratios of teachers to chil?
dren. Second, some provided a large number
of hours of intervention over two to nearly
86 THE FUTURE OF CHILDREN

Early Childhood Development and Social Mobility
five years. The Perry program provided one
teacher (not an assistant) for every six or
seven students. Although it operated only
half-day during the school year (and most,
but not all, children attended for two years),
teachers visited each child at home weekly.
The Abecedarian program had a teacher and
an aide for every twelve children and oper?
ated for up to ten hours a day, fifty weeks a
year, over almost five years. This pattern can
hardly be considered surprising and is consis?
tent with other evidence. It posits that more
highly educated, better prepared, better su?
pervised, and better compensated teachers
are more effective.51 Smaller class sizes and
better teacher-student ratios result in better
teaching and more individual attention,
which produce larger gains in achievement
and school success.52 Finally, more hours of
effective interventions produce larger effects.
The Effects of Early Childhood
Education on Social Mobility
The above evidence on access and outcomes
suggests the following conclusions about the
extent to which preschool, as it now stands,
affects social mobility by breaking down the
links between parental socioeconomic status
and behaviors and children's status and
behaviors.
Although current public programs move in
the direction of equalizing preschool oppor?
tunities across races and income levels, they
fall considerably short of their goal.
Preschool opportunities are not close to
equal for Hispanic children. Nor are pre?
school opportunities equal when mother s ed?
ucation is considered, or when the quality of
the different programs is accounted for, or
when children aged three as well as four are
included. Furthermore, Head Start funding
is so limited that it precludes serving most of
the eligible population, and public preschool
program coverage varies greatly from state to
state. Thus many opportunities exist for ex?
panding preschool, but the form of that ex?
pansion is critical, as we discuss below.
In addition, broader questions might be
raised about the extent to which current pre?
school programs integrate social groups.
Given the separation of children in Head
Start and other compensatory programs,
preschool programs do not appear to be
structured so as to allow disadvantaged chil?
dren to benefit from long-term exposure to
other children. And where preschool pro?
grams are tied to local public schools, resi?
dential patterns may also limit socioeco?
nomic integration.
Preschool Outcomes and Social Mobility
Preschool may enhance social mobility if it
affects children of different races or income
levels in different ways. Based on the ob?
served effects of preschool, one might expect
increased social mobility across various do?
mains. One domain is earnings: if preschool
raises incomes most for those in the lowest
earning deciles, then it may increase social
mobility. As table 2 shows, model programs
do yield reasonable earnings advantages of
approximately $30,000 (in current dollars), a
little less than 10 percent of the lifetime
earnings of a high school dropout. Current
public programs, however, are not as effec?
tive as these model programs, and $30,000 is
the advantage compared to control groups
with no preschooling (that is, it is the effect
of the preschooling, not the difference in ef?
fect on earnings for different groups). The
second domain is education, the focus of
most research. Preschool does indeed raise
achievement. However, current programs are
unlikely to have strongly different effects on
educational attainment for different groups
of children.53
VOL. 16 / NO. 2 / FALL 2006 87

W. Steven Barnett and Clive R. Belfield
The effects of preschool in other domains,
however, appear more conducive to social
mobility. For example, disadvantaged chil?
dren are more likely to engage in crime, be
on welfare, and become teenage parents;
they are also more likely to report ill health.54
Here there is more scope for preschool to
break the link between family behaviors and
child outcomes.
In those areas where
preschool may raise social
mobility the most?criminal
tendency, welfare receipt,
and fertility?it may also
benefit the children of the
preschoolers.
For welfare participation, children may be
"scarred" by their parents' receipt of welfare:
family receipt of welfare may cause poorer
labor market outcomes, break down social
norms against welfare support, or increase
awareness of welfare eligibility, all of which
would perpetuate welfare dependency for in?
dividuals and within families. Researchers
cannot precisely identify welfare heritability,
but they generally find that when parents,
particularly African American women, re?
ceive welfare, their children are more likely
to receive welfare.55 Preschool programs may
therefore raise social mobility by reducing
welfare reliance heritability, although the size
of the effect is questionable (not least be?
cause welfare is increasingly time limited).
Recent evidence also indicates reasonably
strong "heritability" of criminal activity, par?
ticularly for men. Among noncriminals, 6
percent have fathers who were arrested;
among criminals the figure is 15 percent. In
the specific case of partner violence, the Na?
tional Youth Survey shows a strongly positive
correlation between family violence and later
partner violence.56 A full meta-analysis, how?
ever, finds that "violent origins have only a
weak-to-moderate effect on the risk of later
partner violence."57 Most families report zero
crime by the parent and zero crime by the
offspring. This means that the second-
generational elasticity of crime is hard to esti?
mate. Given the limited number of families
that appear to transmit crime from one gen?
eration to another, preschooFs overall effects
on this link cannot but be small. But to the
extent that crime rates are much higher in
some communities than others, preschool
programs might have more of an equalizing
effect by reducing crime rates in higher-
crime-rate communities.
