1. No category

EdPolicyWorks
Working Paper:
Do the Effects of Early Childhood Interventions
Systematically Fade? Exploring Variation in the
Persistence of Preschool Effects
Daphna Bassok1, Chloe R. Gibbs1 & Scott Latham1
Early childhood education receives significant public attention as a cost-effective approach to closing achievement
gaps and improving the life prospects of disadvantaged children. However, critics point to the relatively quick
evaporation of academic benefits for participants demonstrated in several recent experimental studies. Our paper
is the first to employ both kindergarten cohorts of the nationally representative Early Childhood Longitudinal
Study (ECLS), 1998 and 2010, to describe the extent to which preschool effects fade over time and to assess variation in those patterns of convergence. Notably, the association between preschool participation and cognitive
outcomes fades more rapidly in the 2010 kindergarten cohort than in 1998. Whereas in the 1998 data, a statistically significant “preschool advantage” is evident in both reading and math at the end of first grade, in the more
recent cohort no differences are observed by the end of the kindergarten. The rapid fade-out is observed for children who attended both full and half-day preschool programs and does not depend on several proxies of kindergarten classroom quality. The results suggest that the rate of fade-out of preschool effects may be accelerating
over time. Interestingly, across both kindergarten cohorts, we document a persistent positive association between
preschool participation and first grade cognitive outcomes for black children. Policy implications are discussed.
University of Virginia
1
Updated January 2015
EdPolicyWorks
University of Virginia
PO Box 400879
Charlottesville, VA 22904
EdPolicyWorks working papers are available for comment and discussion only. They have not been peer-reviewed.
Do not cite or quote without author permission. This working paper was retrieved from:
http://curry.virginia.edu/uploads/resourceLibrary/36_Preschool_Fade_Out.pdf
Acknowledgements: Scott Latham was supported by a grant from the Institute of Education Sciences, U.S.
Department of Education, through Grant #R305B090002 to the University of Virginia. Opinions reflect those of
the authors and do not necessarily reflect those of the granting agency.
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
Working Paper © 2015 Rector and Visitors of the University of Virginia. For more information please visit www.curry.virginia.edu/edpolicyworks or contact [email protected]
Preschool Fade-Out
DO THE EFFECTS OF EARLY CHILDHOOD INTERVENTIONS SYSTEMATICALLY FADE?
EXPLORING VARIATION IN THE PERSISTENCE OF PRESCHOOL EFFECTS
Daphna Bassok, Chloe R. Gibbs & Scott Latham
Early childhood interventions are touted as powerful, cost-effective and equity enhancing
investments (Bassok & Loeb, 2015; Cunha & Heckman, 2010; Shonkoff & Phillips, 2000). In his
2013 and 2014 State of the Union addresses, for example, President Obama highlighted the longterm impacts of high-quality early childhood interventions as a motivation for his administration’s
heightened focus on preschool expansion. During a White House Summit on Early Childhood in
the winter of 2014, the president again cited the long-term benefits of early childhood opportunities
when he announced over $750 million dollar federal investment in early childhood education
opportunities. Although a large body of research suggests that early childhood interventions indeed
have significant impacts on both short and long-term outcomes, a number of recent experimental
studies show that the benefits from large-scale preschool programs dissipate quickly (Lipsey, Hofer,
Dong, Farran, & Bilbrey, 2013; Puma et al., 2012).
Researchers, policy-makers, and practitioners have struggled to reconcile the observed “fadeout” of preschool effects with the large body of research demonstrating that early childhood
programs have large and persistent benefits. One possibility is that large-scale preschool programs
are of substantially weaker (and more variable) quality than “model” programs that have yielded
long-lasting effects. Citing this rationale, some have argued that more resources are needed to
improve the quality of preschool programs while others have cautioned that investing public
resources in these programs is unlikely to yield the desired benefits (Bartik, 2013; Whitehurst &
Armor, 2013). Another hypothesis is that characteristics of subsequent schooling might play a role in
sustaining preschool effects (Currie & Thomas, 2000; Reynolds, Magnuson, & Ou, 2010). Indeed,
the latest round of the Race to the Top Early Learning Challenge (RTT-ELC), a federal grant
competition to improve early childhood opportunities, identified “creating preschool through third
grade approaches to sustain improved early learning outcomes through the elementary grades” as
one of the key components for a competitive application. The U.S. Departments of Health and
Human Services and Education’s recently initiated Preschool Development Grants similarly
articulated the expectation that grantees “align preschool programs within a birth through thirdgrade continuum of services” (U.S. Department of Education, 2014).
Despite the heightened attention paid to early childhood education in general, and to the
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
1
Preschool Fade-Out
issue of fade-out in particular, our understanding of fade-out and its mechanisms remains
underdeveloped. There is a lack of research examining this issue at a national level using recent data.
Such evidence is critical because the early childhood landscape has changed considerably since 1998,
when the first nationally-representative study of children entering kindergarten took place. Since that
time, enrollment in state-sponsored preschool has increased dramatically, nearly doubling since
2000, and the visibility of early childhood education as a policy issue has grown substantially
(Barnett, Carolan, Squires, & Brown, 2014).
Our study aims to fill this gap. We leverage two large, nationally representative datasets to
describe the extent to which the association between preschool participation and children’s cognitive
and behavioral outcomes fades between kindergarten entry and the spring of first grade. The study
makes two key contributions. First, because the datasets overlap substantially in their content, we are
able to provide the first rigorous examination of whether patterns of fade-out have changed over a
period characterized by heightened investment in early education. Second, we add to a relatively
limited literature exploring whether characteristics of the preschool experience or the subsequent
schooling environment moderate patterns of fade-out. Understanding how the effects of early
childhood education programs differ across children in these different contexts is foundational in
identifying strategies for early policy interventions with lasting impact.
We examine four related research questions:
1. Do children who attended preschool in the year before kindergarten outperform children
who did not participate in formal care with respect to cognitive or behavioral outcomes?
2.
Are there remaining differences between these two groups by the spring of first grade?
3. Are there differences in patterns of persistence depending on the intensity of the preschool
experience or the characteristics of the subsequent kindergarten experience?
4. Do these patterns differ across demographic groups as defined by race, socio-economic
status and English language skills?
We answer each of these questions using both waves of the ECLS-K data and discuss similarities,
differences, and potential implications for policy.
Background
The Effects of Preschool and the Role of “Fade-out”
Two well-known and oft-cited experiments provide compelling evidence of the long-term benefits
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
2
Preschool Fade-Out
of preschool programs (Campbell et al., 2012; Schweinhart et al., 2005). Both demonstrated that
early childhood programs have the potential to yield lasting benefits that far outweigh the costs.
However, it is unclear to what extent the findings from these small, decades-old studies generalize to
today’s context. In particular, the participants in these studies were disproportionately extremely
low-income, black children. In addition, the preschool services provided were far more intensive and
costly than most large-scale programs today. Finally, comparison group children in these studies
generally stayed home or with relatives. By contrast, most young children today receive some type of
non-relative care.
A large body of evidence has emerged over the past decade that addresses some of the
generalizability concerns related to the experimental evidence and enhances our understanding of
the benefits of early childhood education. This research includes a set of careful quasi-experimental
studies evaluating the short-term impacts of more recent preschool programs as well as studies
examining the longer-term impacts of large preschool programs. For the most part, studies
examining the short-term benefits of preschool participation suggest important benefits (Bassok,
2010; Gormley, Gayer, Phillips, & Dawson, 2005; Weiland & Yoshikawa, 2013; Wong, Cook,
Barnett, & Jung, 2008).
Quasi-experimental studies also suggest these benefits persist (Fitzpatrick, 2008; Ladd,
Muschkin, & Dodge, 2014). Notably, Reynolds et al (2011) evaluated the Chicago Child-Parent
Centers, which provided comprehensive services to low-income children and families, using a
matched control design and estimated that the preschool services produced benefits about seven
times greater than their costs. Ludwig & Miller (2007) evaluated the long-term effects of Head Start
using a regression discontinuity design and found that participation increased children’s educational
attainment. Deming (2009) compared children who attended Head Start classrooms to their siblings
who did not and found benefits of Head Start on an index of young adult outcomes. Taken
together, these studies provide evidence that larger-scale programs can have important short and
long-term benefits.
Despite the impressive long-term results found in these studies however, a recurring pattern
in the literature is that the short-term cognitive benefits often dissipate early in elementary school.
This “convergence” or “fade-out” of test scores has been documented across many studies that
include a wide range of preschool settings (Currie, Garces, & Thomas, 2000; Deming, 2009; Lee &
Loeb, 1995; Lipsey et al., 2013; Ludwig & Miller, 2007; Magnuson, Ruhm, & Waldfogel, 2007b).
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
3
Preschool Fade-Out
Notably, results from the national Head Start Impact Study, a large, randomized trial of Head Start
participation, showed that at third grade, Head Start participants did not systematically outperform
their control group peers on any of the developmental domains examined (Puma et al., 2012). In
fact, many of the gains realized by participants at the end of the program year had eroded as soon as
kindergarten or first grade (Puma et al., 2010).
Differences in Preschool Effects by Child Characteristics
Although researchers often focus on measuring the overall effects of preschool participation,
increasingly there is also interest in understanding heterogeneity of effects across a number of
dimensions including child characteristics, preschool characteristics and elementary school
characteristics. Each has concrete policy implications. For instance, if certain types of children
benefit disproportionately from preschool participation, there may be an argument for programs
that “target” these specific types of children. Similarly, if the observed benefits are greatest in
programs that meet a certain level of quality or intensity, it may be important to use policy levers to
ensure these quality standards are met.
While studies have shown that on average all children benefit from preschool participation,
there is evidence that the magnitude of the “preschool advantage” varies across groups. For
instance, researchers have found evidence of more pronounced benefits among low-income children
(Magnuson, Ruhm, & Waldfogel, 2007a; Weiland & Yoshikawa, 2013) and children with low
maternal education (Havnes & Mogstad, 2011; Peisner-Feinberg et al., 2001; Waldfogel, 2002).
Other studies have shown greater benefits among Hispanic children (Gormley et al., 2005; Loeb,
Bridges, Bassok, Fuller, & Rumberger, 2007), black children (Bassok, 2010) and children with low
initial cognitive scores (Zill et al., 2001).
Does the Intensity of Exposure to the Preschool Experience Matter?
