reading and answer 3 questions
Please define school readiness?
Why do scholars believe school readiness is important?
Please list and explain two factors that influence school readiness?
a short paragraph for each question.
BROOKINGS| December 2011
`
Starting School at a
Disadvantage:
The School Readiness of Poor Children
Julia B. Isaacs, Brookings Institution
THE SOCIAL GENOME PROJECT
The author expresses appreciation to Jeffrey Diebold for invaluable assistance with
data analysis and to Brian Levy, Katherine Magnuson, Emily Monea, Stephanie Owen,
Isabel Sawhill for their helpful comments.
MARCH 2012
Executive Summary
Poor children in the United States start school at a disadvantage in
terms of their early skills, behaviors, and health. Fewer than half (48
percent) of poor children are ready for school at age five, compared to 75
percent of children from families with moderate and high income, a 27
percentage point gap. This paper examines the reasons why poor
children are less ready for school and evaluates three interventions for
improving their school readiness.
Poverty is one of several risk factors facing poor children. Mothers
living in poverty are often unmarried and poorly educated, they have
higher rates of depression and poor health than more affluent mothers,
and they demonstrate lower parenting skills in certain dimensions. In fact,
the gap in school readiness shrinks from 27 percentage points to 7
percentage points after adjusting for demographic, health, and behavioral
differences between poor and moderate- and higher-income families.
Even so, poverty remains an important influence on school readiness,
partly through its influence on many of the observed differences between
poor and more affluent families. Higher levels of depression and a more
punitive parenting style, for example, may result from economic stress
and so models controlling for these factors may understate the full
effects of poverty on school readiness.
In addition to poverty, key influences on school readiness include
preschool attendance, parenting behaviors, parents’ education, maternal
depression, prenatal exposure to tobacco, and low birth weight. For
example, the likelihood of being school ready is 9 percentage points
higher for children attending preschool, controlling for other family
characteristics, and is 10 percentage points lower for children whose
mothers smoke during pregnancy and also 10 percentage points lower for
children whose mothers score low in supportiveness during parent-child
interactions. These findings suggest a diverse set of policy interventions
that might improve children’s school readiness, ranging from smoking
cessation programs for pregnant women to parenting programs,
treatments for maternal depression, income support programs and
expansion of preschool programs.
Preschool programs offer the most promise for increasing children’s
school readiness, according to a simple simulation that models the
effects of three different interventions. Expanding preschool programs
for four-year olds has more direct effects on school readiness at age five
than either smoking cessation programs during pregnancy or nurse home
visiting programs to pregnant women and infants, the two other
alternatives considered.
2 BROOKINGS | March 2012
mericans aspire to live in a society where children of humble origins who work hard
can rise to middle-class status. Yet many children who are born into poverty struggle
to make ends meet as adults, failing to earn enough to achieve middle-class status.
Nearly two out of three children born into the bottom fifth of the income distribution
remain in the bottom two-fifths of the income distribution as adults (Isaacs, Sawhill and
Haskins, 2009). Often, lack of economic success can be traced back to failure to complete
college or even high school, which in turn stems from academic and behavioral struggles
during grade school. In fact, there is a large health and skills gap between poor children and
their more affluent peers even before they enter school.
Poor children start school at a disadvantage. Their health, behaviors, and skills make them
less prepared for kindergarten than children growing up under better economic conditions.
Fewer than half (48 percent) of poor children are school ready at age five, under a summary
measure that encompasses early math and reading skills, learning-related and problem
behaviors, and overall physical health. Children born to parents with moderate or higher
incomes are much more likely to enter school ready to learn; three-fourths (75 percent) of
these children are ready for school at age five. In other words, there is a 27 percentage point
gap in school readiness between poor children and those from moderate or higher income
families.
Kindergarten teachers find it easier to teach children if they have pre-academic skills, such
as recognizing letters and numbers, and if they can sit still, follow directions, and pay
attention. Children who are aggressive, have temper tantrums or exhibit other problem
behaviors, as well as children in poor health, pose challenges to kindergarten teachers
struggling to impart basic skills in a classroom setting.
School readiness has effects beyond the first few months of kindergarten; children with
higher levels of school readiness at age five are generally more successful in grade school, are
less likely to drop out of high school, and earn more as adults, even after adjusting for
differences in family background (Duncan et al., 2007, Duncan et al., 2010). Entering school
ready to learn can improve one’s chances of reaching middle class status by age 40 by about
8 percentage points, according to a recent analysis that uses linked data sets to track success
from birth to age 40 (Winship, Sawhill and Gold, 2011).
With growing awareness of the importance of early years, federal and state governments
have expanded their investments in young children. State spending on public pre
–
kindergartens, for example, increased each year from 2000 to 2009. The federal Head Start
program has expanded to serve younger children, Congress enacted the new Maternal, Infant
and Early Childhood Home Visiting Program two years ago, and the most recent “Race to the
Top” competition included some funding for states’ systems of early childhood education.
Even so, early childhood programs receive much less funding than public education.
Moreover, early interventions are at risk for funding cuts, as federal and state budgets are
squeezed by rising spending on health and retirement costs and falling tax revenues. State
funding on public pre-kindergartens actually fell in 2010, the first cut after a decade of
expansion (Barnett et al., 2010). At the federal level, early education programs must compete
with many other programs annually for increasingly scarce federal dollars. In this fiscal
environment, it is important to understand the effects of expanding (or cutting) programs
addressing children’s school readiness.
This paper examines children’s readiness for school at age five, comparing poor children
to children from more affluent families. After an initial section documenting a sizable gap in
school readiness, the next two sections address two important questions. First, why are poor
children less ready to learn than children from more affluent families? Second, does a better
understanding of key explanatory factors suggest targets of opportunity, that is, points of
A
3 BROOKINGS | March 2012
possible intervention to improve the early academic skills and behaviors of low-income
children? Finally, the concluding section describes a simulation that compares the effects and
costs of three different interventions: preschool programs, smoking cessation programs, and
nurse home visiting programs.
Fewer than half (48 percent) of poor children compared to 75 percent of children from
moderate or high-income households are ready for school at age five, resulting in a 27
percentage point gap in school readiness, as shown in Figure 1. This comparison focuses on
the difference between children from households with income below 100 percent of poverty
($18,000 for a family of three or $23,000 for a family of four, in 2011 terms) and children from
households with income above 185 percent of poverty. This latter group spans a broad
spectrum of family income, from incomes just above 185 percent of poverty ($33,000 for a
family of three in 2011) to much higher levels of family income.
Children who are “near poor” (from households with income between 100 and 185 percent
of poverty) also enter kindergarten at a disadvantage, although faring better than poor
children: 59 percent of children with incomes just above the poverty line are ready for school
at age five. School readiness rises to 86 percent for children born into households with
income above $100,000, and falls to 42 percent for children who are persistently poor: not
just at birth, but also at ages two, four and five years (Isaacs and Magnuson, 2011).
Source and Notes: Brookings tabulations of data from the Early Childhood Longitudinal Study – Birth Cohort (ECLS-
B). Near poor is defined as household income between 100 and 185 of poverty percent and moderate or high
income is defined as household income above 185 percent of poverty.
The school readiness patterns analyzed here come from the Early Childhood Longitudinal
Study-Birth Cohort (ECLS-B), which follows a nationally representative sample of children
from birth (in 2001) through entry into kindergarten (in the fall of 2006 or 2007). Nearly one
quarter (23 percent) of children in the sample were born into poor families, another quarter
(25 percent) of children fall into the near-poor group, and the remaining half (52 percent) are
classified as having moderate or higher income.