Preschool may have its strongest differential
effect on fertility. The first-generation effect
is to reduce teenage parenting, which is cor?
related with low economic well-being.58
Given that teenage parenting rates are higher
for women with low education and low in?
come and for African American and Hispanic
women, preschool should have relatively
greater benefits for poor and minority chil?
dren.59 (Women who participated in the
Perry Preschool program as children were
also considerably less likely to have an abor?
tion, suggesting that preschool enhances par?
enthood planning.) There may also be sec?
ond-generation effects.
Second-Generation Effects of Preschool
In those areas where preschool may raise so?
cial mobility the most?criminal tendency,
welfare receipt, and fertility?it may also
benefit the children of the preschoolers.
Given the diffuse benefits of preschool, and
THE FUTURE OF CHILDREN

Early Childhood Development and Social Mobility
the reasonably strong heritability of behav?
iors and circumstances, these second-genera?
tion effects may be a key to social mobility60
The effects on fertility and crime, in particu?
lar, may spill over into the second generation.
Research has documented that teenage par?
enting and single parenting adversely affect
children's attainment.61 Children of teenage
parents are much less likely to graduate from
high school, and a child in a two-parent fam?
ily accumulates on average 0.43 more years
of schooling than a child in a single-parent
family62 Given conventional estimates of the
returns to a year of education, the benefits
from residing in a two-parent household are
the equivalent of a 2-4 percent increase in
annual earnings. Two-parent family status
and smaller family size also reduce criminal
activity, while children of teenage parents are
more likely to engage in crimes such as as?
sault.63 The effects appear to be significant, if
only because any reduction in criminal activ?
ity conveys substantial economic benefits.
Other second-generation effects are probably
weaker. There may be some effect on sec?
ond-generation earnings, to the extent that
preschool weakens the link between parents'
and children's incomes.64 Finally, preschool
may affect educational attainment across
generations. Both mother's and father's edu?
cation are statistically significant influences
on a child's graduation and years of school?
ing.65 One extra year of parental schooling is
associated, on average, with 0.29 years of off?
spring attainment.66
Although these arguments are plausible,
there is no direct evidence on the benefits to
subsequent generations from either state or
model preschool programs. (Because the
sample sizes in the model programs are so
small, it is typically not possible to identify
second-generation effects.)67 Moreover, be?
cause such benefits would be a long time in
the future, they would need to be discounted
(valued less relative to immediate benefits).
Applying a social discount rate of 3.5 percent,
we find that any monetary gain for a child is
worth half that of a gain in the same domain
to the actual participant. So, even with per?
fect heritability, the effects on social mobility
are half as strong for the second generation.
In summary, there is some evidence that di?
rect and indirect heritability effects are sig?
nificant, though there is insufficient research
from which to generalize to an anticipated ef?
fect of participation in early childhood educa?
tion programs. A recent simulation model,
however, suggests that these effects are
meaningful.
Diego Restuccia and Carlos Urrutia generate
a four-period model of parent-child invest?
ments to determine social mobility across
generations, contingent on increased public
spending on elementary and secondary edu?
cation, and separately, on higher education.68
In their policy simulations, they find that in?
creased spending on elementary and second?
ary education (which can include prekinder-
garten and kindergarten) raises social
mobility. The logic is relatively straightfor?
ward. Increased public spending on the early
years of schooling?in the model, the in?
creased spending is used for a universal pro?
gram of preschooling for all children, regard?
less of family background?eases the burden
of borrowing for educational investments for
poorer families (although it also motivates
some wealthier parents to switch from pri?
vate to public schools). The children of
poorer families will then go on to college, and
although they will drop out at relatively high
rates, the children who finish will increase
the number of college graduates from low-
VOL. 16 / NO. 2 / FALL 2006 89

W. Steven Barnett and Clive R. Belfield
income backgrounds. In the model, intergen?
erational earnings and education correlations
both fall as a result. Assuming an increase in
public spending on early education of $90
billion?sufficient to fund preschool for all
children for approximately two years?earn?
ings correlations across generations should
fall from 0.40 to 0.36 (a perfect correlation
would be 1, no correlation at all would be 0)
Because
many low-income
and
minority children are
already enrolled in Head
Start and other programs,
another
way to raise social
mobility would be to upgrade
the
existing program.
and education correlations across generations
should fall from 0.35 to 0.28.
Relative to other educational investments in
the model, these effects are substantial.
Spending on higher education in the model,
for example, has zero or even a negative ef?
fect on these earnings correlations: subsidies
awarded to a college student do not greatly
affect the student's ability to graduate from
college.69 However, in this model the spend?
ing on early education would do little to raise
educational attainment (college enrollment
and completion) for the lowest income quin?
tile. Its main effect would be to equalize col?
lege enrollment rates for the three middle
quintiles of family income.
Targeted or Universal Preschooling?
The above discussion assumes a trend toward
universal preschooling, or at least that any
program expansion would be distributed in
the same way as the present system. In part,
that assumption reflects widespread political
support for universal programs and the prac?
tical challenges of more accurately targeting
programs to the disadvantaged. A universal
program should still reduce inequalities, be?
cause it benefits low-income and minority
children more than it does advantaged chil?
dren, but the effects (especially at current
quality levels) on relative socioeconomic po?
sition may not be strong.