A large body of evidence demonstrates the importance of preschool quality, broadly defined, in
explaining effects (Mashburn et al., 2008; Peisner-Feinberg et al., 2001). The intensity of the
program, as measured by hours per day, is one potentially important feature of the program’s
quality. Attending a program a few hours per week may yield quite different effects than attending
full-time (Herry, Maltais, & Thompson, 2007; Reynolds et al., 2014). Loeb et al. (2007) found that
children who spent more hours per week in center-based care had higher cognitive scores at the start
of kindergarten, and that this result was most prominent among low income children. Walters
(forthcoming) finds similar results in Head Start centers, with centers offering full-day programming
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
4
Preschool Fade-Out
improving cognitive skills more than those on a half-day schedule. A number or large studies also
found that more hours in preschool settings were associated with increases in behavioral problems, a
pattern replicated across a number of other large studies (Belsky, 2002; Loeb et al., 2007; Vandell et
al., 2010). We are not aware of prior studies that examined whether hours in preschool may explain
patterns of fade-out.
Subsequent Schooling Experiences as a Way to Combat Fade-out
In recent years, early childhood advocates have called for preschool to third grade (PK-3rd)
initiatives, which are designed to provide aligned and consistently high-quality care starting before
school entry and continuing through the third grade (Bogard & Takanishi, 2005). The premise of
these initiatives, in part, is that in order to sustain the benefits of even a high-quality preschool
experience, children must have access to high-quality learning experiences in early elementary school
as well. Oftentimes, however, this is not the case. Lee & Loeb (1995) first brought this notion to the
fore with their finding that children who attended Head Start classrooms went on to attend middle
schools that were lower on a number of dimensions of quality than did similar children who did not
attend Head Start. Expanding upon this work, Currie & Thomas (2000) suggested that among black
children, poor quality schooling experiences may have mitigated an initially positive effect of Head
Start.
Reynolds, Magnuson, & Ou (2010) reviewed the available PK-3 literature and found that
programs with coordinated services across preschool and early elementary school had overall
benefits in terms of achievement and school adjustment. Similarly, Reynolds, Ou, & Topitzes (2004)
found that high-quality elementary schools were an important mediator of the long-term benefits of
preschools.
However, counter to the hypothesis that high-quality classrooms, broadly defined, foster
more persistent preschool benefits, Magnuson, Ruhm, & Waldfogel (2007b), which specifically
examined whether preschool fade-out was moderated by later schooling experiences, showed that
the preschool advantage with respect to cognitive test scores was actually more persistent for
children who attended classrooms with larger class sizes and lower levels of reading instruction. Their
finding may suggest that in higher quality classrooms those children who did not attend preschool
experience an effective enough learning environment to compensate for the lack of preschool and
allow them to “catch up.” That said, the specific school characteristics examined in this study are
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
5
Preschool Fade-Out
just two of a host of early elementary school characteristics that have been shown to matter for
learning during the kindergarten year and may therefore mitigate fade-out.
For instance, a number of studies have demonstrated that children benefit from full-day
more than part-day kindergarten settings (Cannon, Jacknowitz, & Painter, 2006; Lee, Burkam,
Ready, Honigman, & Meisels, 2006). Gibbs (2014) showed that random assignment to and
participation in full-day kindergarten was associated with a one-third of a standard deviation
advantage on end-of-kindergarten literacy skills.
There is considerable evidence that smaller kindergarten classes are beneficial for children.
Experimental evidence from Project STAR in Tennessee demonstrates the benefits of small classes
on math and reading scores in elementary school (Nye, Hedges, & Konstantopoulos, 2001), and on
later outcomes such as middle school achievement, college test taking (Krueger & Whitmore, 2001),
and college attendance (Chetty et al., 2011).
Multiple studies have also used Project STAR to explore the impact of kindergarten peers.
Boozer & Cacciola (2001) & Graham (2008) both find that children with more able peers have
higher cognitive scores in later grades. Chetty et al. (2011) showed that children randomly assigned
to peers with higher test scores in kindergarten also went on to have higher earnings and college
attendance rates.
Finally, several studies demonstrate the importance of kindergarten curriculum and show
that children learn more in kindergarten classrooms that spend more time on challenging rather than
basic math and literacy content (Claessens, Engel, & Curran, 2013; Engel, Claessens, & Finch, 2013).
Given the robust body of evidence demonstrating that differences in kindergarten experiences can
have important implications for learning, it is logical to wonder whether the fade-out of preschool
effects may differ depending on subsequent kindergarten experiences.
The Current Study
This paper adds to the existing literature in several important ways. First, we are the first study we
are aware of to leverage newly-released data from the National Center for Education Statistics’
(NCES) Early Childhood Longitudinal Study (ECLS-K 2010), a large, nationally-representative
dataset tracking a cohort of kindergarteners through the end of first grade, to examine to what
extent the association between preschool participation and children’s outcomes persist throughout
the first two years of elementary school. This dataset is a follow-up to the ECLS-K 1998, which
tracked a similar cohort of kindergarteners who entered school nearly two decades ago. The ECLSEdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
6
Preschool Fade-Out
K 1998 has been used extensively to examine a host of questions about the effects of preschool
participation both at school entry and beyond (Loeb et al., 2007; Magnuson et al., 2007a, 2007b).
The years between when these two groups of children entered kindergarten were characterized by
substantial expansion of early childhood opportunities as well as a heightened focus on improved
preschool quality. It is therefore important to revisit questions of fade-out within the more recent
data and examine whether patterns uncovered in the earlier round align with those seen today.
In addition to comparing patterns of fade-out across datasets, we then examine
heterogeneity in these patterns along three dimensions: demographic characteristics of children;
intensity of the preschool intervention; and quality of the kindergarten experience. While existing
research suggests low-income children experience the greatest gains from preschool participation,
and there is some evidence that the initial effects of preschool are larger for black and Hispanic
children relative to their white peers, less is known about differential patterns of fade-out across
groups. We examine this matter here. Full-day preschool is often championed because of its
potential relationship with children’s developmental outcomes and its practicality for families,
particularly low-income, single parents (Robin, Frede, & Barnett 2006), but again, there is little
evidence of whether patterns of preschool fade-out are moderated by the length of the preschool
day. Finally, following the rationale for PreK-3rd interventions we examine the extent to which fadeout of preschool effects by the end of the first grade may be related to characteristics of the
kindergarten experience including structural features like class size and length of day, as well as more
process-oriented features such as “transition practices” to ease the shift from preschool to
elementary school or time spent on literacy instruction.
Methods
This study leverages data from two longitudinal cohorts of the Childhood Longitudinal Study
(ECLS-K 1998 & ECLS-K 2010), which are nationally representative samples of children entering
kindergarten in the fall of 1998 and 2010 respectively. The ECLS-K 1998 tracked children through
8th grade, with 7 waves of data collection occurring in fall and spring of kindergarten, fall and spring
of 1st grade, and the spring of grades 3, 5, and 8. Data collection for the 2010 cohort is ongoing, and
data have so far been released for fall and spring of both the kindergarten and first grade year. Data
collection is also planned for the fall and spring of second grade, and the spring of grades 3-5
(Tourangeau et al. 2013).
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
7
Preschool Fade-Out
The datasets were designed, in part, to facilitate comparisons across cohorts. To a large
extent, the content and data collection procedures from the new cohort were modeled after the
original ECLS-K and, as a result, these datasets provide a unique opportunity to compare the
experiences of children in the United States over time.
Participants
The initial ECLS-1998 and ECLS-2010 samples consist of approximately 21,400 and 18,150 children
respectively. 1 We restrict these samples to first-time kindergarteners. To facilitate longitudinal
analysis, we also restrict our sample to children who had non-missing outcome data in both the fall
and spring of kindergarten and the spring of first grade. Since the sample of children who had valid
cognitive assessments in each wave is substantially larger than the sample that had valid behavioral
assessments, we construct separate analytic samples for analyzing the cognitive and behavioral
outcomes (henceforth the “cognitive” and “behavioral” samples). 2
Our cognitive samples consist of 12,450 and 12,100 children in the 1998 and 2010 cohorts
respectively. The behavioral samples consist of 11,150 and 9,250 children. We weight our analysis to
make it nationally representative of children attending kindergarten in their respective years. The
weights we use were developed as part of the ECLS-K and account for possible non-random
attrition from the study.
We conduct multiple imputation using chained equations to avoid the bias that may arise
when analyzing complete-case data. Our imputation model accounts for all the covariates that we
later include in our analysis (i.e. demographics, preschool participation, kindergarten characteristics,
and variables listed in Table A1), and we impute independent but not dependent variables. Multiple
imputation was conducted using the MI command in Stata, and five imputed datasets were
generated.
Measures
Preschool experience. During the fall kindergarten data collection, parents were asked
detailed questions about the type and amount of care their children received in the prior year.
Specifically, parents were asked to report whether their children attended a “day care center, nursery
school, preschool or prekindergarten program” and how many hours per week they attended.
In keeping with NCES guidelines, we round all sample sizes to the nearest 50.
We also run all analyses on a fixed sample of children who have both cognitive and behavioral outcomes.
Results, available from the authors, are substantively unchanged.
1
2
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
8
Preschool Fade-Out
Although parents across both cohorts were also asked about Head Start participation, these
items are inconsistent across the two waves of data. In the 1998 parent surveys, parents were asked a
detailed set of questions about Head Start participation, analogous to the questions asked about
other types of center-based care (e.g. hours attended). In addition, whenever parents reported their
child attended Head Start, data collectors independently verified whether the program was actually a
Head Start program and whether the child had actually attended. In contrast, in the 2010 survey
parents were only asked a single question about Head Start participation and it was a follow-up
within the larger section on center-based care participation. Specifically, the survey only asked,
“Were any of child’s care arrangement{s} in a day care center, nursery school, preschool, or
prekindergarten program in the year before kindergarten Head Start?" This variable allows
researchers to determine whether a parent reported any Head Start utilization in the year prior to
kindergarten, but there is no way to disentangle if Head Start was their primary care provider or how
many hours the child went to Head Start.
Unfortunately, this discrepancy in survey questions across waves makes it impossible to
make comparable “Head Start” categories across the two cohorts. Our measure of preschool
participation is therefore defined as any child who attended a day care center, nursery school,
preschool or prekindergarten program for five or more hours per week and whose parents reported
they did not report that they ever used Head Start. We use the terms preschool and center-based care
interchangeably in the paper to refer to this definition of preschool participation in the year prior to
kindergarten.