The ECLS-B has rich data on children’s school readiness, including measures of children’s
early academic skills, socio-emotional behavior, and physical health. Each of these domains is
4
8
5
9
7
5
–
10
20
30
40
50
60
70
80
Poor Near Poor Moderate or High Income
P
er
ce
nt
ag
e
S
ch
oo
l R
ea
dy
4 BROOKINGS | March 2012
important for later success and contributes to the school readiness measure used in this
analysis. Specifically, school readiness is measured by combining direct assessments of early
math skills and early reading skills with overall health status taken from parent surveys and
two behavioral measures drawn from kindergarten teacher reports (learning-related
behaviors, such as paying attention, and externalizing or problem behaviors, such as
disrupting others). Children are rated as “school ready” provided they do not score “very low”
on any of these underlying measures; “very low” is defined here as poor/fair on health, and
more than one standard deviation below average on the academic and behavioral measures.
As shown in Figure 2, poor children are much more likely than other children to score very
low on math and reading skills: three out of ten poor children (30 percent) score very low on
early reading skills, compared to only 7 percent of children from moderate- or high-income
families. Differences are smaller but still substantial on the behavioral and health measures.
More than half (52 percent) of poor children score very low on at least one of the five
measures, and so fail to be school ready, compared to one-quarter of children from moderate-
or high-income families.
Figure 2: Likelihood of Scoring Very Low (Failing to Be School Ready) on Measures of School
Readiness, by Poverty Status
Source and Notes: Brookings tabulations of data from the Early Childhood Longitudinal Study – Birth Cohort (ECLS-
B). Very low is defined as more than one standard deviation below average on the academic and behavioral
measures and in poor/fair health on the physical health measure.
Poverty affects school-readiness across a wide range of populations. For example, poor
whites are less school ready than moderate/higher income whites, and the same is true of
blacks, Hispanics, children of married parents, children of unmarried parents, and children
whose mothers have a high school degree or less (see Figure 3). In addition, poor children who
attend preschool programs are less likely to be school ready than preschool attendees from
more economically advantaged backgrounds. Children with college-educated mothers provide
an exception to the general pattern: poor children whose mothers have a college degree or
higher are as well-prepared for school as other children of college-educated mothers (but this
small group represents only 2 percent of all poor children).
2
6
30
23
20
4
5
2
19
2
1
20
15
2
41
7
7
13 13
1
25
–
10
20
30
40
50
60
Math Skills Reading Skills Learning-related
Behaviors
Externalizing
Behaviors
Physical health Any of These
Measures
P
er
ce
nt
ag
e
S
co
ri
ng
V
er
y
Lo
w
(F
ai
lin
g
to
b
e
S
ch
oo
l R
ea
dy
)
Poor Near-Poor Moderate or High Income
5 BROOKINGS | March 2012
46 45
51
43
51
91
55
44
51
79
66
63
53
70
84
78
59
77
–
10
20
30
40
50
60
70
80
90
100
White Black Hispanic < HS (Mother)
HS (Mother) BA+
(Mother)
Married
Mother
Not Married Attend
Preschl
P
er
ce
nt
ag
e
S
ch
oo
l R
ea
dy
Poor Moderate or High Income
46 45
51
43
51
91
55
44
51
79
66
63
53
70
84
78
59
77
–
10
20
30
40
50
60
70
80
90
100
White Black Hispanic < HS
(Mother)
HS (Mother) BA+
(Mother)
Married
Mother
Not Married Attend
Preschl
P
er
ce
nt
ag
e
S
ch
oo
l R
ea
dy
Poor Moderate or High Income
Figure 3: Likelihood of Being Ready for School at Age Five, by Poverty Status at Birth and
Selected Child and Family Characteristics
Source and Notes: Brookings tabulations of data from the Early Childhood Longitudinal Study-Birth Cohort (ECLS-
B). Poor at birth is defined as household income less than 100 percent of poverty and moderate or high income is
defined as household income at or above 185 percent of poverty. School readiness of near-poor children (incomes
100-185 percent) is not shown but generally lies between the two other groups.
At the simplest level, there are two basic theories as to why poor children have worse
outcomes than other children. Proponents of one view focus on the economic differences
between poor families and other families, and argue that many of the negative outcomes
observed in poor children and their families are a by-product of lack of financial resources.
Another explanation is that it is not money itself, but the many parental characteristics that
are associated with poverty that are harmful to children (Mayer, 1997). Analysis of these and
other data suggests that both explanations play a role: poor children do worse in school partly
because their families have fewer financial resources but also because their parents tend to
have less education, higher rates of single and teen parenthood, poorer health, and other
characteristics that place their children at risk for less successful outcomes.
There are several ways that family income can directly influence child development. From
an economic perspective, families with lower incomes have less access to the resources
needed for healthy development, such as nutritious meals, enriched home environments, high-
quality child care settings, and first-class health care resources (Becker, 1981). Poor children
also may suffer from the negative effects of living in neighborhoods with more crime and air
and noise pollution (Evans, 2004). From a psychological perspective, the stress of living in
poverty has a profound effect on parents, contributing to depression, anxiety, and other
forms of psychological stress that can negatively impact their interactions with children. Even
when parental stress does not manifest itself in observed changes in mental health, it can
contribute to a harsh and less supportive parenting style, according to a body of research
dating back to the Great Depression (Mcloyd, 1990; Chase-Lansdale and Pittman, 2002).
While poverty may have myriad influences on family life, it also is true that poor families
differ from other families across a range of characteristics, some of which may be
independent of, or forerunners to, poverty status. As shown in Figure 4, many poor children
live with unmarried parents who have not graduated from college and may not have
completed high school. More than half of their mothers show moderate or severe signs of
6 BROOKINGS | March 2012
12
8
28
4
7
6
5
17
4
30
27
55
14
19
10
23
64
42
0 10 20 30 40 50 60 70
Low on Cognitive Stimulation
Low on Maternal Supportiveness
Mother Depressed
Mother in Poor/Fair Health
Smoking During Pregnancy
Low Birth Weight
Teen Mother
Mother Not Married
Mother has < HS degree
Percentage With Characteristic
Poor
Moderate or High
Income
depression, a depression rate nearly twice that of more affluent mothers. Poor children also
are at higher risk of prenatal tobacco exposure and low birth weight, although these are
relatively infrequent events (see Figure 4). In addition, in comparison to their wealthier
counterparts, more poor children live with mothers who score low on providing cognitive
stimulation (e.g., infrequently reading books, telling stories, singing songs) or score low on
sensitivity, warmth, and general supportiveness during parent-child interactions (based on
videotaped observations of the parent and child at age two).
Figure 4: Poor Families Differ from Moderate/High Income Families on Many Characteristics
that May Affect School Readiness
Source and Notes: Brookings tabulations of data from the Early Childhood Longitudinal Study-Birth Cohort (ECLS-
B). Poor at birth is defined as household income less than 100 percent of poverty and moderate or high income is
defined as household income at or above 185 percent of poverty. Prevalence of characteristics among near-poor
children (incomes 100-185 percent) is not shown but always lies between the two other groups for all
characteristics shown in the figure.
With the range of differences shown in Figure 4, one might well expect that much of the
“poverty gap” in school readiness is influenced by family characteristics rather than family
income. In fact, the poverty-related gap in school readiness drops considerably, from 27
percentage points to 10 percentage points, if we add statistical controls for family
demographics. It shrinks further, to 7 percentage points, after adjusting for other factors that
are more prevalent in poor than in moderate and higher-income families, as shown in Figure 5.
The demographic controls in this analysis include the parents’ level of education, marital
status, and mother’s age at birth, as well race/ethnicity, immigrant status, gender, and age in
months. Parents’ education is a large factor explaining why children from moderate and high
income families enter school with higher reading and math skills—their parents are better
educated. Children’s early academic skills are higher, on average, when parents have more
years of schooling, and this association persists even after controlling for parents’ inherent
abilities, according to evidence from welfare reform evaluations and sophisticated statistical
analyses (Gennetian, Magnuson & Morris, 2008; Carneiro et al., 2007). In addition, the
“education” effect also reflects underlying differences in parents’ skills and preferences,
which are often passed on to their children, by both inherited traits and upbringing.