In theory, programs targeted at the most dis?
advantaged children would increase social
mobility the most. A targeted program would
obviously generate benefits for those who en?
rolled. Indeed, existing public preschool pro?
grams do raise social mobility. But many chil?
dren who would enroll in a new targeted
program would either be white non-Hispanic
or Hispanic or be in the lower-middle quar?
tile of income distribution. Thus Hispanic
children might gain more than African Amer?
ican children (who have much higher pre?
school participation rates), and children in
poverty would not benefit much more than
children in families with higher but still mod?
est incomes. Moreover, it may not be easy to
identify the enrollees who might benefit most
from a targeted program (particularly chil?
dren of mothers who are high school
dropouts) and exhort them to participate.70
Screening, regulating, and monitoring eligi?
bility would also raise unit costs. With imper?
fect targeting, many disadvantaged children
would miss out on programs. If the chal?
lenges of targeting could be overcome, how?
ever, social mobility effects might be greater.
Because many low-income and minority
children are already enrolled in Head Start
and other programs, another way to raise so?
cial mobility would be to upgrade the exist-
90 THE FUTURE OF CHILDREN

Early Childhood Development and Social Mobility
ing program. Janet Currie and Matthew Nei-
dell have found that increased spending on
Head Start does appear to enhance out?
comes.71 Also, state programs (most of which
are funded at rates below Head Start) might
be upgraded. Here the challenge is to get
sufficient resources for high-quality targeted
programs. Another option is to expand Head
Start and state programs to serve all children
for two years, which would generate stronger
effects. At present most children attend such
programs for only one year.
The dilemma is the old efficiency-equity
trade-off. A targeted program would have a
greater impact on social mobility, but it
would not generate as high a public return on
investment as a universal program.72 If a pro?
gram targeted to the lowest quintile is only
50 percent accurate?that is, if half of the
participants are not from the lowest quin?
tile?then it would generate smaller returns
than a universal program (even as the average
benefits from such a program would be sig?
nificantly lower). Universal programs are also
much more likely to garner political support,
as well as generate spillover benefits such as
better school discipline. And any fiscal sav?
ings these programs yield will be passed on to
taxpayers. Thus a useful strategy for increas?
ing social mobility might be to target within a
universal system by providing more intensive
programs, with smaller classes and longer
hours, to disadvantaged children. However,
the amount of extra resources needed to
yield sufficient social mobility cannot be eas?
ily specified.
Conclusions
U.S. preschool programs are effective across a
wide set of outcomes. But participation rates
are lower for children with lower incomes and
low parental education, for Hispanics, and for
those residing in the western states than for
other children. Together, these facts suggest
that increased investment in preschool could
raise social mobility. Program expansions tar?
geted to disadvantaged children would help
them move up the ladder, as would a more
universal set of policies from which disadvan?
taged children gained disproportionately. In?
creasing the educational effectiveness of
these programs would provide for greater
gains in social mobility than would increasing
participation rates alone. At the same time,
expectations of what can be accomplished on
this front should be modest.
Under current policies, preschool participa?
tion rates are not vastly different across races
and income levels. Future expansions may
end up serving all children, not just the poor?
est. In this scenario, society as a whole would
gain. Benefits would exceed costs, and there
would be more economic growth and thus
more upward mobility, but not necessarily
substantially greater opportunities for those
at the bottom of the economic ladder.
VOL. 16 / NO. 2 / FALL 2006 91

W. Steven Barnett and Clive R. Belfield
Notes
1. See Jack Shonkoff and Deborah Phillips, From Neurons to Neighborhoods: The Science of Early Childhood
Development (Washington: National Academy Press, 2000); Donald Rock and A. Jackson Stenner, "Assess?
ment Issues in the Testing of Children at School Entry," Future of Children 15, no. 1 (2005): 15-34.
2. Administration for Children and Families, "Head Start Program Fact Sheet: Fiscal Year 2006" (2006),
www.acf.hhs.gov/programs/hsb/research/2006.htm (accessed April 13, 2006).
3. Parents as Teachers, "2003-04 Parents as Teachers Born to Learn Annual Program Report Summary"
(2004), www.parentsasteachers.org/atf/cf/%7B00812ECA-A71B-4C2C-8FF3-8F16A5742EEA%7D/PAT%20
NEWS%20FINAL%20VERSION%20-%20National%20Annual%20Program%20Report%20Summary.pdf
(accessed April 13, 2006).
4. W. Steven Barnett and Donald J. Yarosz, "Who Goes to Preschool and Why Does It Matter?" Preschool
Policy Matters 7 (New Brunswick, N.J.: National Institute for Early Education Research, 2004).
5. Debra J. Ackerman, W Steven Barnett, and Kenneth Robin, "Making the Most of Kindergarten: Present
Trends and Future Issues in the Provision of Full-Day Programs," NIEER Policy Report (March 2005),
http://nieer.org/docs/index.php?DocID=118 (accessed April 13, 2006).