We also subdivide the preschool participation variable into “part time” (between 5 and 20
hours/week) and “full time” (more than 20 hours/week). By construction, the sum of children in
part- and full-time care equals the number of children in preschool.
Kindergarten characteristics. We explore interactions between preschool participation and
a number of kindergarten characteristics that we hypothesize may influence children’s learning
during kindergarten to determine whether these characteristics moderate patterns of fade-out. Our
measures of kindergarten quality include full-day kindergarten, small class size, peer exposure to
preschool, along with an index of kindergarten quality.
Information about kindergarten classrooms was drawn from teacher surveys conducted in
the fall of the kindergarten year. We construct an indicator for full-day kindergarten from teacher
reports of the number of hours children spend in class each day, with full-day defined as five or
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
9
Preschool Fade-Out
more hours, consistent with the literature on full-day kindergarten (Gibbs, 2014). We define small
classes as those with 17 children or fewer, which represent the bottom quartile of kindergarten class
sizes. We measure peer exposure preschool as the percentage of children in a kindergarten
classroom who attended preschool (excluding the focal child). We hypothesize that the benefits of
preschool participation may persist more for children with more peers who also experienced
preschool.
We also construct a crude index of kindergarten quality defined as the sum of six indicator
variables. These include whether the child is in a full-day kindergarten, whether the child is in a small
class, whether the student’s kindergarten school is co-located with a preschool, and three other
indicators defined as whether a student is above the sample medians for peer preschool attendance,
for use of kindergarten transition practices to ease the transition from preschool, and the amount of
time spent on reading in kindergarten. Each of these measures is meant to map onto measures of
the kindergarten experience that have been shown to affect learning (Claessens et al., 2013;
Magnuson et al., 2007a; Schulting, Malone, & Dodge, 2005).
Outcomes. We consider the relationship between children’s participation in preschool and
their cognitive as well as behavioral outcomes. The cognitive outcomes are drawn from direct
assessments of student literacy and math skills, conducted in the fall and spring of kindergarten, and
in the spring of first grade. The literacy assessment was designed to measure basic skills such as
letter recognition and print familiarity along with vocabulary and listening comprehension. The math
assessment measured conceptual knowledge, procedural knowledge, and problem solving ability. To
minimize burden on children, the tests were administered in two parts. First, children took a routing
test, which provided a crude estimate of ability. They were then given an easy, medium, or hard test
based on their score on the routing test. For this reason, children did not all answer the same set of
items. Instead, scores were equated across children using item response theory, and the resulting
scale scores represent estimates of what children would have scored if they had been administered
every item in the assessments.
Our behavioral outcomes are drawn from teacher reports of student behavior in
kindergarten and first grade. Specifically, we use measures of student externalizing behavior and selfcontrol. The externalizing scale asks how often children argue, fight, get angry, act impulsively, or
disturb classroom activities. For this scale, lower scores represent more positive outcomes. The selfcontrol scale asks whether children respect others’ property rights, control their temper, accept peer
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
10
Preschool Fade-Out
ideas, and respond appropriately to peer pressure. Higher scores represent more positive outcomes
for this scale. We standardized both cognitive and behavioral outcomes to have a mean of zero and
standard deviation of one.
Control variables. Both ECLS-K datasets contain a rich set of potential covariates. We
include many of these in our analysis to account for the possible selection bias associated with child
and family characteristics. For a complete list of these variables, see Appendix Table 1. We use an
identical set of controls in our analysis of the 1998 and 2010 data to facilitate comparison across
cohorts.
We control for a number of demographic characteristics that have demonstrated associations
to cognitive and behavioral outcomes. These include race, socioeconomic status, parental education
(both mother’s and father’s), parental employment status, maternal age and whether English is the
primary language spoken at home. We also control for region of the country and urbanicity.
In addition, we include measures of many family characteristics that may be associated with
child outcomes. These account for parental involvement in school, expectations for student
achievement, and beliefs about what is important for school readiness. We also control for home
activities (e.g., reading books, singing songs, and playing games) and extra-curricular activities (e.g.,
dance class, athletics, and music). Other measures of household inputs include the number of books
in a child’s home, whether the family owns a computer, and time spent watching TV. Finally, we
have measures of the relationship between parents and their children. This includes how often
parents hit or spank their child, whether the family regularly eats breakfast/dinner together, and
measures of warmth and affection.
Analytic Approach
We use ordinary least squares to estimate the effect of preschool participation on our four outcome
variables (reading scores, math scores, externalizing, self-control). For each relationship of interest,
we estimate two sets of regressions, one that treats preschool participation as a single construct and
one that separates preschool participation into part- and full-time care. These equations take the
form:
𝑌𝑖 = 𝛽0 + 𝛽1 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖 + 𝛾𝑗 𝑿𝑖𝑖 + 𝜀𝑖
𝑌𝑖 = 𝛽0 + 𝛽1 𝑃𝑃𝑃𝑃𝑖 + 𝛽2 𝐹𝐹𝐹𝐹𝑖 + 𝛾𝑗 𝑿𝑖𝑖 + 𝜀𝑖
(1)
(2)
where Y is the outcome of interest for individual child i, PRESCHOOL is an indicator that takes on
a value of one if the child’s care experience in the year before kindergarten meets the definition of
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
11
Preschool Fade-Out
preschool and zero otherwise, and X is a vector of covariates. Included in X is an indicator for
whether a child attended a Head Start program, allowing us to interpret the counterfactual as
children who did not attend any type of formal center-based care, including Head Start, in the year
before kindergarten. All models include heteroskedasticity-robust standard errors clustered at the
kindergarten classroom level.
We estimate models 1 and 2 three times for each outcome variable, once each for scores
from fall and spring of kindergarten as well as the spring of first grade. These estimates explore the
extent to which outcomes converge between preschool participants and children who experienced
no formal care, both during the kindergarten year and at the end of first grade. We also run these
models separately by demographic subgroups, restricting the regression estimation to black children,
Hispanic children, children in the lowest SES quintile, and children whose primary language at home
is not English.
Next, we explore whether the association between preschool participation and child
outcomes differs for children who attend different types of kindergarten classrooms. We run a
separate model for each measure of the kindergarten experience (full-day kindergarten, small class
size, peer preschool exposure, and the kindergarten quality index). In each regression, we interact
preschool participation with the measure of kindergarten experience. The interacted models are as
follows:
𝑌𝑖 = 𝛽0 + 𝛽1 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖 + 𝛽2 𝐾𝐾𝐾𝐾𝑖 + 𝛽3 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 ∗ 𝐾𝐾𝐾𝐾𝑖 + 𝛾𝑗 𝑿𝑖𝑖 + 𝜀𝑖
𝑌𝑖 = 𝛽0 + 𝛽1 𝑃𝑃𝑃𝑃𝑖 + 𝛽2 𝐹𝐹𝐹𝐹𝑖 + 𝛽3 𝐾𝐾𝐾𝐾𝑖 + 𝛽4 𝐹𝐹𝐹𝐹 ∗ 𝐾𝐾𝐾𝐾𝑖 + 𝛾𝑗 𝑿𝑖𝑖 + 𝜀𝑖
(3)
(4)
where Y is the outcome of interest for individual child i, PRESCHOOL is an indicator that takes on
a value of one if the child’s care experience in the year before kindergarten meets the definition of
preschool and zero otherwise, Kexp is a measure of child i’s kindergarten classroom experience,
PRESCHOOL*Kexp is the interaction term of preschool and kindergarten experience, and X is a
vector of covariates. As with the previous models, we include in X an indicator for whether children
attended a Head Start program. In equation (3), 𝛽3 is the coefficient of interest, capturing the
composite relationship of preschool participation and subsequent kindergarten classroom
experience. Equation (4) replicates the same approach with the part- and full-time indicators to
capture preschool participation, and interactions of full-time care and kindergarten experience. 𝛽4
captures the combination of full-time preschool participation and kindergarten experiences in their
association with outcomes of interest.
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
12
Preschool Fade-Out
To the extent that selection into either the initial preschool experience or the subsequent
kindergarten experiences is confounded by unobservable characteristics over and above those
included in our vector of covariates, our coefficients may be biased estimates of the impact of
preschool or kindergarten experiences. That said, our analytic approach and our reliance on the
richness of the ECLS datasets to address selection issues are consistent with the observational
literature and previous studies employing the ECLS-K data.
Results
The Early Childhood Landscape, 1998 & 2010
We begin by presenting descriptive statistics about our analytic sample, separately across waves, to
provide context about changes in the early childhood landscape over the period studied. In the first
panel of Table 1, we present the percentage of children who attended (non-Head Start) preschool,
both overall and disaggregated by part- and full-time preschool participation. Overall, participation
was stable over the period examined. The lack of increase is notable in light of the documented
increase in state-funded pre-kindergarten, and may reflect a shift from private to public preschool
without any overall participation increase. Participation in full-day preschool was also largely
unchanged.
The lower panels of Table 1 disaggregate these findings across demographic subgroups.
Here too, the patterns are relatively stable across waves. For instance, both in 1998 and 2010, white
children were far more likely to attend (non-Head Start) preschool programs, and particularly partday programs, than either black or Hispanic children. For instance, in 2010, 40 percent of white
children were enrolled in half-day preschool, but only 9 percent of black children and 19 percent of
Hispanic children did so. One notable change across waves is the 7 percentage point drop in
preschool participation among Hispanic children, and the smaller drop among black children from
42 to 38 percent. These drops are not fully explained by shifts towards enrollment in Head Start.
The percentage of children enrolled in any formal care including those whose parents indicated some
enrollment in Head Start (not shown in the table), dropped from 60 to 57 percent for Hispanic
children and from 76 to 66 percent among black children.
Children in the lowest SES quintile participate in (non-Head Start) preschool programs in
the year before kindergarten at much lower rates than middle and high-SES children, and that
pattern has been remarkably stable over time. This pattern still holds, though it is much less
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
13
Preschool Fade-Out
pronounced, if we consider any formal care including Head Start. Although there has been no
overall increase in preschool participation, we do note that among children in the highest SES
quintile there has been a shift from part to full-day programs.
Table 2 provides analogous information about each cohort’s kindergarten experience. The
most notable change across the two waves is the dramatic increase in full-day kindergarten
participation from 1998 to 2010 from 55 to 80 percent. While all student subgroups experienced
notable growth in full-day kindergarten participation, the increases are particularly pronounced
among Hispanic children, from 48 to 85 percent.