The large number of poor children living with an unmarried mother also contributes to the
poverty gap in school readiness. Children living with single and even cohabiting parents tend
to have worse outcomes, particularly behavioral outcomes, than similar children living with
7 BROOKINGS | March 2012
married parents, according to several studies (Waldfogel, Craigie and Brooks-Gunn, 2010).
Living with teen parents also may put children at additional risk, although researchers find
little evidence regarding effects on early academic skills and mixed evidence regarding effects
on behavioral outcomes (Levine et al., 2007).
The poverty gap is much smaller after controlling for demographic factors, but still
substantial, in both statistical significance and policy relevance (10 percentage points). The
gap shrinks further after adding controls for low birth weight and preschool attendance, two
potential pathways for how poverty may affect children’s development. Families with fewer
economic resources may have less access to good health care or nutrition (which may show
up in lower birth weight babies) and also may have decreased ability to send their children to
preschool programs. In fact, although low birth weight and preschool participation exert
independent effects on school readiness, they do not explain much of the remaining poverty-
related gap in school readiness, and the gap remains at 9 percentage points, even after
adding controls for low birth weight and preschool attendance.
Source and Notes: Brookings analyses of data from the Early Childhood Longitudinal Study-Birth Cohort (ECLS-B).
Poor is defined as household income less than 100 percent of poverty and moderate or high income is defined as
household income at or above 185 percent of poverty. Demographics include parents’ level of education, marital
status, mother’s age at birth and immigrant status, and child’s gender, age in months and race/ethnicity. Parental
health and behaviors include physical health, smoking during pregnancy, breastfeeding, depression, parenting
behaviors (parental supportiveness and provision of cognitive stimulation) and maternal employment; the analysis
also controls for use of non-parental care, and number of children and adults in the household.
The poverty gap reduces somewhat further, to 7 percentage points, in the full model,
which adds additional controls for maternal health and behavior, specifically, measures of
physical health, smoking during pregnancy, breastfeeding, depression, and parenting
behaviors (parental supportiveness and provision of cognitive stimulation).
Parental health and behaviors play a complex role in the analysis. These may be caused, at
least partially, by the parents’ lack of financial resources: impoverished circumstances,
overcrowded housing, and lack of access to good health care can result in health problems,
higher levels of depression, and a harsher parenting style. On the other hand, a chronic
physical or mental health condition can exist independent of poverty status, and may even be
a cause, rather than a result of poverty (with poor health dragging down employment and
earnings, and thus increasing chances of entering poverty).
-7
-9
-10
-27
-30 -25 -20 -15 -10 -5 0
Above & Parental Health
& Behaviors
Demographic & Low Birth
Weight & Preschool
Demographics Alone
Raw Gap
Difference in Likelihood of Being School Ready (Poor vs. Moderate/High Income)
8 BROOKINGS | March 2012
If one focuses on the ways that economic stress affects parents’ health and their
interactions with their children, one is likely to view these parental health and behavior
measures as pathways or mediators through which the true underlying cause—poverty—
operates. Under this view, the 7 percentage-point estimate from full model, which includes
controls for these factors, understates the extent to which poverty is a root cause of poor
children’s diminished school readiness. Alternatively, some might argue that parental
education, marital status, health, and behaviors—rather than the amount of money families
have to spend on their children—are the real causes of the diminished school readiness of poor
children. Statistical analyses such as these cannot tell us which interpretation carries more
weight.1
The best evidence as to whether income has a causal effect on school readiness comes
not from regression analyses, but from experimental programs, such as welfare-to-work
experiments, where two otherwise identical groups of families with children receive different
levels of income as a result of a policy intervention. An analysis of several such random-
assignment experiments has found that children’s math and reading skills were indeed
improved by programs that increased parental income and employment, but not by programs
that only increased employment (Duncan et al., 2011). In addition, a quasi-experimental study
of varying child benefits in Canada found higher levels of child achievement (e.g., higher
vocabulary scores) in children whose families received higher income supplements (Milligan
and Stabile, 2008). These quasi-experimental studies provide convincing evidence that money
matters, although it is not the only influence on young children’s developmental outcomes.
Poverty is one of multiple, inter-related influences on children’s school readiness. This
section considers influences other than money by first looking at simple differences across
groups of children (see Figure 6), and then by examining which differences persist when
adding statistical controls (see Figure 7). While the full model with statistical controls provides
some guidance as to which factors contribute to children’s school readiness, the model may
overstate or understate the causal impacts of specific variables, and so it is important to
compare estimates from the statistical model to the broader social science literature.
Girls are markedly more school ready than boys; the average 5-year old girl is 16
percentage points more likely to be school ready than the average boy (see Figure 6). The
gender gap is largely independent of family background and remains at 14 percentage points
even with statistical controls (see Figure 7). The gender difference in school readiness is
driven by behavioral differences: girls score a half-standard deviation higher than boys on
behavioral measures, on average. (See the appendix, Table A-1, for regression estimates for
math, reading, learning-related behaviors, externalizing behaviors, and health, as well as
overall school readiness.)
1 As a final note to Figure 5, the “unexplained portion of the poverty gap” could have been dropped even further,
approaching zero, if the analysis had included an even richer set of mediators for the effects of poverty, such as
the family’s ownership of assets (e.g., home, savings account), material possessions (e.g., computer, car), food
insecurity, housing conditions, and neighborhood crime and safety. When economists Waldfogel and Washbrook
included such controls in their careful analysis of income-related gaps in children’s school readiness, they found no
significant residual gap in academic skills and behavioral outcomes between children from the bottom and the
middle fifth of the income (Waldfogel and Washbrook, 2011).
9 BROOKINGS | March 2012
-16
-22
-18
-13
-12
15
-9
21
-17
-20
21
-7
-15
-17
16
-17
-27
-30 -20 -10 0 10 20 30
Low on Cognitive Stimulation
Low on Maternal Supportiveness
Smoking During Pregnancy
Mother Depressed
Mother in Poor/Fair Health
Attended Preschool
Low Birth Weight
Married Mother
Teen Mother
Mother without HS degree (vs HS)
Mother with BA (vs. HS)
Immigrant Mother
Hispanic (vs. White)
Black (vs. White)
Girls (vs.Boys)
Near-Poor (vs > 185% of poverty)
Poor (vs. > 185% of poverty)
Difference in Likelihood of Being School Ready (in Percentage Points)
Figure 6: Various School Readiness Gaps (Before Controlling for Confounding Factors)
Children of college-educated mothers and children attending preschool enter school with
more skills than their counterparts without these advantages. The positive effects of these
factors persist after adding controls, though the magnitude is smaller. The impact of
attending preschool, for example, drops from 15 percentage points to 9 percentage points
after accounting for family characteristics associated with selection into preschool. Random-
assignment and regression-discontinuity studies have also found positive effects of preschool
programs on various dimensions of school readiness, with the most extensive evidence and
largest effects regarding pre-academic skills (Camilli et al., 2010; Isaacs, 2008).
Differences by race/ethnicity, immigrant status, family structure, maternal age at birth,
and maternal physical health are initially large. For example, black and Hispanic children are
less likely to be school ready than white children (by 15 to 17 percentage points) and children
of married mothers are more likely to be school ready than children of unmarried mothers (by
22 percentage points). However, these differences reduce to insignificant levels after
controlling for income and other confounding factors. In contrast, the effects of poverty
remain significant in the full model, with poor children 7 percentage points less likely to be
school ready, as already discussed.
10 BROOKINGS | March 2012
Source and Notes: Brookings analyses of data from the Early Childhood Longitudinal Study-Birth Cohort (ECLS-B).
Bars with numerical values represent characteristics that are statistically significant (at the 95 percent level,
except for low birth weight and cognitive stimulation (90 percent level) and gender and preschool (99 percent
level). Additional controls not shown in figure include paternal education, race (other), maternal employment, use
of non-parental care, breastfeeding, child’s age in months, number of children and adults in the household, and
dummies for missing values on selected variables. Paternal education, children’s age in months and use of
preschool or center-based care before age 4 have statistically significant effects on school readiness (see appendix
for more details).