6. John Wirt and others, The Condition of Education 2004: 111 (U.S. Department of Education, 2004).
7. Jill Walston and Jerry West, Full-Day and Half-Day Kindergarten in the United States: Findings from the
Early Childhood Longitudinal Study, Kindergarten Class of 1998-99 (Washington: National Center for
Education Statistics, 2004), p. 73. There are some surprising differences between this study and the Cur?
rent Population Survey in estimates of full-day kindergarten attendance by ethnicity and other background
characteristics of the children.
8. We analyzed data from the 2001 National Household Education Survey for reported Head Start atten?
dance by income. Although there appears to be substantial error in reported Head Start participation,
those whose participation can be verified do not significantly differ in income from those whose participa?
tion cannot be verified. Mary Hagedorn and others, National Household Education Surveys Program of
2001: Data Files and Electronic Codebook, NCES 2003078 (U.S. Department of Education, 2003).
9. W Steven Barnett, Kirsty Brown, and Rima Shore, "The Universal vs. Targeted Debate: Should the United
States Have Preschool for All?" Preschool Policy Matters 6 (New Brunswick, N.J.: National Institute for
Early Education Research, 2003).
10. W Steven Barnett and others, The State of Preschool: 2005 State Preschool Yearbook (New Brunswick,
N.J.: NIEER, 2006).
11. Ibid.
12. Administration for Children and Families, "FFY 2004 CCDF Data Tables and Charts (Preliminary Esti?
mates)" (2005), www.acf.hhs.gov/programs/ccb/research/04acf800/list.htm (accessed April 13, 2006).
13. Barnett and Yarosz, "Who Goes to Preschool?" (see note 4); Barnett and others, The State of Preschool (see
note 10).
14. Hagedorn and others, National Household Education Surveys Program of 2001 (see note 8).
92 THE FUTURE OF CHILDREN

Early Childhood Development and Social Mobility
15. Barnett and Yarosz, "Who Goes to Preschool?" (see note 4).
16. Ibid.
17. Ruth McKey and others, The Impact of Head Start on Children, Families, and Communities (Washington:
Head Start Evaluation Synthesis and Utilization Project, 1985); Craig Ramey, Donna Bryant, and Tanya
Suarez, "Preschool Compensatory Education and the Modifiability of Intelligence: A Critical Review," in
Current Topics in Human Intelligence, edited by Douglas Detterman (Norwood, N.J.: Ablex, 1985), pp.
247-96; Michael Guralnick and Forrest C. Bennett, The Effectiveness of Early Intervention for At-Risk and
Handicapped Children (New York: Academy Press, 1987); Michael Guralnick, "Second Generation Re?
search on the Effectiveness of Early Intervention," Early Education and Development 4, no. 4 (1993):
366-78; Jack Shonkoff and Samuel Meisels, Handbook of Early Childhood Intervention (Cambridge Uni?
versity Press, 2000).
18. Deanna Gomby, Patti Culross, and Richard Behrman, "Home Visiting: Recent Program Evaluations?
Analysis and Recommendations," Future of Children 9, no.l (1999): 4-26.
19. Amy Baker, Chaya Piotrkowski, and Jeanne Brooks-Gunn, "The Home Instruction Program for Preschool
Youngsters (HIPPY)," Future of Children 9, no. 1 (1999): 116-33; Gloria Boutte, "The Effects of Home In?
tervention on Rural Children's Home Environments, Academic Self-Esteem, and Achievement Scores"
(Ann Arbor, Mich.: UMI Dissertation Services, 1992); Mary Wagner and Serena Clayton, "The Parents as
Teachers Program: Results from Two Demonstrations," Future of Children 9, no. 1 (1999): 91-116; Sandra
Scarr and Kathleen McCartney, "Far from Home: An Experimental Evaluation of the Mother-Child Home
Program in Bermuda," Child Development 59 (1988): 531-43; Christine Powell and Sandra Grantham-
McGregor, "Home Visiting of Varying Frequency and Child Development," Pediatrics 84 (1989): 157-64.
20. Robert St. Pierre and Jean Layzer, "Using Home Visits for Multiple Purposes: The Comprehensive Child
Development Program," Future of Children 9, no.l (1989): 134-51.
21. Glenn Casto and Anne Lewis, "Parent Involvement in Infant and Preschool Programs," Journal of the Di?
vision of Early Childhood 9 (1984): 49-56; Donna Bryant and Craig Ramey, "Prevention-Oriented Infant
Education Programs," Journal of Children in Contemporary Society 7 (1987): 17-35.
22. Robert St. Pierre, Jean Layzer, and Helen Barnes, "Regenerating Two-Generation Programs," in Early
Care and Education for Children in Poverty: Promises, Programs, and hong-Term Results, edited by W
Steven Barnett and Sarane Boocock (SUNY Press, 1998), pp. 99-121.
23. David Olds and others, "Prenatal and Infancy Home Visitation by Nurses: Recent Findings," Future of
Children 9, no.l (1999): 44-66.