Overall, the proportion of children experiencing a small class size in kindergarten and the
proportion of kindergarten peers that experienced preschool care did not change over this time
period. However, black and Hispanic children in 2010 were more likely to experience a small
kindergarten class size than in 1998. Our crude kindergarten quality index increased only slightly,
and this change was driven almost entirely by the increase in full-day care kindergarten.
Preschool Participation and Children’s Outcomes at School Entry
To explore the relationships between preschool participation and fall and spring academic and
behavioral outcomes, we deploy the models described in equations (1) and (2). The regression
results for the 2010 wave are presented in Table 3 and analogous results for 1998 are in Table 4.
Consistent with the earlier literature leveraging the ECLS-K, children who participated in
preschool in the year prior to kindergarten outperform their peers in math and literacy skills at
kindergarten entry. For example, in 2010, preschool participants started kindergarten performing .10
and .09 of a standard deviation higher on reading and math respectively. The patterns in 1998 were
similar, though somewhat larger (.14 and .12). As has been documented in prior research, preschool
participants in both iterations of the ECLS exhibit more externalizing behavior and less self-control
in the fall of kindergarten relative to peers who did not experience any formal care. However, the
strength of these associations was roughly half as pronounced for the more recent cohort.
The Persistence of Preschool Effects
In the 2010 data, we observe no association between preschool participation and reading or math
scores by the end of the first grade year. In fact, already by the spring of the kindergarten year the
positive, statistically significant associations observed at school entry have dissipated entirely. This
pattern differs from what we (and others) have documented using the 1998 data where children who
participated in preschool were still outperforming their peers in informal care by about one
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
14
Preschool Fade-Out
twentieth of a standard deviation on both reading and math direct assessments. Although we
observe a diminishing association between preschool participation and children’s outcomes for both
cohorts, for the most recent cohort the association falls to zero within the kindergarten year,
whereas in the older cohort statistically significant positive associations persist. Notably, in both
datasets the negative effects on socio-emotional/behavioral skills are also still present and in some
cases more pronounced by the spring of first grade.
Heterogeneity in Persistence by Preschool and Kindergarten Characteristics
One goal of the current analysis was to explore whether the relationship between preschool
participation and child outcomes differed depending on characteristics of the child’s preschool or
kindergarten.
Full-time preschool. We hypothesized that children attending full-day preschool programs
would display a greater advantage in the fall of kindergarten and higher levels of persistence. In
general, this is not the pattern seen in the 2010 data. Children in both part- and full-time care
outperform their peers in reading and math at kindergarten entry, and the magnitude of the
coefficients is actually slightly larger for children who attended part-time preschools. By the spring
of first grade, these associations have dissipated across the board (with the exception of a marginally
significant positive relationship between part-time preschool participation and reading outcomes.)
We do find that the negative relationship between preschool attendance and socio-emotional
and behavioral measures is driven entirely by full-time attendees as the associations for part-time
attendees are not significantly different from zero.
The patterns in 1998 are fairly similar. Again we find that children in both part- and full-time
care outperform their peers in reading and math at kindergarten entry. In addition, similar to the
2010 data, we also observe that the negative associations between preschool participation and
behavioral outcomes are far more pronounced for children who attend full-day programs. For the
earlier cohort, we do find some modest support for our initial hypothesis, in that that the persistence
of a preschool advantage is somewhat stronger, particularly for math, among full-day relative to halfday attendees.
Kindergarten quality. We also hypothesized that the persistence of the associations
between preschool participation and child outcomes may differ depending on the characteristics of
children’s kindergarten experiences (e.g. the length of the kindergarten day, the kindergarten class
size, and the proportion of kindergarten peers that experienced preschool). Table 5 presents results
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
15
Preschool Fade-Out
from models (3) and (4). We regressed outcomes from the spring of first grade for the 2010 cohort
on indicators of preschool participation, kindergarten quality and the interaction of the two. The
coefficient of interest is on the interaction between preschool participation and each measure of the
kindergarten classroom. These interaction terms indicate whether the association between preschool
participation and children’s outcomes is more pronounced for children who experienced higher
quality kindergarten classrooms.
Panel A of Table 5 shows no evidence of such interactions. None of the 16 interactions
examined (4 outcomes by 4 kindergarten quality measures) were statistically different from zero. In
the bottom panel we present similar results, but interact full-day preschool participation with each of
the measures of kindergarten quality. Again, we find no evidence of any interactions between the
preschool experience and subsequent kindergarten quality measures. Notably, the results shown in
Table 5 are substantively unchanged if we focus on the 1998 rather than the 2010 cohort, and are
also similar if we consider outcomes at the end of the kindergarten year rather than the end of the
first grade year.
Heterogeneity across Demographic Groups
Results presented thus far focused on average patterns across all children. Our final research
question explores heterogeneity in patterns across children focusing in particular on race, socioeconomic status and English Language knowledge. In this section we explore whether either the
initial association between preschool participation and child outcomes or its persistence varies across
groups. Tables 3 and 4 present these results. We caution that the standard errors in these analyses
are much larger due to reduced sample size and that in some cases the cell sizes for these subgroup
analyses may be restricted given our extensive controls and specification of the independent
variable(s).
Among the 2010 cohort, we observe, nearly uniformly, that children who attended preschool
in the year prior to kindergarten outperform their peers in both reading and math in the fall of
kindergarten (the exception is that ELL children who attend preschool do not outperform peers in
reading). However, by the spring of kindergarten this “preschool advantage” has vanished for
Hispanic, low-SES and ELL children. We do find that black children who attended preschool
continue to outperform their black peers at the end of first grade in both reading and math, and that
this association is particularly pronounced for black children in full-day preschool. Specifically, by
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
16
Preschool Fade-Out
spring of the preschool year, black children who attended full-day preschool are scoring about .13 of
a standard deviation above children who did not participate in formal care.
These results largely mirror those observed in the 1998 data, where we also find positive
associations across the board for all subgroups at Kindergarten entry, with generally larger
coefficients for children enrolled in full-day care. In addition, we observe that the associations
between preschool participation and cognitive outcomes at the end of first grade are particularly
pronounced for black children. Black children who attended full-time preschool before starting
preschool were still outperforming their black peers at the end of kindergarten by .17 of a standard
deviation in both reading and math.
In Panel B of Tables 3 and 4 we show the associations between preschool participation and
behavioral outcomes in each cohort. Recall that for the full population we found that preschool
participation was associated with worse behavioral ratings both in the fall of kindergarten and in the
spring of first grade, and that these patterns were driven by participation in full-day preschool. When
we disaggregate the analyses by subgroups, the results are fairly similar, though the large standard
errors in these smaller subgroup analyses yield fewer statistically significant coefficients. In general,
the magnitude of the negative associations between preschool participation and child outcomes are
less pronounced in the more recent wave.
Discussion
In this study, we leverage newly-released data on a 2010 cohort of kindergarteners to explore
the relationship between preschool participation in the year prior to kindergarten and cognitive and
behavioral skills at school entry. We also investigate whether any relationship between preschool
participation and child outcomes remains evident by the spring of the first grade. Employing two
large datasets, collecting largely the same measures in 1998 and 2010, we are able to make
comparisons in the persistence of preschool effects across the two waves.
We build on an existing literature which leverages the 1998 data to document positive shortterm benefits from preschool participation on cognitive skills. Our current analysis replicates those
findings and shows that among the 2010 cohort, children who participated in preschool outperform
their non-participating peers on measures of cognitive skill but are rated lower on behavioral
measures. We also demonstrate that the cognitive skills of participants and non-participants
converge throughout elementary school, but that the negative behavioral associations remain.
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
17
Preschool Fade-Out
Notably the rate of convergence on cognitive measures is more pronounced in 2010. In the
1998 cohort, children who attended preschool still outperformed those who had not participated in
preschool by the spring of kindergarten and they continued to do so through the end of first grade.
In contrast, among the 2010 kindergarteners no cognitive differences were observed between
preschool participants and non-participants by the end of kindergarten (and none have reemerged
by the end of the first grade, which is the most recent data currently available).
We also explored heterogeneity in the persistence of preschool associations by the intensity
of the preschool experience or by aspects of the subsequent kindergarten experience. In general we
did not find evidence in the 2010 data that persistence patterns differ depending on the observable
measures of preschool or kindergarten explored in this study. Specifically, we found no meaningful
differences in the rate of fade-out for children who attended full rather than part-time care and also
did not find differential patterns of persistence based on children’s subsequent kindergarten
experiences including the length of the kindergarten school day, the kindergarten class size, the
exposure of kindergarten peers to preschool and a number of other quality measures. These general
patterns were mirrored in the 1998 data where we similarly documented little evidence of differences
in persistence depending on kindergarten experience. That said, in the 1998 data we do show that
the observed association between preschool participation and first grade cognitive outcomes is
somewhat stronger for children who attended full rather than half-day preschool programs.
Finally, our study explored whether there was heterogeneity in preschool persistence
patterns across child demographic characteristics, and found strong support for this hypothesis.
Specifically, we demonstrate that although for the full sample of children the associations between
preschool participation and reading and math scores have fully dissipated by the spring of
kindergarten, this is not the case for black children, particularly those who attended full-day
preschool. In fact, by the end of the first grade, black children who attended full-day preschool
outperform other black children who had no formal center-based care in the year before school by
about .13 standard deviations in both reading and math. This general pattern—of larger, positive
associations for black children in full-day care—also emerged in the 1998 data. For that cohort, we
find that black children in full-day programs outperformed their peers who had no formal centerbased care by .17 standard deviations at the end of first grade. These coefficients are nearly 2.5 times
larger than those observed for the full sample. It is worth noting that among the 1998 cohort,
children classified as low-SES showed more persistent associations between preschool participation
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
18
Preschool Fade-Out
and cognitive outcomes than did peers from higher-SES homes. However, this pattern was not
evident in the 2010 data.
Contributions of the Current Study
This study makes several notable contributions to the existing literature. It is the first in the
literature to employ very similar datasets and the same research design across two waves of
kindergarten entrants in answering questions about the persistence or convergence of preschool
effects. In this way, the study provides unique insights about whether these patterns have changed
over time, and does so over a period characterized by significant increases in policy interest and
investment in early childhood.