In addition to poverty, a number of other risk factors have negative effects that persist
after adding controls for other family characteristics. Prenatal exposure to tobacco and low
birth weight have negative effects on school readiness, consistent with other literature
(Wakschlag et al., 2002; Johnson and Schoeni, 2007). In addition, children are less likely to be
school ready if their mothers showed signs of depression during early childhood, their
mothers showed little supportiveness during video-taped observations of parent-child
interactions, or their mothers reported that they read, sang, and told stories to their children
infrequently. Again, these results are consistent with other studies, with parenting behaviors
likely influenced by maternal depression, but also each variable—maternal depression and
parenting behaviors—showing some degree of independent effects on child development
(Kiernan and Huerta, 2008).
The dark bars in Figure 7 suggest multiple targets of opportunity for improving the school
readiness of poor children. In addition to traditional policies aimed at expanding preschool
participation or increasing family income, policy makers interested in improving children’s
school readiness might consider a more diverse set of interventions, such as smoking
cessation programs targeted at pregnant women, health and nutrition policies designed to
reduce low birth weight and other adverse pregnancy outcomes, parenting programs for low-
income parents of young children, mental health treatment options for mothers of young
children, or policies to improve the educational attainment of low-income mothers.
The concluding section of this paper discusses three different possible interventions to
improve school readiness: 1) expanding voluntary preschool programs to all poor four-year
9
-7
14
6
-5
-5
-10
-5
-10
-30 -20 -10 0 10 20 30
Low on Cognitive Stimulation
Low on Maternal Supportiveness
Smoking During Pregnancy
Mother Depressed
Mother in Poor/Fair Health
Attended Preschool
Low Birth Weight
Married Mother
Teen Mother
Mother without HS degree (vs HS)
Mother with BA (vs. HS)
Immigrant Mother
Hispanic (vs. White)
Black (vs. White)
Girls (vs.Boys)
Near-Poor (vs > 185% of poverty)
Poor (vs. > 185% of poverty)
Difference in Likelihood of Being School Ready (in Percentage Points)
11 BROOKINGS | March 2012
olds not currently served, 2) providing smoking-cessation programs to all poor pregnant
women who smoke, and 3) offering nurse home visiting programs to all first-time mothers
with income below poverty. Preschool programs were selected because of the strength of the
evidence demonstrating that such programs improve early math and reading skills. Smoking
cessation programs targeted at pregnant women were selected because of the magnitude of
the negative effects of prenatal exposure to tobacco on school readiness in this analysis
(controlling for low birth weight), as well as additional effects through low birth weight. Nurse
home visiting programs were selected because they address parenting behaviors, as well as
maternal health during pregnancy. Each of these is discussed below, and then their effects on
school readiness are compared through a simulation exercise. Finally, there is a brief
discussion of policies to raise family incomes.
Children who attend some form of preschool program at age four are 9 percentage points
more likely to be school ready than other children, according to the regression model
summarized in Figure 6. This overall impact is driven by positive effects on early math and
reading skills, and smaller, insignificant effects on behaviors (positive for learning-related
behaviors and negative for problem behaviors), and no effect on health, according to the
detailed analyses shown in the appendix (Table A-1).
Preschool
Programs
These findings are largely consistent with the extensive literature on early childhood
education, which finds strong positive impacts on early academic skills, and smaller,
sometimes mixed, findings on behavioral measures. For example, a recent meta-analysis of
studies of a diverse set of early childhood, center-based programs found an average effect
size of 0.23 standard deviations on cognitive skills, and 0.15 standard deviations for socio-
emotional skills (Camilli et al., 2010). These effects are larger than those shown in Table A-1,
suggesting that preschool programs may, in fact, have larger effects than the 9 percentage
point difference used in the simulation exercise.
The effects of preschool program attendance vary, of course, depending on the particular
program attended. Children “attending preschool” in the United States attend a diverse mix of
programs, including public pre-kindergarten programs, Head Start programs, private nursery
schools, and center-based child care settings. Quality varies considerably, both across and
within these various types of settings. Substantial impacts have been found for certain model
programs (e.g., Perry Preschool, Abecedarian, Chicago Public Schools), as well as some public
pre-kindergarten programs. Impacts of Head Start on school readiness have been smaller, and
prone to fade-out. Little is known about the effects of private nursery schools and center-
based child care settings, and the quality of these settings is quite uneven. The intervention
simulated below will be an expansion of the existing mix of preschool programs, varying in
both quality and setting.
Smoking cessation programs appear to be a promising intervention, given the strong
negative effect of tobacco use during pregnancy seen in Figure 7 (a 10 percentage point
reduction in school readiness of children of smokers, in addition to whatever effects may
occur through the effect of smoking on low birth weight). As shown in the appendix, maternal
smoking during pregnancy is strongly associated with only one domain, behavior problems,
with no effects on math, reading, learning-related behaviors or physical health (after
controlling for low birth weight). This finding is consistent with other research that finds
children of smokers at risk for severe antisocial behavior (Wakcshlag et al., 2002). As a result,
the positive effects of smoking cessation programs might not show up in school readiness
measures that put more emphasis on academic skills. In addition, the causal connection
between smoking and children’s behaviors has not been conclusively established; smoking
may simply signal the presence of another risk factor that might be associated with smoking
during pregnancy and worse child outcomes at age five. However, because negative results of
Smoking Cessation Programs
12 BROOKINGS | March 2012
smoking are found consistently across diverse settings and because biological evidence
suggests that nicotine crosses the placental barrier and is associated with fetal neurotoxicity,
reducing tobacco use during pregnancy remains an important target of intervention.
Interventions with pregnant smokers generally involve one-on-one counseling of pregnant
women, sometimes supplemented with incentives, feedback about fetal health status, and
medications or therapies to reduce nicotine cravings. Such programs have only limited
success: reducing the number of women who continue smoking throughout their pregnancy
by 6 out of 100 women, according to a Cochrane Collaboration review of 72 controlled trials
(Lumley et al., 2010). These percentages are in comparison to women in the control group,
who are generally exposed to advice to stop smoking and often quit on their own. A similar
modest rate of success was reported by the U.S. Public Health Service, which reported that
the abstention rate for pregnant smokers increased from 7.6 percent under “usual care” to
13.3 percent after psycho-social interventions (US HHS, 2008). Lumley and colleagues
reported a roughly 15 percent reduction in low birth weight and pre-term births for infants
born to treated mothers, leading the authors to recommend that smoking cessation programs
be adopted in all maternity care settings.
Nurse home visiting programs address a number of different risk factors identified in
Figure 7 through one intervention. Under what is called the Nurse Family Partnership
program, nurses visit low-income, first-time mothers in their homes during pregnancy and
infancy, following a curriculum that attempts to reduce smoking and other unhealthy
behaviors during pregnancy; improve birth outcomes (with specific attention to reducing pre-
term and low-birth weight births); provide coaching and teaching of parenting skills; and
encourage mothers to focus on their own future, including setting education and career goals,
planning any subsequent births, and building social supports. Nurse Family Partnership has
had more documented success than other home visiting programs; some its success may be
due to its attention to the prenatal period (clients are enrolled during the second trimester)
and its focus on a particularly disadvantaged group (low-income, first-time mothers, many of
whom are unmarried and/or teen parents). It is one of the program models eligible for funding
under the new Maternal, Infant and Early Childhood Home Visiting Program.