24. Deanna Gomby, Home Visitation in 2005: Outcomes for Children and Parents (Washington: Committee for
Economic Development, 2005), www.ced.org/projects/kids.shtml (accessed February 12, 2006).
25. David Olds and others, "Effects of Nurse Home-Visiting on Maternal Life Course and Child Development:
Age 6 Follow-Up Results of a Randomized Trial," Pediatrics 114 (2004): 1550-59.
26. Sandra Scarr, Marlene Eisenberg, and Kirby Deater-Deckard, "Measurements of Quality in Child Care
Centers," Early Childhood Research Quarterly 9, no 2 (1994): 131-52; Sandra Hofferth, "Child Care in the
First Three Years of Life and Preschoolers' Language and Behavior," paper presented at the biennial meet?
ing of the Society for Research in Child Development, Albuquerque, N.M., April 1989; Katherine Magnu-
VOL. 16 / NO. 2 / FALL 2006 93

W. Steven Barnett and Clive R. Belfield
son, Chris Ruhm, and Jane Waldfogel, "Does Prekindergarten Improve School Preparation and Perfor?
mance?" Economics of Education Review (2006, forthcoming); NICHD Early Child Care Research Net?
work and Greg Duncan, "Modeling the Impacts of Child Care Quality on Children's Preschool Cognitive
Development," Child Development 74, no. 5 (2003): 1454-75.
27. Bengt-Erik Andersson, "Effects of Public Day Care?A Longitudinal Study," Child Development 60
(1989): 857-66; Tiffany Field, "Quality Infant Day Care and Grade School Behavior and Performance,"
Child Development 62 (1991): 863-70.
28. Sonalde Desai, P. Lindsay Chase-Lansdale, and Robert Michael, "Mother or Market? Effects of Maternal
Employment on the Intellectual Ability of Four-Year-Old Children," Demography 26 (1989): 545-61; Nazli
Baydar and Jeanne Brooks-Gunn, "Effects of Maternal Employment and Child-Care Arrangements on
Preschoolers' Cognitive and Behavioral Outcomes: Evidence from the Children of the National Longitudi?
nal Survey of Youth," Developmental Psychology 27 (1991): 932-45.
29. NICHD Early Child Care Research Network, "Early Child Care and Children's Development Prior to
School Entry" and "Further Explorations of the Detected Effects of Quantity of Early Child Care on So-
cioemotional Development," both papers presented at the biennial meeting of the Society for Research in
Child Development, Minneapolis, Minn., April 2001.
30. Eliana Garces, Duncan Thomas, and Janet Currie, "Longer-Term Effects of Head Start," American Eco?
nomic Review 92 (2002): 999-1012.
31. John Love and others, Early Head Start Research?Ruilding Their Futures: How Early Head Start Pro?
grams Are Enhancing the Lives of Infants and Toddlers in Low-Income Families (Princeton, N.J.: Mathe-
matica Policy Research, 2001).
32. William Gormley and others, "The Effects of Universal Pre-K on Cognitive Development," Developmental
Psychology 41, no. 6 (2005): 533-58.
33. W Steven Barnett, Cynthia Lamy, and Kwanghee Jung, "The Effects of State Prekindergarten Programs on
Young Children's School Readiness in Five States," http://nieer.org/docs/index.php?DocID=129 (accessed
February 13, 2006).
34. Studies relying on the ECLS-K, however, should be viewed cautiously as researchers must infer program
types from parents' descriptions ("prekindergarten" may include some ordinary child care) and have lim?
ited means for adjusting for the reasons why parents select programs. The ECLS-K data also suffer from
attrition of test score information over time.
35. Frances Campbell and others, "The Development of Cognitive and Academic Abilities: Growth Curves
from an Early Childhood Educational Experiment," Developmental Psychology 37, no. 2 (2001): 231-42;
Larry Schweinhart and others, Lifetime Effects: The High/Scope Perry Preschool Study through Age 40,
Monographs of the High/Scope Educational Research Foundation 14 (Ypsilanti, Mich.: High/Scope Edu?
cational Research Foundation, 2005).
36. In studies from before 1985, the estimates might be somewhat larger because the control group had little
access to alternative services. But even in the Abecedarian study by Barnett and Masse (see note 49), the
control group had considerable access to center-based child care, so that change in control group experi?
ence is unlikely to have much influence on comparisons to more recent studies.
94 THE FUTURE OF CHILDREN

Early Childhood Development and Social Mobility
37. W S. Barnett, "Early Childhood Education," in School Reform Proposals: The Research Evidence, edited
by Alex Molnar (Greenwich, Conn.: Information Age Publishing, 2002), pp. 1-26.
38. Jean Larsen and Clyde Robinson, "Later Effects of Preschool on Low-Risk Children," Early Childhood Re?
search Quarterly 4 (1989): 133-44.