The quasi-experimental literature on the persistence of Head Start effects is suggestive of
increasingly more rapid fade-out for more recent cohorts of Head Start participants (Gibbs, Ludwig,
& Miller, 2013). However, the multiple studies that were compared in that analysis were focused on
distinct populations and applied different empirical approaches towards isolating effects. These
differences made it difficult to assess the extent to which fade-out patterns have truly changed over
time. Using two very similar national datasets, we find comparable associations between preschool
participation and cognitive outcomes for the more recent cohort of children as compared to a
cohort entering kindergarten twelve years prior. However, for the more recent cohort these
associations are no longer evident by the end of the kindergarten year. In contrast, among the 1998
cohort, we continue to observe statistically significant, positive relationships between preschool
participation and both reading and math outcomes through the first grade. In other words, this
study provides compelling new evidence that the phenomenon of fade-out of preschool effects may
be more pronounced for more recent cohorts of children.
A second key contribution of this study is that we examine whether characteristics of the
preschool or subsequent kindergarten experience moderate the associations between preschool
participation and subsequent child outcomes. In contrast to some earlier studies (Magnuson et al.,
2007b; Reynolds et al., 2010), we find very little evidence that specific characteristics of the
preschool or kindergarten experience are related to the persistence of preschool associations.
We do show that the negative associations between preschool participation and externalizing
behaviors and self-control are more pronounced for children in full rather than half-day preschool.
We also show that participating in full-day preschool is positively associated with longer-term
cognitive outcomes for some subgroups examined. However, we find no evidence in either cohort
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
19
Preschool Fade-Out
that the observable measures of kindergarten quality examined in this study serve to moderate the
associations between preschool participation and child outcomes. This finding does not suggest that
subsequent kindergarten experiences are unimportant for maintaining any benefits accrued from
preschool participation. Rather, it may be that the type of kindergarten experiences that may be
particularly beneficial are not easily measured within large surveys.
A final contribution of the current study is our exploration of differences across subgroups
both in the initial association between preschool participation and child outcomes, and the
persistence of these associations. By demonstrating greater persistence of cognitive benefits among
black children attending full-day preschool programs, our study adds to the research base suggesting
larger benefits for particular subgroups, particularly black children (Bassok, 2010).
Limitations
Although the current study makes several important contributions, it also suffers from a number of
limitations that are typical in studies that leverage large national datasets to make comparisons over
time. First, to the extent that our extensive set of controls does not account fully for the endogeneity
of preschool participation, intensity of preschool participation (part- or full-time care), and
subsequent kindergarten experiences, we have to be cautious in attributing any of the relationships
documented in this study as causal. Rather, we can explore associations in the data, and identify
those characteristics of children, preschool participation, and kindergarten experience associated with
the persistence of early cognitive skill advantages or behavioral disadvantages. We acknowledge
there is likely selection—on both observable and unobservable dimensions—into preschool and
subsequent kindergarten experiences. That said, the 1998 wave of the ECLS data has been leveraged
extensively in the literature to document the effects of preschool participation (Loeb et al., 2007;
Magnuson et al., 2007a, 2007b). Our estimates for the earlier wave are consistent with that existing
evidence. Further, we have no reason to believe that patterns of selections differ across waves of the
study.
A second limitation of the current work is that we are constrained somewhat by differences
in data collection between 1998 and 2010. For instance, differences across surveys with respect to
how Head Start participation was treated make it impossible for us to examine whether the
association between Head Start participation and either initial or later outcomes have changed over
time. Although we have done our best to construct very similar “preschool” groups across studies, it
is possible that differences in the wording of the Head Start items across surveys could impact our
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
20
Preschool Fade-Out
findings. Further, because the test score standardization occurs within sample, we are somewhat
cautious about the comparability of the magnitude of effects for preschool participants across the
two waves of survey data.
Finally, when exploring heterogeneity in patterns by preschool and kindergarten experiences,
we focused on a number of measures that were selected based on their demonstrated relevance in
the existing literature. However, these measures are only a small subset of the characteristics that
may be relevant. In other words, it is certainly possible that characteristics of preschool programs
and subsequent schooling experiences do moderate the relationship between preschool participation
and child outcomes, but that we are not picking up the relevant dimensions of kindergarten quality
with our measures. This is particularly true with regards to the quality of preschool experience,
where the ECLS data is particularly weak. A large body of research has demonstrated the
importance of engaged and caring child-teacher interactions in predicting gains from preschool
participation (Hamre & Pianta, 2006; Mashburn et al., 2008). The ECLS-K datasets have very
limited information about children’s preschool experiences, and none about the quality of childteacher interactions, so we were unable to explore the extent to which various measures of
preschool quality moderate the relationships we examined. Similarly, it may be that in kindergarten
classrooms with particular curriculums, pedagogical approaches, or teacher characteristics, children
who attended preschool programs are able to maintain their initial gains.
Conclusion
This project has addressed several important questions in the field of early childhood education: Do
children who attend preschool outperform their peers as they progress through elementary school?
Do these patterns depend on observable characteristics of the child, the preschool program or the
subsequent kindergarten program? Our findings demonstrate that, on average, by the end of
kindergarten children who attended preschool are no longer outperforming on measures of reading
and math relative to their peers. This finding seems to hold irrespective of whether children are
attending part- or full-time preschool, and irrespective of a number of proxies for kindergarten
quality. We do find encouraging evidence that black children, particularly those who attended fullday preschool, are still outperforming their peers through first grade. However, the bulk of our
analysis is consistent with an account of rapid “fade-out” of any preschool advantage with respect to
cognitive assessments.
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
21
Preschool Fade-Out
It is unclear to what extent either this fade out or its possible acceleration should raise
concerns. Gibbs, Ludwig, & Miller (2013) suggest that the convergence in test score outcomes
between preschool participants and their peers is not necessarily inconsistent with long-term
preschool effects. They note that patterns of fade-out have been documented in studies of early
childhood interventions that still realize long-term improvements in important outcomes. For
instance, Deming (2009) found large, long-term benefits of Head Start despite fade-out of the testscore benefits. Ludwig and Miller (2007) found that Head Start increased students’ educational
attainment, despite having no measurable effect on 8th grade test scores. In addition, Chetty et al.
(2011) found effects of high-quality kindergarten on earnings and employment despite fade-out of
effects on test scores.
These studies suggest that the lack of persistent associations between preschool and child
outcomes we observe at the end of first grade does not necessarily mean that preschool participation
does not yield important long-term outcomes. For fade-out to be problematic, it would have to
correspond to changes in the way early childhood interventions affect the realization of long-term
benefits. If preschool participation no longer impacts the unmeasured pathways through which
benefits are generated, then fade-out and the acceleration of fade-out we document could be
worrisome. Currently, we know very little about whether and how patterns of fade-out are related to
longer-term outcomes. On this question, there is much more to be learned. Similarly, more work is
needed to better understand the conditions that sustain the short-term benefits of preschool
participation, and the reasons why rates of fade-out may be hastening.
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
22
Preschool Fade-Out
References
Barnett, W. S., Carolan, M. E., Squires, J. H., & Brown, K. C. (2014). The State of Preschool 2013: State
Preschool Yearbook. New Brunswick, New Jersey: National Institute for Early Education
Research.
Bartik, T. (2013). The effectiveness of many state and local pre-K programs has been backed by
sound research. Retrieved from http://investinginkids.net/2013/07/24/the-effectivenessof-many-state-and-local-pre-k-programs-has-been-backed-by-sound-research/
Bassok, D. (2010). Do Black and Hispanic Children Benefit More From Preschool? Understanding
Differences in Preschool Effects Across Racial Groups. Child Development, 81(6), 1828–1845.
Bassok, D., & Loeb, S. (2015). Early Childhood and the Achievement Gap. In H. F. Ladd & M.
Goertz (Eds.), Handbook of Research in Education Finance and Policy (2nd ed.).
Belsky, J. (2002). Quantity Counts: Amount of Child Care and Children’s Socioemotional
Development. Journal of Developmental and Behavioral Pediatrics, 23(3), 167–170.
Bogard, K., & Takanishi, R. (2005). PK-3: An Aligned and Coordinated Approach to Education for
Children 3 to 8 Years Old. Social Policy Report. Volume 19, Number 3. Society for Research in
Child Development. Retrieved from http://eric.ed.gov/?id=ED521747
Boozer, M., & Cacciola, S. E. (2001). Inside the’Black Box’of Project STAR: Estimation of peer
effects using experimental data. Yale Economic Growth Center Discussion Paper, (832). Retrieved
from http://papers.ssrn.com/sol3/papers.cfm?abstract_id=277009
Campbell, F. A., Pungello, E. P., Burchinal, M., Kainz, K., Pan, Y., Wasik, B. H., … Ramey, C. T.
(2012). Adult outcomes as a function of an early childhood educational program: an
Abecedarian Project follow-up. Developmental Psychology, 48(4), 1033.
Cannon, J. S., Jacknowitz, A., & Painter, G. (2006). Is full better than half? Examining the
longitudinal effects of full-day kindergarten attendance. Journal of Policy Analysis and
Management, 25(2), 299–321.
Chetty, R., Friedman, J. N., Hilger, N., Saez, E., Schanzenbach, D. W., & Yagan, D. (2011). How
Does Your Kindergarten Classroom Affect Your Earnings? Evidence from Project Star. The
Quarterly Journal of Economics, 126(4), 1593–1660.
Claessens, A., Engel, M., & Curran, F. C. (2013). Academic Content, Student Learning, and the
Persistence of Preschool Effects. American Educational Research Journal, Forthcoming., 1–32.
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
23
Preschool Fade-Out
Cunha, F., & Heckman, J. J. (2010). Investing in Our Young People (Working Paper No. 16201).
National Bureau of Economic Research. Retrieved from
http://www.nber.org/papers/w16201
Currie, J., Garces, E., & Thomas, D. (2000). Longer-Term Effects of Head Start. The American
Economic Review, 999–1012.
Currie, J., & Thomas, D. (2000). School quality and the longer-term effects of Head Start. Journal of
Human Resources, 755–774.
Deming, D. (2009). Early childhood intervention and life-cycle skill development: Evidence from
Head Start. American Economic Journal: Applied Economics, 111–134.
Engel, M., Claessens, A., & Finch, M. A. (2013). Teaching Students What They Already Know? The
(Mis)Alignment Between Mathematics Instructional Content and Student Knowledge in
Kindergarten. Educational Evaluation and Policy Analysis, 35(2), 157–178.
doi:10.3102/0162373712461850
Fitzpatrick, M. D. (2008). Starting School at Four: The Effect of Universal Pre-Kindergarten on
Children’s Academic Achievement. The B. E. Journal of Economic Analysis & Policy, 8(1), 1897.