Nurse Home Visiting
While it does not have much effect on maternal education or depression, nurse home
visiting has some success in reducing smoking during pregnancy, reducing pre-term and low
birth weight births, and improving mother’s observed responsiveness and sensitivity toward
children in parent-child interactions.2 As a result of these and other changes, children of
home-visited mothers show some signs of improved school readiness in the two most recent
trials, including gains in early reading and math skills, as well as increases in executive
functioning, a measure related to learning-related behaviors. However, these gains were only
significant for about half the children in the programs, those children with “low-resource”
mothers, defined as mothers scoring low in intelligence, mental health, or self-confidence. 3
2 The smoking effects were found in two of three random-control trials (and there were very few
smokers in the third trial); the reduction in low birth was found in one trial, among younger mothers
and smoking mothers; and the improvements in maternal sensitivity were found in all three samples (in
full sample for two trials and in a subgroup of low resource mothers in the third trial). U.S. DHHS, 2011;
Olds et al., 1986.
3 The positive effects were found in Denver (observations at age four) and Memphis (observations at
age six). The original study in Elmira, New York did not find many differences during early childhood,
but it relied on IQ tests, rather than the newer measures of school readiness. It did find lower rates of
juvenile delinquency at age fifteen, as well as lower rates of child abuse and neglect throughout
childhood. U.S. DHHS, 2011; Olds et al., 1998.
13 BROOKINGS | March 2012
For each of these three interventions, I simulated possible effects on school readiness,
using the estimates shown in Figure 7, combined with a number of different assumptions, as
described below and summarized in Table 1. In brief, preschool programs offer the most
promise for increasing children’s school readiness, with more direct effects on school
readiness at age five than nurse home visiting or smoking cessation programs.
Simulated Effects of Interventions
Preschool
Programs
Smoking
Cessation
Programs
Nurse Home
Visiting
Programs
1. Poor Population (single-year
cohort) 23% of 4 million=920,000
2. Treated Group 221,000a 184,00b 128,000c
3. Affected Group
All newly
enrolled
(100% of treated)
Smokers who
quit due to
intervention
(6% of treated)
Mothers with low
psychological
resources
(50% of treated)
4. Increased Chances of Being
School Ready (for affected group) 8.9% d 4.9-9.7% e 6.7% f
5. Increased Chances of Being
School Ready (for treated group)
[row 4 x percentages in row 3]
8.9% 0.3%-0.6% 3.4%
Additional information:
Average Cost Per Treatment $ 7,400 $ 350 $7,200
Total Cost (in billions)
[row 2 x Average Costs ] 1.6 0.1 0.9
Increased Chances of Being School
Ready (across all poor children)
[row 5*(row 2/row 1)] 2.1% 0.1% 0.5%
Other Effects (in addition to effects
on school readiness)
Education
Earnings
Maternal
Health
Neonatal Costs
Child Abuse
Neonatal Costs
Juvenile Arrests
Note: a. The treated preschool population is 24% of all poor year-olds, assuming 66% already enrolled and 10% fail
to enroll in a voluntary program. b. The treated population for smoking cessation programs is 20% of poor
pregnant women initially smoking (building off the observed 18.5 percent still smoking in the third trimester and
assuming a 7.6 percent rate of quitting under usual care). c. The treated population for nurse home visiting is 14%
of poor mothers, assuming 30% of poor births are to first-time mothers, 50% of that 30% voluntarily enroll and
1% are already served. d. Estimate from Figure 7 (Table A-1). The 90% confidence interval around that estimate is
4.8 to 13.0%. e. The high estimate from Figure 7 (Table A-1); the low estimate assumes that quitting smoking before
third trimester has only half the effect of not smoking at all during pregnancy. The 90% confidence interval around
the high estimate is 4.4 to 15.0%. f. Simulated effect assuming math skills of children of low-resource mothers
increase by 0.25 standard deviations, reading skills by 0.2 and learning-related behaviors by 0.25, as explained in
the text and footnote 7.
Expanding preschool programs to all four-year old children who choose to participate is
estimated to affect 221,000 children annually, or 24 percent of poor four-year olds. Other
poor preschoolers already participate in preschool programs (66 percent) or are assumed not
to participate under a voluntary program (10 percent). As already discussed, preschool
attendance is estimated to increase the likelihood of school readiness by about 9 percentage
14 BROOKINGS | March 2012
points for the affected group.4
Providing smoking-cessation programs to all poor pregnant women could potentially
affect as many as 184,000 pregnant smoking women and their infants per year. The results
summarized in Figure 7 suggest large potential impacts; maternal smoking during pregnancy
is associated with a 10 percent change in school readiness. However, mothers who quit
smoking at some point during the pregnancy may not reap the full benefit of not smoking, and
so the simulation includes an alternate low estimate of about 5 percent, assuming only half
the full impact for mothers who quit during the pregnancy. Whether the impact is 5 or 10
points for those mothers who quit, the overall impact for all treated smokers is quite small,
because children are not affected unless their mothers quit and, as noted above, only 6 out of
100 women enrolled in such programs quit as a result of the intervention. The total impact for
children of smoking mothers thus falls to an 0.3 to 0.6 percentage point in school readiness (6
percent x 5-10 percent). While modest, these school readiness gains are in addition to the
more obvious health gains for both mothers and infants.
Consistent with other research, this analysis suggests that
preschool programs would have a substantial impact on the affected children, those newly
enrolled in preschool. Note that this simple measure of school readiness probably understates
the full benefits of preschool participation, because it ignores benefits for poor children who
are just above the cut-off for school readiness, but still could benefit from seeing their math
and reading skills increase with a year of preschool participation.
5
Making nurse home visiting programs available to all first-time poor mothers would
expand services to roughly 128,000 pregnant women, or 14 percent of all poor mothers. This
estimate of the treated population assumes that first-time mothers represent 30 percent of
all poor mothers (based on the ECLS-B data), that 50 percent of these mothers would
participate, based on experience with existing programs, and that 1 percent of poor mothers
already receive nurse home visiting services.
Unlike the first two simulations, the simulation of nurse home visiting impacts does not
rely on estimates from Figure 7, because there is no variable for “participation in nurse home
visiting” in the regression analysis. Instead, the simulation draws upon research from studies
conducted in Denver and Memphis. Specifically, nurse home visiting programs are assumed to
increase early reading skills by 0.2 standard deviations, early math skills by 0.25 standard
deviations, and learning-related behaviors by 0.25 standard deviations—but only among
children of low-resource mothers, with no observed effects on other children.6
4 The point estimate is actually 8.7 percentage points, as shown in Table A-1, with a 90 percent
confidence interval ranging from 4.8 to 10.3 percentage points. This simulation uses the estimate for all
children and assumes it applies to poor children. An alternate approach would be to re-estimate the
model for the sample of poor children; the point estimate of preschool attendance was slightly higher
(10.1 vs. 8.9) but the standard errors were also much higher, and so the simulations relied on the
estimates for the full population.
Such increases
would translate into a 6.7 percentage point increase in school readiness according to a
5 This simulation ignores the potential positive effects of reducing cigarette intake, without quitting
completely. It also ignores the indirect effects of smoking on school readiness operating through
effects on low birth weight. The simulation originally included these additional indirect effects, but they
were too small to change the final estimate and so were dropped.
6 The early reading skill effect of 0.2 is based on averaging effect sizes for language development in
Denver (0.31), receptive vocabulary (PPVT) in Memphis (0.21) and reading in Memphis (0.09), math skills
were only measured in Memphis (where they were 0.25 for arithmetic and also 0.25 for composite of
arithmetic and reading, and the 0.25 effect on learning-related behaviors was based on an average
effects on executive functioning in Denver (0.49) and academic engagement in Memphis (0.02). These
effects are for children of low-resource mothers; effects were much smaller for the full population. No
effects were assumed for externalizing behaviors because the effect sizes in both Denver and Memphis
were small, even for the children of low-resource mothers. (.03 and .09, respectively, according to the
child behavior check list on externalizing behaviors). DHHS, 2011.