39. Ramey, Bryant, and Suarez, "Preschool Compensatory Education" (see note 17); Martin Woodhead,
"When Psychology Informs Public Policy: The Case of Early Childhood Intervention," American Psycholo?
gist 43 (1988): 443-54; Ron Haskins, "Beyond Metaphor: The Efficacy of Early Childhood Education,"
American Psychologist 44 (1989): 274-82; Charles Locurto, "Beyond IQ in Preschool Programs?" Intelli?
gence 15 (1991): 295-312; Herman Spitz, "Commentary on Locurto s 'Beyond IQ in Preschool Programs?"'
Intelligence 15 (1991): 327-33.
40. Frances Campbell and Craig Ramey, "Cognitive and School Outcomes for High-Risk African American
Students at Middle Adolescence: Positive Effects at Early Intervention," American Educational Research
Journal 32 (1995): 743-72; Jeanne Brooks-Gunn, Jane Waldfogel, and Wen-Jui Han, "Maternal Employ?
ment and Child Outcomes in the First Three Years of Life: The NICHD Study of Early Childcare," Child
Development 73 (2002): 1052-72.
41. W. Steven Barnett, "Long-Term Effects on Cognitive Development and School Success," in Early Care
and Education for Children in Poverty: Promises, Programs, and Long-Term Outcomes, edited by W
Steven Barnett and Sarane Boocock (SUNY Press, 1998), pp. 11-44; Janet Currie, "Early Childhood Pro?
grams," Journal of Economic Perspectives 15 (2001): 213-38.
42. Anne McGill-Franzen and Richard Allington, "Flunk 'em or Get Them Classified: The Contamination of
Primary Grade Accountability Data," Educational Researcher 22, no. 1 (1993): 19-22.
43. W. Steven Barnett and Greg Camilli, "Compensatory Preschool Education, Cognitive Development, and
'Race,'" in Race and Intelligence: Separating Science from Myth, edited by Jeff Fish (Mahwah, N.J.:
Lawrence Erlbaum Associates, 2002), pp. 368-406.
44. W. Steven Barnett, John Young, and Larry Schweinhart, "How Preschool Education Influences Long-Term
Cognitive Development and School Success," in Early Care and Education for Children in Poverty, edited
by W Steven Barnett and Sarane Boocock (SUNY Press, 1998).
45. W. Steven Barnett, "Preschool Education for Economically Disadvantaged Children: Effects on Reading
Achievement and Related Outcomes," in Handbook of Early Literacy Research, edited by Susan Neuman
and David Dickinson (New York: Guilford Press, 2001), pp. 421-43.
46. W. Steven Barnett, "Long-Term Effects of Early Childhood Programs on Cognitive and School Outcomes,"
Future of Children 5, no. 3 (1995): 25-50; Hiro Yoshikawa, "Prevention as Cumulative Protection: Effects
of Early Family Support and Education on Chronic Delinquency and Its Risks," Psychological Bulletin 115
(1994): 27-54.
47. Arthur Reynolds and others, "Long-Term Effects of an Early Childhood Intervention on Educational
Achievement and Juvenile Arrest: A 15-year Follow-Up of Low-Income Children in Public Schools," Jour?
nal of the American Medical Association 285 (2001): 2339-46; Dale Johnson and Todd Walker, "A Follow-
Up Evaluation of the Houston Parent-Child Development Center: School Performance," Journal of Early
Intervention 15, no. 3 (1991): 226-36; Clive Belfield and others, "Cost-Benefit Analysis of the High/Scope
Perry Preschool Program Using Age 40 Follow-Up Data." Journal of Human Resources 41 (2006): 162-91.
VOL. 16 / NO. 2 / FALL 2006 95

Steven Barnett and Clive R. Belfield
48. Victoria Seitz and Nancy Apfel, "Parent-Focused Intervention: Diffusion Effects on Siblings," Child Devel?
opment 56 (1994): 376-91.
49. W Steven Barnett and Leonard Masse, "Comparative Benefit-Cost Analysis of the Abecedarian Program
and Its Policy Implications," Economics of Education Review (2006, in press).
50. Garces, Thomas, and Currie, "Longer-Term Effects of Head Start" (see note 30).
51. Richard Murnane and Barbara Phillips, "What Do Effective Teachers of Inner-City Children Have in
Common?" Social Science Research 10 (1981): 83-100; Ronald Ferguson, "Can Schools Narrow the Black-
White Test Score Gap?" in The Black-White Test Score Gap, edited by Christopher Jencks and Meredith
Phillips (Brookings, 1998), pp. 318-74; Alison Clarke-Stewart, Christian Gruber, and Linda Fitzgerald,
Children at Home and in Day Care (Hillsdale, N.J.: Lawrence Erlbaum Associates, 1994); Carollee Howes
and Michael Olenick, "Child Care and Family Influences on Toddlers' Compliance," Child Development
57 (1986): 202-16; Marcy Whitebook, Deborah Phillips, and Carollee Howes, National Child Care Staffing
Study Revisited: Four Years in the Life of Center-Based Child Care (Oakland, Calif.: Child Care Employee
Project, 1993); Carollee Howes, "Children's Experiences in Center-Based Child Care as a Function of
Teacher Background and Adult-Child Ratio," Merrill-Palmer Quarterly 43, no. 3 (1997): 404-25; Leslie
Phillipsen and others, "The Prediction of Process Quality from Structural Features of Child Care," Early
Childhood Research Quarterly 12 (1997): 281-304; Barbara Bowman, M. Suzanne Donovan, and Susan
Burns, eds., Eager to Learn: Educating Our Preschoolers (Washington: National Academy Press, 2001).