Gibbs, C. (2014). Experimental evidence on early intervention: The impact of full-day kindergarten. (Batten
School Working Paper No. 2014-04). Charlottesville, Virginia: University of Virginia.
Gibbs, C., Ludwig, J., & Miller, D. L. (2013). Does Head Start Do Any Lasting Good? In M. Bailey
& S. Danzinger (Eds.), The Legacy of the War on Poverty: A 50-Year Retrospective. New York, NY:
Russel Sage. Retrieved from http://www.nber.org/papers/w17452
Gormley, W. T., Gayer, T., Phillips, D., & Dawson, B. (2005). The effects of universal pre-k on
cognitive development. Developmental Psychology, 41(6), 872–884.
Graham, B. S. (2008). Identifying social interactions through conditional variance restrictions.
Econometrica, 76(3), 643–660.
Hamre, B., & Pianta, R. C. (2006). Student-Teacher Relationships. Retrieved from
http://doi.apa.org/?uid=2006-03571-005
Havnes, T., & Mogstad, M. (2011). No child left behind: Subsidized child care and children’s longrun outcomes. American Economic Journal: Economic Policy, 97–129.
Herry, Y., Maltais, C., & Thompson, K. (2007). Effects of a full-day preschool program on 4-yearold children. Early Childhood Research and Practice, 9(2). Retrieved from
http://ecrp.uiuc.edu/v9n2/herry.html
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
24
Preschool Fade-Out
Krueger, A. B., & Whitmore, D. M. (2001). The effect of attending a small class in the early grades
on college-test taking and middle school test results: Evidence from Project STAR. The
Economic Journal, 111(468), 1–28.
Ladd, H. F., Muschkin, C. G., & Dodge, K. A. (2014). From Birth to School: Early Childhood
Initiatives and Third-Grade Outcomes in North Carolina. Journal of Policy Analysis and
Management, 33(1), 162–187. doi:10.1002/pam.21734
Lee, V. E., Burkam, D. T., Ready, D. D., Honigman, J., & Meisels, S. J. (2006). Full-Day versus HalfDay Kindergarten: In Which Program Do Children Learn More? American Journal of
Education, 112(2), 163–208.
Lee, V. E., & Loeb, S. (1995). Where do Head Start attendees end up? One reason why preschool
effects fade out. Educational Evaluation and Policy Analysis, 17(1), 62–82.
Lipsey, M. W., Hofer, K. G., Dong, N., Farran, D. C., & Bilbrey, C. (2013). Evaluation of the
Tennessee voluntary prekindergarten program: Kindergarten and first grade follow-up
results from the randomized control design. Vanderbilt Peabody College, August. Retrieved from
https://my.vanderbilt.edu/tnprekevaluation/files/2013/10/August2013_PRI_Kand1stFollo
wup_TN-VPK_RCT_ProjectResults_FullReport1.pdf
Loeb, S., Bridges, M., Bassok, Fuller, B., & Rumberger, R. (2007). How much is too much? The
influence of preschool centers on children’s social and cognitive development. Economics of
Education Review, 26(1), 52–66.
Ludwig, J., & Miller, D. L. (2007). Does Head Start Improve Children’s Life Chances? Evidence
from a Regression Discontinuity Design. The Quarterly Journal of Economics, 122(1), 159–208.
Magnuson, K. A., Ruhm, C., & Waldfogel, J. (2007a). Does prekindergarten improve school
preparation and performance? Economics of Education Review, 26(1), 33–51.
Magnuson, K. A., Ruhm, C., & Waldfogel, J. (2007b). The persistence of preschool effects: Do
subsequent classroom experiences matter? Early Childhood Research Quarterly, 22(1), 18–38.
Mashburn, A. J., Pianta, R. C., Hamre, B. K., Downer, J. T., Barbarin, O. A., Bryant, D., … Howes,
C. (2008). Measures of Classroom Quality in Prekindergarten and Children’s Development
of Academic, Language, and Social Skills. Child Development, 79(3), 732–749.
Nye, B., Hedges, L. V., & Konstantopoulos, S. (2001). Are Effects of Small Classes Cumulative?
Evidence From a Tennessee Experiment. The Journal of Educational Research, 94(6), 336–345.
doi:10.1080/00220670109598771
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
25
Preschool Fade-Out
Peisner-Feinberg, E. S., Burchinal, M. R., Clifford, R. M., Culkin, M. L., Howes, C., Kagan, S. L., &
Yazejian, N. (2001). The Relation of Preschool Child-Care Quality to Children’s Cognitive
and Social Developmental Trajectories through Second Grade. Child Development, 72(5),
1534–1553. doi:10.2307/3654403
Puma, M., Bell, S., Cook, R., Heid, C., Broene, P., Jenkins, F., … Downer, J. (2012). Third Grade
Follow-Up to the Head Start Impact Study: Final Report. OPRE Report 2012-45. Administration for
Children & Families. US Department of Health and Human Services.
Puma, M., Bell, S., Cook, R., Heid, C., Shapiro, G., Broene, P., … Spier, E. (2010). Head Start
Impact Study. Final Report. Administration for Children & Families, 611.
Reynolds, A. J., Magnuson, K. A., & Ou, S.-R. (2010). Preschool-to-third grade programs and
practices: A review of research. Children and Youth Services Review, 32(8), 1121–1131.
Reynolds, A. J., Ou, S.-R., & Topitzes, J. W. (2004). Paths of Effects of Early Childhood
Intervention on Educational Attainment and Delinquency: A Confirmatory Analysis of the
Chicago Child-Parent Centers. Child Development, 75(5), 1299–1328. doi:10.1111/j.14678624.2004.00742.x
Reynolds, A. J., Richardson, B. A., Hayakawa, M., Lease, E. M., Warner-Richter, M., Englund, M.
M., … Sullivan, M. (2014). Association of a full-day vs part-day preschool intervention with
school readiness, attendance, and parent involvement. JAMA, 312(20), 2126–2134.
doi:10.1001/jama.2014.15376
Reynolds, A. J., Temple, J. A., White, B. A. B., Ou, S.-R., & Robertson, D. L. (2011). Age 26 Cost–
Benefit Analysis of the Child-Parent Center Early Education Program. Child Development,
82(1), 379–404.
Robin, K. B., Frede, E. C., & Barnett, W. S. (2006). Is More Better? The Effects of Full-Day vs
Half-Day Preschool on Early School Achievement. Retrieved from
http://works.bepress.com/william_barnett/17/
Schulting, A. B., Malone, P. S., & Dodge, K. A. (2005). The effect of school-based kindergarten
transition policies and practices on child academic outcomes. Developmental Psychology, 41(6),
860.
Schweinhart, L., Montie, J., Xiang, Z., Barnett, W. S., Belfield, C. R., & Nores, M. (2005). Lifetime
effects: The High/Scope Perry preschool study through age 40. Ypsilanti: High Scope Press.
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
26
Preschool Fade-Out
Shonkoff, J. P., & Phillips, D. A. (Eds.). (2000). From Neurons to Neighborhoods : The Science of Early
Childhood Development. Washington, D.C.: National Academies Press.
U.S. Department of Education. (2014, December 10). Preschool Development Grants Program,
Fact Sheet [Program Home Page; Programs]. Retrieved January 6, 2015, from
http://www2.ed.gov/programs/preschooldevelopmentgrants/index.html
Vandell, D. L., Belsky, J., Burchinal, M., Steinberg, L., Vandergrift, N., & NICHD Early Child Care
Research Network. (2010). Do Effects of Early Child Care Extend to Age 15 Years? Results
From the NICHD Study of Early Child Care and Youth Development. Child Development,
81(3), 737–756. doi:10.1111/j.1467-8624.2010.01431.x
Waldfogel, J. (2002). Child care, women’s employment, and child outcomes. Journal of Population
Economics, 15(3), 527–548. doi:10.1007/s001480100072
Walters, C. R. (forthcoming). Inputs in the Production of Early Childhood Human Capital:
Evidence from Head Start. American Economic Journal: Applied Economics.
Weiland, C., & Yoshikawa, H. (2013). Impacts of a Prekindergarten Program on Children’s
Mathematics, Language, Literacy, Executive Function, and Emotional Skills. Child
Development, 84(6), 2112–2130.
Whitehurst, G. J. “Russ,” & Armor, D. J. (2013). Obama’s Preschool Proposal is Not Based on
Sound Research. Retrieved December 17, 2014, from
http://www.brookings.edu/research/papers/2013/07/24-preschool-proposal-whitehurst
Wong, V. C., Cook, T. D., Barnett, W. S., & Jung, K. (2008). An effectiveness-based evaluation of
five state pre-kindergarten programs. Journal of Policy Analysis and Management, 27(1), 122–154.
Zill, N., Resnick, G., Kim, K., McKey, R. H., Clark, C., Pai-Samant, S., … Vaden-Kiernan, M.
(2001). Head Start FACES: Longitudinal Findings on Program Performance. Third Progress
Report.
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
27
Preschool Fade-Out
Table 1. Center-based Care Participation in 1998 and 2010 (First grade sample).
Panel A. Overall Center Participation
Center
Center
Part-time center
Full-time center
Panel C. Participation by SES
Center
Part-time center
Full-time center
Panel D. Participation by ELL status
Center
Part-time center
Full-time center
2010
0.32
0.31
0.54
Part-time center
Full-time center
Panel B. Participation by Race
1998
0.23
0.55
0.24
White
Black
Hispanic
White
Black
Hispanic
0.22
0.30
0.18
0.25
0.29
0.17
0.63
0.41
0.42
0.13
0.43
0.24
Low
Middle
High
0.13
0.23
0.28
0.24
0.11
ELL
0.41
0.25
0.16
0.54
0.31
0.79
0.52
Non ELL
0.56
0.33
0.23
0.65
0.40
0.38
0.36
0.09
0.19
Low
Middle
High
0.13
0.23
0.35
0.24
0.11
ELL
0.33
0.18
0.54
0.79
0.31
0.44
Non ELL
0.15
0.57
0.32
0.25
Note. Means of the different types of care are displayed as column percentages, so that they represent the
proportion of students from each column who experienced each type of care (e.g. Sixty-three percent of all
white students in the 1998 cohort attended center-based care vs. 42% of all black students).
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
28
Preschool Fade-Out
Table 2. Kindergarten Characteristics in 1998 and 2010 (First grade sample).