15 BROOKINGS | March 2012
simulation that provided such increases to the sample of poor first-born children in the ECLS-
B whose mothers were ever depressed (a proxy for mothers with low psychological
resources). The effect would be half as large, only 3.4 percentage points, for the full
treatment group, assuming that low resource mothers represent half of the poor mothers
participating in the program.7 An alternate approach, which attempted to model the indirect
effects of nurse family partnership through its effects on smoking, low birth weight and
parenting behaviors found considerably smaller effects, 0.6 rather than 3.4 percentage
points, but it was not used, because it required a much more extensive set of assumptions.8
To sum up the simulations, preschool attendance is estimated to increase school readiness
by about 9 percentage points for newly enrolled poor children, compared to an increase of
about 3 percentage points in school readiness for children in nurse home visiting programs
and 0.3 to 0.6 percentage points for children whose mothers are provided smoking cessation
services. While these estimates are quite uncertain, they provide a rough sense of the
differential effects of these three interventions.
Finally, the bottom half of Table 1 compares the costs and effects of the programs across
the poverty population. Ball-park costs range from $0.1 billion for smoking-cessation programs
for all poor pregnant women (assuming costs of $350 per woman) to $1.6 billion for expanding
preschool programs to 4-year olds not currently served (assuming costs of $7,200 per child,
or the average of costs for public pre-kindergarten programs and Head Start programs). Costs
for nurse home visiting for all poor, first-time mothers choosing to participate are estimated
as $0.9 billion, assuming average annual costs of $4,500 and average treatment length of 1.6
years.
The effects summarized above were for the treated groups (e.g., poor preschoolers newly
enrolling in preschool programs, poor smoking mothers, etc.), which vary in size across the
interventions. For an apples-to-apples comparison of total effects as compared to total costs,
it is important to also look at effects across the full poverty population. Effects drop in size
when spread across the untreated as well as the treated population, falling to 2.1 percentage
points for preschool programs, 0.1 percentage points for smoking programs (the rounded
estimate for both the low and high estimates), and 0.5 percentage points for nurse home
visiting programs.
It may not be surprising that preschool programs—provided to four year olds and often
with the express purpose of preparing children for school—have the highest effects on school
readiness, whether measured for the treated group or the full poverty population. Smoking-
cessation programs are primarily undertaken to improve maternal and fetal health. This
analysis has highlighted the fact that such programs can have additional effects on school
readiness—particularly by reducing the number of children with severe antisocial programs—
but that is not the main rationale for their adoption. Nurse home visiting programs also have a
strong focus on maternal and child health, including child abuse and neglect. However, they
7 Low-resource mothers represented half of the mothers in Memphis, and 40 percent of mothers in
Denver. Note that the programs served near-poor as well as poor mothers, and so might have served a
slightly less disadvantaged population than the population in these simulations. Olds, Kitzman et al.,
2004; and Olds, Robinson et al., 2004.
8 The alternate approach required making assumptions such as the following: NFP reduces smoking by
6 percent, as in smoking cessation programs, NFP reduces low birth weight by 10 percent, based on the
author’s manipulation of data reported from the trials, and NFP reduces the incidence of low-
supportiveness among parents by 10 percent, a fairly arbitrary assumption. These assumptions,
combined with estimates from Figure 7, resulted in an overall effect of 0.6 percentage points on those
who are home visited, considerably below the 3.4 percent estimate eventually used. The alternate
estimate may be low because it ignores effects on non-smokers who had children of normal birth
weight and have more normal rates of maternal supportiveness. The difference in estimates also
highlights the uncertainty of these simulations, particularly for nurse home visiting programs.
16 BROOKINGS | March 2012
also have been promoted for their ability to improve overall child development, including
school readiness, even though home visits stop on the child’s second birthday. This simulation
suggests that home visiting programs do indeed have some effects on school readiness,
though not as large as effects from center-based preschool programs.
A fourth intervention, not directly modeled here, would be to provide families with
sufficient income to raise them out of poverty. Moving families out of poverty and into
moderate-income status would be fairly costly. For a family of three, there is a gap of more
than $15,000 between being in poverty (household income less than $18,000) and having
income above 185 percent of poverty (greater than $33,000 in household income). However,
if poor families were to experience this substantial increase in income, their children’s
likelihood of being school ready is estimated to increase by 10 percentage points (using the
middle estimate from Figure 5, controlling for demographics, but not for factors that may be
influenced by poverty).
Smaller increases in family income also would improve poor children’s school readiness.
The quasi-experimental evidence discussed earlier suggests that a $1,000 increase in family
income, sustained over two to five years, would result in a 0.06 to 0.07 standard deviation
increase in early academic skills (Duncan et al., 2011; Milligan and Stabile, 2008). In an earlier
analysis of the ECLS-B data used in this paper, Isaacs and Magnuson (2011) estimated that a
$1,000 increase in annual household income during early childhood was associated with about
a 1 percentage point increase in the probability that very low-income child would be school
ready (using the same school readiness measure as in this analysis).9
Such an increase would
cost roughly $5,000 over the first five years of life plus additional costs for administration.
Program costs might be lower or higher, depending on whether the program encouraged or
discouraged higher earnings among parents. Possible interventions to supplement family
income, while at the same time encouraging parents to increase their earnings, include
welfare reform programs that contain include earnings supplements and targeted expansions
of the Earned Income Tax Credit.
This paper examined children’s readiness for school at age five and found a 27 percentage
point gap in school readiness between poor children and moderate/high income children. The
data presented show that it is not poverty alone that places poor children at risk, but also the
fact that their parents have low levels of education, higher rates of smoking, higher rates of
depression, and lower parenting skills than children from moderate- and high-income families.
These findings suggest a diverse set of policy interventions that might improve children’s
school readiness, ranging from smoking cessation programs for pregnant women, parenting
programs, treatments for maternal depression, and policies to improve the educational
attainment of low-income mothers, to income support programs and expansion of preschool
programs. A simple comparison of the simulated effects of three interventions—preschool
programs, smoking cessation programs, and nurse home visiting programs—suggests that the
traditional approach of preschool expansion has the most promise for increasing the school
readiness of poor children.
9 Specifically, Isaacs and Magnuson estimate a 0.7 percentage point increase in the likelihood of school
readiness for children from families whose annual household income during early childhood averaged
$8,100, before the $1,000 increase. Note that family income had considerably smaller effects on
reading and math skills in the Isaacs and Magnuson analysis than suggested by the quasi-experimental
findings, suggesting that this estimate may be quite conservative.
17 BROOKINGS | March 2012
Data for this analysis are taken from the Department of Education’s Early Childhood
Longitudinal Study-Birth Cohort (ECLS-B). Assessments were conducted when the children
were 9 months old, 2 years old, 4 years old, and when they entered kindergarten (either fall
2006 or fall 2007). The analysis file consists of 4,300 children with complete data on the five
school readiness measures and certain key descriptive measures. The analysis was conducted
using weights developed by NCES to correct for attrition and sampling design.
Children were rated school ready as long as they scored no more than one standard
deviation below average on measures of early math skills, early reading skills, learning-related
behaviors and externalizing behaviors, provided they also had a health status of good, very
good or excellent (rather than poor or fair). The five underlying measures were
operationalized as follows:
• The math
• The
assessment is based on an Item Response Theory (IRT) instrument that
includes questions regarding number sense, properties, and operations;
measurement; geometry and spatial sense; data analysis, statistics, and probability;
and patterns, algebra, and functions. The math score was standardized to have a
mean of zero and standard deviation of one across children in the sample.
reading
• The
assessment was derived in a similar manner, based on questions
regarding basic/phonological skills, initial understanding, developing interpretation,
demonstrating a critical stance, and vocabulary.
attention/learning-related behaviors
• The
measure reflects kindergarten teacher
responses to seven questions assessing behaviors such as a child’s ability to
concentrate, work independently, and work until finished, as well as a child’s
eagerness to learn. It was standardized in the same way as the math measure.
externalizing problem behaviors
• The
measure, also based on teacher reports, was
drawn from six questions about whether or not a child acts impulsively, disrupts
others, is overly active, is physically aggressive, annoys other children, and has
temper tantrums. It was reverse coded (and standardized) so that a higher score
indicates a better outcome, consistent with the other measures.
physical health
measure is based on parents’ rating of the child’s overall
health, from a scale of one to five. The health variable was then dichotomized into
either poor/fair health or good/very good/excellent health.