52. Harry McGurk and others, Staff-Child Ratios in Care and Education Services for Young Children (Lon?
don: HMSO, 1995); Jean Layzer, Barbara Goodson, and Marc Moss, Life in Preschool?Volume One of an
Observational Study of Early Childhood Programs for Disadvantaged Four-Year-Olds: Final Report (Cam?
bridge, Mass.: Abt Associates, 1993); Susan Kontos, Carollee Howes, and Ellen Galinsky, "Does Training
Make a Difference to Quality in Family Child Care?" Early Childhood Research Quarterly 12 (1997):
351-72; Ann Smith, "Quality Child Care and Joint Attention," International Journal of Early Years Educa?
tion 7, no. 1 (1999): 85-98; Charles Achilles, Patrick Harman, and Paula Egelson, "Using Research Results
on Class Size to Improve Pupil Achievement Outcomes," Research in the Schools 2, no. 2 (1995): 23-30;
Harold Wenglinsky, "How Money Matters: The Effect of School District Spending on Academic Achieve?
ment," Sociology of Education 70, no. 3 (1997): 377-99; Fred Mosteller, "The Tennessee Study of Class
Size in the Early School Grades," Future of Children 5, no. 2 (1995): 113-27; Jeremy Finn, Susan Gerber,
and Jayne Boyd-Zaharias, "Small Classes in the Early Grades, Academic Achievement and Graduating
from High School," Journal of Educational Psychology 97, no. 2 (2005): 214-23; Ellen Frede, "Preschool
Program Quality for Children in Poverty," in Early Care and Education for Children in Poverty: Promises,
Programs, and Long-Term Outcomes, edited by W Steven Barnett and Sarane Boocock (SUNY Press,
1998), pp. 77-98.
53. See Stacey Dale and Alan Krueger, "Estimating the Payoff to Attending a More Selective College: An Ap?
plication of Selection on Observables and Unobservables," Quarterly Journal of Economics 98 (2002):
1491-527. Magnuson and Waldfogel simulate the effects on achievement from expanding early childhood
education programs to clarify how wider access or upgraded preschooling can redress inequities. Expand?
ing enrollment of black and Hispanic children to 80 percent (that is, one-third higher than the rate for
white children) would close the initial gap by 4-20 percent (12-52 percent) for black (Hispanic) children.
Expanding enrollments to cover all children below the poverty line would reduce the black-white
(Hispanic-white) gap by at most 12 percent (16 percent). In additional simulations, upgrading all types of
96 THE FUTURE OF CHILDREN

Early Childhood Development and Social Mobility
preschooling has no effect on gaps across ethnic groups; improving the quality of Head Start reduces racial
gaps by at most 10 percent (8 percent) for black (Hispanic) children. Katherine Magnuson and Jane Wald-
fogel, "Early Childhood Education: Effects on Ethnic and Racial Gaps in School Readiness," Future of
Children 15, no. 1 (2005): 169-96.
54. Long-term health effects may be significant. Single motherhood is associated with higher rates of physical
abuse and child neglect, and parental income has a strong impact on child health. See Christina Paxson and
Jane Waldfogel, "Parental Resources and Child Abuse and Neglect," American Economic Review 89
(1999): 239-44; Anne Case, Darren Lubotsky, and Christina Paxson, "Economic Status and Health in
Childhood: The Origins of the Gradient," American Economic Review 92 (2003): 1308-34.
55. Peter Gottschalk, "Is the Correlation in Welfare Participation across Generations Spurious?" Journal of
Public Economics 63 (2003): 1-25; David Green and William Warburton, "Tightening a Welfare System:
The Effects of Benefit Denial on Future Welfare Receipt," Journal of Public Economics 88 (2004):
1471-93. But the correlation may be muted, as the extent of welfare receipt scarring is debatable: scarring
effects appear weak, and welfare payments are increasingly becoming time limited (most new EITC
claimant families lose eligibility within two years).
56. Chris Lackey, "Violent Family Heritage, the Transition to Adulthood, and Later Partner Violence," Journal
of Family Issues 24 (2003): 74-98.
57. Sandra Stith and others, "The Intergenerational Transmission of Spouse Abuse: A Meta-Analysis," Journal
of Marriage and the Family 62 (2000): 640-54; Jenny Williams and Robin Sickles, "An Analysis of the
Crime as Work Model: Evidence from the 1958 Philadelphia Birth Cohort Study," Journal of Human Re?
sources 37 (2002): 479-509.
58. Rebecca Maynard, "The Costs of Adolescent Childbearing," in Kids Having Kids: Economic Costs and So?
cial Consequences of Teen Pregnancy, edited by Rebecca Maynard (Washington: Urban Institute Press,
1996).