Panel A. Kindergarten Characteristics
Full-day K
1998
2010
0.21
0.21
0.55
Small class size
Peer center attendance
White
0.23
Small class size
0.59
Peer center attendance
K index
Panel C. Kindergarten Characteristics by SES
2.36
Peer center attendance
Peer center attendance
K index
Panel E. Kindergarten Characteristics by ELL
0.16
0.45
2.14
2.67
2.77
0.85
0.19
0.40
2.49
0.36
0.53
0.69
2.51
2.22
Peer center attendance
2.49
0.24
0.22
0.68
0.62
K index
Small class size
0.46
0.94
0.77
0.52
2.14
0.49
0.59
0.16
0.35
0.22
Full-day K
0.44
0.17
High
0.55
ELL
0.43
0.48
0.78
Hispanic
Middle
0.56
0.42
Black
Low
No center Center
0.20
White
High
2.22
Panel D. Kindergarten Characteristics by Center Participation
Hispanic
Middle
0.21
Small class size
Black
2.64
Low
0.67
Full-day K
Small class size
2.34
0.51
Full-day K
Full-day K
0.53
0.53
K index*
Panel B. Kindergarten Characteristics by Race
K index
0.80
0.55
0.21
0.49
0.23
2.49
2.32
0.90
0.21
0.81
0.21
2.56
2.62
Part-time Full-time No center Center
0.46
0.68
0.83
0.78
2.42
2.64
2.49
2.75
0.23
0.21
0.64
0.61
Non ELL
0.21
0.21
0.42
0.62
ELL
0.56
0.83
2.35
2.43
0.22
0.54
0.19
0.38
0.71
0.22
2.73
Part-time Full-time
0.71
0.88
2.64
2.90
0.19
0.24
0.62
0.63
Non ELL
0.80
0.22
0.55
2.66
Note. "K index" variable ranged from 0-6 with a SD of approx 1.1 that was consistent across all groups considered. Means of the
individual kindergarten characteristics are displayed as column percentages, so that they represent the proportion of students from
each column who experienced each characteristic. e.g. Fifty-one percent of all white students attended full-day K in 1998 vs. 78% of
all black students.
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
29
Preschool Fade-Out
Table 3. Main Effects of Center-based Care on Kindergarten Outcomes, 2010 Cohort.
All
Fall
Spring Spring
K
K
1st
Panel A. Cognitive Outcomes
Reading
Center
0.10*** 0.03
0.03
(0.02) (0.02) (0.03)
Part-time
0.11*** 0.07** 0.05+
(0.02) (0.03) (0.03)
Full-time
0.06*
-0.02
-0.03
(0.02) (0.03) (0.03)
Math
Center
0.09*** 0.01
0.02
(0.02) (0.02) (0.02)
Part-time
0.08*** 0.02
0.04
(0.02) (0.02) (0.03)
Full-time
0.06*
-0.01
-0.02
(0.02) (0.03) (0.03)
0.37
Approx R^2
11300
Approx N
Panel B.Behavioral Outcomes
Externalizing
Center
0.09** 0.12*** 0.09**
(0.03) (0.03) (0.03)
Part-time
0.03
0.04
0.02
(0.03) (0.03) (0.03)
Full-time
0.17*** 0.18*** 0.17***
(0.03) (0.03) (0.04)
Self Control
Center
-0.06+ -0.09** -0.11***
(0.03) (0.03) (0.03)
Part-time
0.00
-0.01
-0.04
(0.03) (0.03) (0.03)
Full-time
-0.10** -0.13***-0.20***
(0.03) (0.03) (0.04)
Approx R^2
0.13
Approx N
9250
Fall
K
Black
Spring Spring
K
1st
Fall
K
Hispanic
Spring Spring
K
1st
Fall
K
Low SES
Spring Spring
K
1st
Fall
K
ELL
Spring Spring
K
1st
0.14+
(0.08)
0.04
(0.10)
0.16*
(0.07)
0.10
(0.09)
0.04
(0.12)
0.10
(0.08)
0.14
(0.09)
0.06
(0.12)
0.13+
(0.08)
0.09*
(0.04)
0.06
(0.04)
0.14**
(0.05)
0.02
(0.05)
0.00
(0.05)
0.05
(0.06)
0.03
(0.06)
0.05
(0.06)
-0.04
(0.06)
0.10*
(0.05)
0.08
(0.07)
0.08
(0.06)
0.01
(0.05)
0.01
(0.07)
-0.02
(0.07)
0.03
(0.06)
0.02
(0.08)
-0.01
(0.08)
0.03
(0.07)
-0.06
(0.08)
0.13
(0.10)
0.01
(0.07)
-0.03
(0.09)
0.04
(0.10)
0.00
(0.07)
0.04
(0.09)
-0.09
(0.09)
-0.04
(0.10)
0.00
(0.14)
-0.11
(0.11)
-0.05
(0.10)
-0.03
(0.16)
-0.13
(0.11)
0.24*
(0.11)
0.23
(0.17)
0.20+
(0.12)
0.14*
(0.06)
0.07
(0.07)
0.17*
(0.07)
0.14*
(0.06)
0.11+
(0.06)
0.12+
(0.07)
0.05
(0.06)
0.05
(0.08)
0.04
(0.07)
0.10
(0.08)
0.04
(0.10)
0.14
(0.10)
0.18*
(0.07)
0.14
(0.09)
0.18+
(0.09)
0.06
(0.07)
-0.03
(0.09)
0.16
(0.10)
0.07
(0.09)
0.08
(0.10)
0.05
(0.11)
0.11
(0.08)
0.13
(0.10)
0.09
(0.09)
0.06
(0.09)
0.03
(0.10)
0.10
(0.11)
0.14+
(0.07)
0.10
(0.10)
0.13+
(0.07)
0.04
(0.11)
0.02
(0.15)
0.10
(0.10)
0.11
(0.08)
0.04
(0.10)
0.09
(0.07)
0.37
1350
0.05
(0.10)
0.02
(0.18)
0.14
(0.11)
0.19
1100
0.12+
(0.07)
0.10
(0.10)
0.13+
(0.07)
-0.16+
(0.10)
-0.20
(0.14)
-0.17
(0.10)
0.09+
(0.05)
0.07
(0.05)
0.10+
(0.05)
-0.12+
(0.07)
-0.04
(0.08)
-0.12+
(0.07)
0.00
(0.06)
-0.01
(0.05)
0.04
(0.06)
0.32
2700
-0.14*
(0.07)
-0.10
(0.08)
-0.10
(0.07)
0.14
2150
0.03
(0.05)
0.08
(0.06)
-0.04
(0.06)
-0.11+
(0.06)
0.00
(0.07)
-0.18*
(0.08)
0.11+
(0.05)
0.05
(0.07)
0.13+
(0.07)
0.04
(0.06)
0.01
(0.07)
0.05
(0.08)
0.27
1900
0.00
(0.06)
0.03
(0.07)
-0.06
(0.08)
-0.11 -0.19** -0.11
(0.09) (0.07) (0.07)
0.01
-0.11
-0.04
(0.10) (0.08) (0.09)
-0.17 -0.23* -0.16+
(0.11) (0.09) (0.09)
0.17
1500
0.14*
(0.07)
0.06
(0.08)
0.22*
(0.09)
-0.09
(0.10)
0.05
(0.12)
-0.27*
(0.12)
0.02
(0.08)
-0.08
(0.09)
0.12
(0.10)
0.41
1050
-0.13
(0.10)
-0.05
(0.11)
-0.20
(0.13)
0.25
850
0.04
(0.07)
0.08
(0.08)
-0.04
(0.09)
-0.11
(0.09)
0.00
(0.10)
-0.25*
(0.11)
Note. Each "Center" coefficient comes from a separate regression. "Part time" and "Full time" estimates are drawn from the same
regression for each outcome. Low SES students are defined as those in the bottom quintile of socioeconomic status. ELL students are
defined as those for whom English is not their primary language. *** p <.001, **p <.01, *p <.05, +p <.1
30
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
Preschool Fade-Out
Table 4. Main Effects of Center-based Care on Kindergarten Outcomes, 1998 Cohort.