Other key variables are measured as follows:
• Children’s poverty status
•
is measured near birth, based on cash income received by
household members over the past twelve months, as reported in a parent survey
conducted when the infant is about nine months. Children are classified as poor if
income is below 100 percent of poverty, near-poor if income is between 100 and
185 percent of poverty and moderate/high income if income is greater than or
equal to 185 percent of poverty.
Maternal education was collapsed into three variables: Less than high school; High
school graduate; College degree, based on the 9-month parent survey. Father’s
18 BROOKINGS | March 2012
education was measured the same way, with a dummy variable for when the
father’s data was missing.
• Marital Status
•
is based on marital status at birth, and is dichotomized as married
mothers compared to unmarried mothers (single or cohabiting).
Teen Parent
•
was coded as 1 if the mother was less than 20 at the time of birth.
Race/Ethnicity
•
was coded as Non-Hispanic White, Non-Hispanic Black, Hispanic, and
Other (Asian, Mixed or other). Findings for other race are not included in the
summary figures but are shown in the full regressions in Table A-2.
Immigrant Mother
•
was coded 1 if the mother was foreign-born.
Low Birth Weight
•
was coded 1 if the child’s birth weight was less than 5.5 pounds or
2500 grams.
Preschool Attendance
•
was coded 1 if the child attended Head Start, a nursery
school or pre-kindergarten program, or any form of center-based child care at age
four.
Maternal Health
•
was dichotomized as poor/fair vs. good/very good/excellent
health, measured at 9 months.
Moderate or Severe Depression
•
was based on the CES-D depression scale, with
score “1” if scored a 10 or higher at ages 9 months, 4 years or 5 years
Smoking during Pregnancy
•
was coded 1 if mother said yes to a question about
smoking during the third trimester.
Low in Maternal Supportiveness
•
was coded 1 if the mother scored below 3.5 on a 7-
point index measuring parental sensitivity, parental regard, and appropriateness of
parental cognitive stimulation during a videotaped parent-child observation at age
two (the Two Bag test). Fifteen percent of the sample with data on this measure
received this low score.
Low in Cognitive Stimulation
•
was coded 1 if the mother scored 7 or lower in a 12
point scale measuring frequency of reading out loud, telling stories or singing to
children, based on parents’ reports at age two; this represented 18 percent of the
sample.
Breastfeeding
The regression analysis also has variables for mother’s employment, operationalized as
three dummy variables for whether a child’s mother has ever worked (reporting having
worked in the past week at any of the four waves of the survey); whether she has ever worked
full time (reporting having worked at least 30 hours in the past week at any of the waves; and
whether she worked full time when the child was an infant (reporting having worked at least
30 hours in the past week at the nine-month wave). It also controls for non-parental care with
three variables, one for whether the child was ever in non-relative care, one for whether the
child was ever in relative care, and one for whether the child was ever in center-based care
before age four. Since data were collected when children were two and four but not three, the
“ever in center care before age four” variable misses some center-based experiences that
may have occurred between the age two and age four survey waves. Finally, there are also
controls for the number of adults and the number of children in the household as well as the
child’s birth month, from January-December 2001.
was coded 1 if mother breastfed at any time in first 9 months
19 BROOKINGS | March 2012
Results from the regression analyses are shown in Table A-1. The School Readiness and
Health models are linear probability models estimating the change in percentage points that a
child will be school ready (or in good/very good/excellent health. For example, a coefficient of
-0.67 in the School Readiness model means that children in poverty a 6.7 percentage points
less likely to be school ready than children from moderate/high income households, after
controlling for other characteristics. The Math, Reading, Learning-Related Behaviors and
Externalizing Behaviors variable are measured continuously, as Z-scores with a mean of 0 and
a standard deviation of 1. For example, a coefficient of -.286 in the Math model means that
children in poverty score -.286 standard deviations lower than children from moderate/high
income households, all other things held constant.
Table A-1: Coefficients on Regression Analyses
School
Ready
Math Reading
Learning-
Related
Behavior
External-
izing
Behavior
Health
Poor (vs. > 185%) -.067** -.286*** -.276*** -.129* -.061 -.001
(.031) (.049) (.052) (.073) (.071) (.010)
Near-Poor (vs. >185%) -.018 -.129** -.137** -.031 .003 .011
(.030) (.046) (.051) (.062) (.056) (.007)
Mother BA or More (vs HS) .064** .218*** .189*** .137** .129** .005
(.025) (.042) (.047) (.061) (.043) (.004)
Mother Less than High School -.037 -.203*** -.170*** .059 .034 -.013
(.037) (.037) (.039) (.065) (.070) (.009)
Father BA or More (vs. HS) .082*** .239*** .206*** .141** .088* .004
(.024) (.051) (.060) (.056) (.046) (.004)
Father Less than High School -.077** -.126** -.150** -.022 -.049 -.012
(.033) (.058) (.055) (.069) (.077) (.011)
Father Education Missing -.048 -.144** -.126* -.079 -.120* .008
(.041) (.067) (.065) (.077) (.069) (.009)
Married .047 .048 .110** .129* .091 -.001
(.036) (.049) (.051) (.076) (.078) (.008)
Teenage Mother -.008 .049 .017 -.039 -.165** .004
(.035) (.061) (.069) (.075) (.073) (.007)
Female .144*** .042 .140*** .471*** .514*** -.007*
(.018) (.028) (.031) (.037) (.035) (.004)
Black (vs. White) -.024 -.146** .085* .015 -.058 -.023**
(.029) (.056) (.047) (.070) (.072) (.009)
Hispanic (vs. White) -.030 -.188*** -.069 -.022 .050 -.023**
(.029) (.051) (.052) (.053) (.058) (.007)
Other Race (vs. White) .033 -.001 .102 .015 .052 -.007
(.031) (.064) (.071) (.053) (.077) (.005)
Immigrant Mother -.024 .015 .025 .004 .004 -.005
(.028) (.053) (.061) (.069) (.057) (.008)
(Continued on next page)
20 BROOKINGS | March 2012
School
Ready
Math Reading
Learning-
Related
Behavior
External-
izing
Behavior
Health
Low Birth Weight -.049* -.222*** -.102** -.157** -.010 -.014*
(.025) (.038) (.040) (.052) (.051) (.007)
Attended Preschool .089*** .126** .181*** .052 -.019 -.002
(.025) (.041) (.046) (.045) (.046) (.006)
Mother Smoked while Pregnant -.097** .038 .014 -.111 -.249** .011*
(.032) (.054) (.058) (.074) (.090) (.006)
Child was Breastfed -.007 .077** .090** .023 -.040 .006
(.020) (.032) (.038) (.045) (.043) (.006)
Mother in Poor/Fair Health .000 -.125* -.063 -.214** .002 -.039**
(.043) (.065) (.066) (.101) (.074) (.015)
Low in Cognitive Stimulation -.047* -.122** -.126** -.124** -.113* -.005
(.026) (.037) (.043) (.046) (.059) (.007)
Mother Low in Supportiveness -.098** -.256*** -.191*** -.152** -.063 -.002
(.030) (.055) (.054) (.070) (.067) (.009)
Notes: Significance levels: *** p<.01, ** p<.05, and * p< .10. Though not shown in the table, the full model also controls for maternal employment, use of relative care, use of non-relative care, use of center-based care before age four, child’s age in months, number of children and adults in the household, and dummies for missing values on selected variables. The child’s age in months and use of preschool or center-based care before age 4 have statistically significant effects on school readiness: babies born in April – August are 10-14 percentage points less likely to be school ready than babies born in January, and children using center-based care before age 4 are 4 percentage points less likely to be school ready than other children (significant at the 90 percent level). Use of center-based care prior to age four is measured roughly in this model but appears to be associated with small positive effects on math and reading skills, which are outweighed by larger negative effects on behavioral measures.