59. Preschooling does not have a strong effect on "steady-state" family size. It delays or reduces childbearing,
by raising the opportunity cost of time spent on child care and lowering the probability of unplanned par?
enthood; but it raises childbearing, because of its association with higher incomes. Typically, the opportu?
nity cost and planning effects are slightly greater than the income effect.
60. Terrie Moffitt, "Adolescence-Limited and Life-Course-Persistent Antisocial Behavior?A Developmental
Taxonomy," Psychological Review 100 (1993): 674-701; Casey Mulligan, "Galton versus the Human Capi?
tal Approach to Inheritance," Journal of Political Economy 107 (1999): S184-224.
61. They also adversely influence children's test scores: being born to a teen mother reduces children's test
scores at age six by 0.07 effect sizes; independently, a two-parent family is associated with test scores that
are 0.1 effect size higher. See Roland Fryer and Steven Levitt, "Understanding the Black-White Test Score
Gap in the First Two Years of School," Review of Economics and Statistics 86 (2004): 447-64.
62. Robert Haveman, Barbara Wolfe, and Elaine Peterson, "Children of Early Childbearers as Young Adults,"
in Kids Having Kids: Economic Costs and Social Consequences of Teen Pregnancy, edited by Rebecca May?
nard (Washington: Urban Institute Press, 1996); Christian Belzil and Jorgen Hansen, "Structural Estimates
of the Intergenerational Education Correlation," Journal of Applied Econometrics 18 (2003): 679-96.
VOL. 16 / NO. 2 / FALL 2006 97

W. Steven Barnett and Clive R. Belfield
63. Heather Antecol and Kelly Bedard, "Does Single Parenthood Increase the Probability of Teenage Promis?
cuity, Substance Abuse, and Crime?" Working Paper, University of California, http://econ.ucsb.edu/~kelly/
youth.pdf; Cesar Rebellon, "Reconsidering the Broken Homes/Delinquency Relationship and Exploring
Its Mediating Mechanism(s)," Criminology 40 (2002): 103-35; Jennifer Hunt, "Teen Births Keep American
Crime High," Working Paper 9632 (Cambridge, Mass.: National Bureau of Economic Research, 2003).
64. Based on sixteen studies of intergenerational earnings correlations, an increase in parental income of
$1,000 raises offspring income by approximately $340. See Mulligan, "Galton versus the Human Capital
Approach" (see note 60). The intergenerational earnings elasticity between fathers and sons is 0.4; that is,
if the father s earnings are 10 percent above the average for his generation, his son's earnings will be 4 per?
cent higher than the average for his own generation. See Gary Solon, "Intergenerational Income Mobility
in the United States," in Handbook of Labor Economics, vol. 3A, edited by Orley Ashenfelter and David
Card (Amsterdam: North-Holland, 1999).
65. Robert Haveman and Barbara Wolfe, "The Determinants of Children's Attainment: A Review of Methods
and Findings," Journal of Economic Literature 33 (1995): 1829-77.
66. See Mulligan, "Galton versus the Human Capital Approach" (see note 60). Income effects on offspring at?
tainment are similarly strong; and a two-parent family is associated with higher attainment by 0.43 years;
Belzil and Hansen, "Structural Estimates" (see note 62).
67. Perry Preschool program participants had an average of 2.4 children by age forty-two. There are data on
whether the first or second child has ever been arrested, repeated a grade, or been on welfare, and if the
child is currently employed. There is no clear evidence of offspring advantages across these dimensions.
However, sample sizes are very small.
68. Diego Restuccia and Carlos Urrutia, "Intergenerational Persistence of Earnings: The Role of Early and
College Education," American Economic Review 94 (2004): 1354-78.
69. This result accords with other research that finds investments in youth to be less efficient than investments
in young children. See Steve Cameron and James Heckman, "The Dynamics of Educational Attainment for
Black, Hispanic, and White Males," Journal of Political Economy 109 (2001): 455-99. Our primary focus is
on the efficacy of early investments per se, rather than on their efficacy relative to other interventions.
70. Families will have to invest resources, even if the program is publicly provided. Even with zero fees, some
families do not enroll, so presumably the costs and inconvenience of enrollment must outweigh the bene?
fits. It is possible that there is an informational problem: families do not appreciate the benefits of pre?
schooling. However, the more likely explanation is that pre-K is not convenient for many families or that
even relatively small direct expenses (such as transportation) are too much.
71. Janet Currie and Matthew Neidell, "Getting Inside the 'Black Box' of Head Start Quality: What Matters
and What Doesn't," Economics of Education Review (2006, forthcoming).
72. W Steven Barnett, "Maximizing Returns from Prekindergarten Education," Education and Economic De?
velopment, Federal Reserve Bank of Cleveland (November 2004), www.clevelandfed.org/Research/Ed-
Conf2004/Nov/PapersPresntns .cfm.
98 THE FUTURE OF CHILDREN

Plain text is unavailable for this page.