Fall
K
All
Spring Spring
K
1st
Panel A. Cognitive Outcomes
Reading
Center
0.14***0.08***
(0.02) (0.02)
0.13***0.08***
Part-time
(0.02) (0.02)
0.14*** 0.08**
Full-time
(0.02) (0.03)
Math
0.12***0.09***
Center
(0.02) (0.02)
0.12***0.08***
Part-time
(0.02) (0.02)
0.12***0.11***
Full-time
(0.02) (0.02)
0.32
Approx R^2
12500
Approx N
0.06*
(0.03)
0.05+
(0.03)
0.07*
(0.03)
0.05*
(0.02)
0.04
(0.03)
0.07*
(0.03)
Panel B.Behavioral Outcomes
Externalizing
0.22***0.21***0.17***
Center
(0.02) (0.02) (0.03)
0.12***0.10*** 0.07*
Part-time
(0.03) (0.03) (0.03)
0.37***0.35***0.32***
Full-time
(0.03) (0.03) (0.04)
Self Control
-0.13***-0.14***-0.13***
Center
(0.03) (0.02) (0.03)
Part-time
-0.05+ -0.05+ -0.04
(0.03) (0.03) (0.03)
-0.26***-0.27***-0.25***
Full-time
(0.03) (0.03) (0.04)
0.13
Approx R^2
Approx N
11050
Fall
K
Black
Spring Spring
1st
K
0.15**
(0.05)
0.08
(0.08)
0.18**
(0.06)
0.16** 0.12
(0.06) (0.08)
0.11
0.03
(0.08) (0.10)
0.18** 0.17*
(0.06) (0.08)
0.14**
(0.04)
0.10+
(0.06)
0.16**
(0.05)
0.20***
(0.05)
0.17*
(0.07)
0.21***
(0.06)
0.32
1800
0.16*
(0.07)
0.16+
(0.09)
0.17*
(0.08)
0.30*** 0.29** 0.35***
(0.09) (0.09) (0.10)
0.12
0.08 0.24+
(0.11) (0.11) (0.13)
0.37***0.38***0.40***
(0.09) (0.10) (0.11)
-0.24**-0.31*** -0.21*
(0.08) (0.08) (0.09)
-0.17 -0.13 -0.07
(0.11) (0.10) (0.12)
-0.28**-0.40***-0.27**
(0.09) (0.09) (0.10)
0.15
1450
Fall
K
Hispanic
Spring Spring
K
1st
0.16** 0.12*
0.00
(0.05) (0.06) (0.06)
0.13*
0.09
-0.06
(0.06) (0.07) (0.07)
0.19** 0.16*
0.07
(0.07) (0.07) (0.07)
0.12*
0.07
0.02
(0.05) (0.05) (0.05)
0.10+
0.06
0.02
(0.06) (0.06) (0.06)
0.13*
0.10
0.03
(0.06) (0.06) (0.07)
0.33
1500
0.21***0.22***
(0.06) (0.06)
0.19** 0.24**
(0.07) (0.07)
0.24** 0.19*
(0.08) (0.08)
-0.09
(0.06)
-0.05
(0.07)
-0.14+
(0.08)
-0.11+
(0.06)
-0.09
(0.07)
-0.15+
(0.08)
0.14
1650
0.19**
(0.07)
0.11
(0.07)
0.30**
(0.09)
-0.17*
(0.07)
-0.16*
(0.08)
-0.21*
(0.10)
Low SES
Spring Spring
K
1st
Fall
K
0.15***
(0.04)
0.12*
(0.05)
0.18***
(0.05)
0.09+
0.12
(0.05) (0.07)
0.03
0.09
(0.06) (0.09)
0.14
0.14*
(0.07) (0.10)
0.17***
(0.04)
0.16**
(0.05)
0.18***
(0.05)
0.14** 0.16*
(0.05) (0.07)
0.11+ 0.19*
(0.06) (0.09)
0.17*
0.14
(0.07) (0.10)
0.21
1650
0.28***
(0.07)
0.26**
(0.10)
0.30**
(0.10)
0.19*
(0.07)
0.19+
(0.10)
0.20*
(0.10)
-0.14+
(0.07)
-0.06
(0.09)
-0.22*
(0.09)
-0.07
(0.07)
0.02
(0.09)
-0.16+
(0.09)
0.17
1650
0.18*
(0.08)
0.07
(0.10)
0.30**
(0.11)
-0.15+
(0.08)
-0.08
(0.11)
-0.24*
(0.10)
ELL
Fall
K
Spring Spring
K
1st
0.30***
(0.08)
0.30**
(0.09)
0.28*
(0.11)
0.23** 0.03
(0.09) (0.09)
0.26*
0.13
(0.10) (0.11)
0.18
-0.11
(0.12) (0.12)
0.14*
0.05
-0.01
(0.07) (0.08) (0.09)
0.12
0.07
0.03
(0.08) (0.09) (0.10)
0.17+
0.02
-0.08
(0.10) (0.12) (0.12)
0.52
700
0.24** 0.14
0.20*
(0.09) (0.09) (0.09)
0.26*
0.12
0.12
(0.11) (0.11) (0.11)
0.22+
0.17 0.34**
(0.12) (0.12) (0.12)
-0.24**
(0.09)
-0.23*
(0.09)
-0.27*
(0.13)
-0.18* -0.04
(0.08) (0.09)
-0.18+ -0.04
(0.10) (0.10)
-0.18 -0.05
(0.12) (0.12)
0.22
900
Note. Each "Center" coefficient comes from a separate regression. "Part time" and "Full time" estimates are drawn from the
same regression for each outcome. Low SES students are defined as those in the bottom quintile of socioeconomic status. ELL
students are defined as those for whom English is not their primary language. *** p<.001, **p<.01, *p<.05, +p<.1
31
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
Preschool Fade-Out
Table 5. Interactions Between Center Care and Kindergarten Experiences
on first grade outcomes.
Panel A. Center Care Interactions
Center
Full-day K
Center*Full-day K
Center
Small K class
Center*Small K class
Center
Peer preschool
Center*Peer preschool
Center
Index of K quality
Center*Index
Panel B. Full-time Center Care Interactions
Full-time center
Full-day K
Full-time center*Full-day K
Full-time center
Small K class
Full-time center*Small K class
Full-time center
Peer preschool
Full-time center*Peer preschool
Full-time center
Index of K quality
Full-time center*Index
Reading
-0.01
(0.05)
-0.03
(0.04)
0.05
(0.05)
0.02
(0.03)
-0.10 *
(0.04)
0.01
(0.06)
0.05
(0.04)
0.08
(0.05)
-0.05
(0.07)
0.05
(0.06)
0.00
(0.02)
-0.01
(0.02)
-0.11 +
(0.07)
-0.04
(0.05)
0.11
(0.07)
-0.01
(0.04)
-0.10 *
(0.04)
0.02
(0.07)
0.08
(0.05)
0.08
(0.05)
-0.17 +
(0.09)
0.07
(0.08)
0.00
(0.02)
-0.03
(0.03)
Math
0.06
(0.05)
0.03
(0.04)
-0.05
(0.06)
0.02
(0.03)
-0.05
(0.04)
-0.03
(0.05)
0.01
(0.05)
0.03
(0.05)
0.00
(0.07)
0.10 +
(0.06)
0.01
(0.01)
-0.03
(0.02)
0.02
(0.06)
0.03
(0.04)
-0.04
(0.07)
-0.00
(0.04)
-0.05
(0.04)
-0.04
(0.06)
0.03
(0.06)
0.02
(0.05)
-0.05
(0.09)
0.14 +
(0.08)
0.01
(0.01)
-0.05 *
(0.03)
Externalizing Self-Control
0.06
(0.06)
0.01
(0.05)
0.03
(0.06)
0.09 **
(0.03)
0.03
(0.04)
0.02
(0.06)
0.14 **
(0.05)
0.07
(0.06)
-0.10
(0.08)
0.15 *
(0.07)
0.03
(0.02)
-0.02
(0.02)
0.27
(0.09)
0.01
(0.05)
-0.10
(0.10)
0.17
(0.04)
0.04
(0.05)
0.04
(0.08)
0.23
(0.07)
0.05
(0.06)
-0.10
(0.11)
0.29
(0.10)
0.03
(0.02)
-0.04
(0.03)
**
***
**
**
+
-0.13
(0.06)
-0.08
(0.05)
0.03
(0.06)
-0.12
(0.03)
-0.07
(0.05)
0.05
(0.06)
-0.18
(0.05)
-0.10
(0.06)
0.13
(0.09)
-0.18
(0.07)
-0.03
(0.02)
0.03
(0.02)
-0.33
(0.09)
-0.08
(0.05)
0.15
(0.10)
-0.22
(0.04)
-0.07
(0.05)
0.08
(0.08)
-0.23
(0.07)
-0.09
(0.06)
0.08
(0.11)
-0.29
(0.10)
-0.03
(0.02)
0.03
*
***
**
**
***
***
***
**
(0.03)
Note. Full day center care was defined as more than 20 hours/wk. Full day kindergarten
was defined as 5 or more hours/day. *** p<.001, **p<.01, *p<.05, +p<.1
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
32
Preschool Fade-Out
Table A1. List of variables included in analysis
Variable
Gender
Race
Child height/weight in fall of kindergarten
Socioeconomic status
Region of residence
Urbanicity
Low birthweight
Premature birth
Number of places child has lived
Maternal education
Paternal education
Mother's age in fall of kindergarten
Language spoken at home
Maternal work status
Paternal work status
Parental expectations of child's academic attainment
Parent chose home location for the school
Parent spanks child
Frequency that parent spanked child last week
Family type
Number of days/week family eats breakfast/dinner together
Number of days child eats breakfast/dinner at regular time
Family has a home computer
Variable
Child watches TV on weekdays
Number of hours child watches TV on weekdays
Number of living grandparents
Number of grandparents that have close relationship with child
Number of household members younger than 18
Number of household members 18 or older
Number of books that child owns
Frequency that child reads outside of school
Frequency that child looks at picture books outside of school
How variable is coded
1 = male
Separate indicators for black, Hispanic, Asian, other
continuous
Indicators for SES quintiles 1-4
Northeast, Midwest, South
City, suburb
< 5.5 pounds
3-7 weeks, more than 7 weeks
continuous
Indicators for dropped out of HS/ graduated HS
Indicators for dropped out of HS/ graduated HS
< 25, between 25 and 35, between 35 and 45
Indicator for only English spoken at home
Indicators for part time/full time employment
Indicators for part time/full time employment
HS degree, >=2 years of college, BA, masters degree
1 = yes
1 = yes
continuous
Indicators for single parent, one biological
parent, two adopted parents
Two separate continuous variables
Two separate continuous variables
1 = yes
Reference group (if any)
Female
White
Quintile 5 (Highest SES)
West
rural
>= 5.5 pounds
< 3 weeks
Some college or higher
Some college or higher
> 45
At least one other language spoken
Mother is not in the workforce
Father is not in the workforce
< HS degree, > masters degree
Two biological parents
1 = Child does not watch TV
continuous
continuous
continuous
continuous
continuous
continuous
1 = never, 2 = once/twice a week
3 = 3-6 times/week, 4 = every day
33
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
Preschool Fade-Out
34
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia
Preschool Fade-Out
Table A1. (Continued)
Variable
Parental school involvement
Attended an open house at school
Attended a PTA meeting
Attended a parental advisory meeting
Attended a parent group
Volunteered at school
Participated in a school fundraiser
Parental stress composite
Too busy to play with child
Being parent is harder than expected
Child does things that bother me
I have to sacrifice to meet child's needs
I feel trapped as a parent
I often feel angry with child
Child is harder to care for than most
Being a parent is more work than pleasure
Parental depression composite
Bothered by things that don't usually bother you
Felt like not eating, or your appetite was poor
Felt that you could not shake off the blues
Had trouble keeping your mind on task
Felt depressed
Felt that everything you did was an effort
Felt fearful
Felt that your sleep was restless
Felt that you talked less than usual
Felt lonely
Felt sad
Felt like you could not get going
Parent-child relationship
Most of the time I feel that child likes me
Even when I'm in a bad mood, I show child a lot of love
I express affection by hugging, kissing, and holding child
How variable is coded
Reference group (if any)
1 = yes
Each item is rated on a scale of 1-4
1 = Completely true
2 = Mostly true
3 = Somewhat true
4 = Not at all true
Stress composite is the mean of these 8 items
Parents were asked to report how they have felt
about themselves and their life in the past week.
Each item is rated on a scale of 1-4
1 = Never
2 = Some of the time
3 = A moderate amount of the time
4 = Most of the time
Depression composite is the mean of these 12 items
1 = Completely true
2 = Mostly true
3 = Somewhat true
4 = Not at all true
35
EdPolicyWorks Working Paper Series No. 36. January 2015.
Available at http://curry.virginia.edu/edpolicyworks/wp
Curry School of Education | Frank Batten School of Leadership and Public Policy | University of Virginia