21 BROOKINGS | March 2012
Barnett, W.S. et al., The State of Preschool 2010. Rutgers, NJ: The National Institute for Early
Education Research.
Camilli, G., Vargas, S., Ryan, S., and Barnett. W., “Meta-Analysis of the Effects of Early “Meta-
Analysis of the Effects of EarlyEducation Interventions on Cognitive and Social
Development” Teachers College Record Volume 112, Number 3, March 2010, pp. 579–620.
Carneiro, P., Meghir, C., & Parey, M. (2007). Maternal education, home environments and the
development of children and adolescents. IZA Discussion Papers 3072, Institute for the
Study of Labor (IZA).
Chase-Lansdale, P. L., & Pittman, L. D. (2002). Welfare reform and parenting: Reasonable
expectations. Future of Children, 12 (1), 167-185.
Duncan, G. J., Ziol-Guest, K. M., & Kalil, A. (2010). Early childhood poverty and adult
attainment, behavior, and health. Child Development 81(1), 306-325.
Duncan, G.J., Dowsett, C.J., Claessens, A., Magnuson, K., Huston, A.C., Klebanov, P., Pagani,
L.S., Feinstein, L., Engel, M., Brooks-Gunn, J., Sexton, H., & Duckworth, K. (2007). School
readiness and later achievement. Developmental Psychology, 43(6), 1428-1446.
Duncan, G.J., Morris, P.A., & Rodrigues, C. (2011). Does money really matter? Estimating
impacts of family income on young children’s achievement with data from random-
assignment experiments. Developmental Psychology, 47(5), 1263–1279.
Evans, Gary W. (2004). “The environment of childhood poverty. American Psychologist, 2004,
59(2), 77-92.
Gennetian, L., Magnuson, K., & Morris, P.A. (2008). From statistical associations to causation:
What developmentalists can learn from instrumental variables techniques coupled with
experimental data. Developmental Psychology, 44(2), 381-394.
Isaacs, J. B., Sawhill, I.V., & Haskins, R. (2008). Getting ahead or losing ground: Economic
mobility in America. Washington, DC: The Brookings Institution
Isaacs, J. and Magnuson, K. (2011). Income and Education as Predictor’s of Children’s School
Readiness. Washington, DC: The Brookings Institution.
Isaacs, J. (2008). Impacts of Early Childhood Programs. Washington, DC: Brookigns Institution.
Johnson, Rucker C. and Robert F. Schoeni. 2007. “The Influence of Early-Life Events on
Human Capital, Health Status, and Labor Market Outcomes over the Life Course.”
Population Studies Center Report 07-616.
Levine, Judith A., Clifton R. Emery, and Harold Pollack. 2007. The Well-Being of Children Born
to Teen Mothers. Journal of Marriage and Family, 69, 105-122.
Lumley J, Chamberlain C, Dowswell T, Oliver S, Oakley L, Watson L, Interventions for
promoting smoking cessation during pregnancy (Review) The Cochrane Library 2009,
Issue 3.
Mayer, Susan (1997). What Money Can’t Buy: The Effect of Parental Income on Children’s
Outcomes. Cambridge: Harvard University Press.
McLoyd, V.C. (1990) The impact of economic hardship on black families and children:
Psychological distress, parenting, and socioemotional development. Child Development,
61(2), 311-346.
22 BROOKINGS | March 2012
Milligan, K., & Stabile, M. (2008). Do child tax benefits affect the wellbeing of children?
Evidence from Canadian child benefit expansions. NBER Working Paper Series: Working
Paper 14624. Cambridge, MA: National Bureau of Economic Research.
Nurse Family Partnership (2011). Nurse Family Partnership Snap Shot. Downloaded 2/14/2012
at http://www.nursefamilypartnership.org/assets/PDF/Fact-sheets/NFP_Snapshot.
Olds, D. L., Henderson Jr., C. R., Chamberlin, R., & Tatelbaum, R. (1986). Preventing child abuse
and neglect: A randomized trial of nurse home visitation. Pediatrics, 78, 65–78.
Olds, D. L., Henderson, C. R., & Kitzman, H. (1994). Does prenatal and infancy nurse home
visitation have enduring effects on qualities of parental caregiving and child health at 25
to 50 months of life? Pediatrics, 93(1), 89–98.
Olds, D. L., Henderson, C. R., Cole, R., Eckenrode, J., Kitzman, H., Luckey, D., et al. (1998). Long-
term effects of nurse home visitation on children’s criminal and antisocial behavior: 15-
year follow-up of a randomized controlled trial. JAMA: The Journal of the American
Medical Association, 280(14), 1238–1244.
Olds, D. L., Kitzman, H., Cole, R., Robinson, J., Sidora, K., Luckey, D. W., et al. (2004). Effects of
nurse home-visiting on maternal life course and child development: Age 6 follow-up results
of a randomized trial. Pediatrics, 114(6), 1550–1559.
Olds, D. L., Robinson, J., Pettitt, L., Luckey, D. W., Holmberg, J., Ng, R. K., et al. (2004). Effects
of home visits by paraprofessionals and by nurses: Age 4 follow-up results of a
randomized trial. Pediatrics, 114(6), 1560-1568.
U.S. Department of Health and Human Services, Administration for Children and Families.
Home Visiting Evidence of Effectiveness (HomVEE) Web Site, Tables on “Child
Development and School Readiness Outcomes,” July 2011. Downloaded February 3, 2012,
at http://homvee.acf.hhs.gov/Default.aspx.
U.S. Department of Health and Human Services, Public Health Service, Treating Tobacco Use
and Dependence: 2008 Update. Washington, DC: Author
Waldfogel,J., Craigie, T., and Brooks-Gunn, J. Fragile Families and Child Wellbeing.” The Future
of Children, Volume 20, Number 20, Fall 2010.
Waldfogel, J. and Washbrook, E. (2011) “Income-related gaps in school readiness in the U.S.
and U.K” in Smeeding et al., ed., Persistence, Privilege and Parenting. New York: Russell
Sage Foundation.
Wakschalg, L., Pickett, K. Cook, E., Benowitz, N., and Levnthal, B.(2002), “Maternal Smoking
During Pregnancy and Severe Antisocial Behavior in Offspring: A Review” American
Journal of Public Health June 2002, Vol 92, No. 6
Winship, S., Sawhill, I., and Gold, A. (2011) “Pathways to the Middle Class, “ Draft paper
presented November 8, 2011 at the Social Genome Project Conference at Brookings.
http://www.nursefamilypartnership.org/assets/PDF/Fact-sheets/NFP_Snapshot�
http://homvee.acf.hhs.gov/Default.aspx�
- Introduction
- Figure 2: Likelihood of Scoring Very Low (Failing to Be School Ready) on Measures of School Readiness, by Poverty Status
- Figure 3: Likelihood of Being Ready for School at Age Five, by Poverty Status at Birth and Selected Child and Family Characteristics
- /Figure 4: Poor Families Differ from Moderate/High Income Families on Many Characteristics that May Affect School Readiness
- /Figure 6: Various School Readiness Gaps (Before Controlling for Confounding Factors)
I. Poor Children Are Less Ready for School at Age Five than Other Children
Figure 1: Likelihood of Being Ready for School at Age Five, by Poverty Status at Birth
II. Why Are Poor Children Less Ready for School?
Figure 5: School Readiness Gap (Poor vs. Moderate/High Income), Various Levels of Controls
III. Multiple Targets of Opportunity for Addressing School Readiness of Poor Children
Source: Brookings analyses of data from the Early Childhood Longitudinal Study-Birth Cohort (ECLS-B)
Figure 7: School Readiness Gaps (Before and After Controlling for Confounding Factors)
IV. Comparison of Three Interventions to Improve School Readiness
Table 1: Simulated Effects of Three Different Interventions
Conclusion
Appendix
Measurement of School Readiness
Independent Variables
Regression Results
References