ArticlePDF Available

The nature and impact of early achievement skills, attention skills, and behavior problems

A slightly revised version of this paper was published as a chapter:
Greg J. Duncan and Katherine Magnuson, “The Nature and Impact of Early Achievement
Skills, Attention Skills, and Behavior Problems”, in Greg J. Duncan and Richard J.
Murnane (eds.), Whither Opportunity: Rising Inequality, Schools, and Children's Life
Chances, New York: Russell Sage, 2011, pp. 47-69.
Chapter 3
The Nature and Impact of Early Achievement Skills,
Attention Skills, and Behavior Problems
Greg J. Duncan
Katherine Magnuson
The authors are grateful to the Spencer Foundation, the Russell Sage Foundation, the Foundation
for Child Development, and the NSF-supported Center for the Analysis of Pathways from
Childhood to Adulthood (grant no. 0322356) for research support.
Executive Summary
Our chapter investigates links between young children’s skills and behaviors and their
later attainments. We begin with a conceptual framework for understanding early skills. We
argue that the skill categories of “cognitive” and “noncognitive” used by many economists are
both too simplistic and inaccurate. “Cognitive” skills mix mental acuity (IQ) with concrete
achievements such as knowing letters, beginning word sounds, and numbers. “Noncognitive”
skills encompass diverse capacities such as paying attention (an inherently cognitive task),
getting along with classmates and teachers, and maintaining good mental health. We propose and
defend the early-skill classification of achievement, attention, behavior problems and mental
health while at the same time acknowledging that these broad categories are related and can be
broken down further into more narrowly defined component parts.
Children from different social groups enter school with very different skills and
behaviors. Comparing children in the bottom and top quintiles of socioeconomic status (SES),
we show that low-SES children are 1.3 standard deviations lower than high-SES children in their
kindergarten-entry math skills, nearly two-thirds of a standard deviation below in teacher ratings
of attention skills, and one-fourth of a standard deviation worse in terms of teacher-reported
antisocial behavior. None of these gaps shrinks over the course of elementary school, and in the
case of antisocial behavior, the SES-based gap nearly doubles. More than half of the SES gaps
occurred within schools, which suggests that the very different kinds of schools attended by poor
and affluent children do not begin to account for all of the gaps.
Next, we summarize what is known about the developmental course of these capacities.
Cross-time achievement correlations tend to be higher than correlations for either attention or
behavior, but this may be due in part to the fact that achievement is measured more reliably than
attention or behavior. Interesting work on behavior problems identified upon school entry shows
that they persist for a small but significant number of children. Behavior problems that arise in
adolescence also generally fail to persist much beyond the adolescent period.
The heart of our chapter is a review of associations between early achievement, attention,
and behavior and later school achievement and such late-adolescent schooling outcomes as
dropping out and college attendance. We also consider early-adult criminal behavior as measured
by the likelihood of having been arrested.
We find that although school-entry achievement skills prove quite predictive of later
school achievement, the persistence dimension of early skills and behavior problems matters
most for later attainment and crime. Point-in-time assessments of primary school children are, at
best, weakly predictive of where children will end up in late adolescence or early adulthood.
Associations between skills and outcomes were generally stronger after age ten than before.
Using measures based on persistent problems across elementary school boosts the explanatory
power of early measures considerably. Children with either persistent math problems or behavior
problems were much less likely to graduate from high school or attend college. In the case of
early-adult crime, early antisocial behaviors were most predictive. But even when we judged
persistent early skill and behavior problems to have strong effects on our outcomes, there were
still many exceptions to the rule.
The Nature and Impact of Early Achievement Skills, Attention
Skills, and Behavior Problems
During the 1960s, the High/Scope Perry preschool intervention program provided one or two
years of high-quality part-day educational services and home visits to three- and four-year-old
low-income, low-IQ African American children in Ypsilanti, Michigan. At program entry, the
Perry children averaged 80 on an IQ test normed to a population mean of 100.1 Shortly after
these children completed the program, and around the time they entered kindergarten, their
scores had jumped to 95. For the children randomly assigned to a control group, scores increased
very little, from 79 to 84. The differential Perry advantage amounted to nearly one standard
deviation—a huge advantage. Perry children went on to get better grades, complete more
schooling, commit fewer crimes, and, through middle age, enjoy higher earnings and rely less on
social services.2
It is tempting to draw two conclusions from the Perry evidence. The first is that the skills
children develop prior to school entry can have important impacts on lifelong success. Abundant
theory and evidence from neuroscience and developmental psychology, as well as evaluations of
a number of intensive early-childhood interventions, support the contention that early skills and
behavior can indeed matter a great deal for later academic achievement and attainment.
A second possible conclusion is that boosting childhood IQ was the key reason for the
Perry program’s long-run successes. This is likely false: by third grade, the average IQs of Perry
children had fallen to 88—a statistically insignificant single IQ point higher than the third-grade
IQs of control-group children. If not IQ, then what other skill or behavior, consequential alone or
in combination with early cognitive skills, conveyed the benefits from the Perry “treatment”?
One possibility is that the Perry program improved key literacy and numeracy skills that,
independently of pure cognitive ability, can lay the foundation for future success in school and
beyond. In fact, measures of school achievement continued to show significant advantages for
the Perry children well beyond third grade, although later achievement impacts were certainly
smaller than early impacts. Early cognitive and achievement gains might have helped children to
avoid early school failure; indeed, children who attended the Perry program were also less likely
to receive special education services or to have been retained. Progressing through the early
school years without being held back or placed in special education increased the likelihood that
they would later go on to complete high school (Deming 2009).
Perhaps it was something about the Perry children’s ability to pay attention and become
more engaged with their school tasks. A few years after the Perry study ran its preschools,
Mischel, Shoda, and Rodriguez (1989) measured impulse control by observing whether four-
year-olds from affluent Californian families, when left alone with a marshmallow, could wait
long enough before eating it to earn a second marshmallow. He found that children who were
better able to control their impulses went on to get higher SAT scores, graduate from better
colleges, and have better adult outcomes. Regrettably, the Perry evaluation did not measure
children’s self-regulation skills, so it cannot answer this question.
Or perhaps Perry taught children inclined toward aggressive behavior how to get along
better with their peers and teachers. A number of longitudinal studies have found that adults who
commit crimes repeatedly were much more likely to have been aggressive as young children than
adults with no criminal records (Leschied et al. 2008). Analyses of the Perry evaluation by
Cunha and Heckman (2009) consider whether children’s misbehavior during elementary school
as measured in the Perry evaluation accounted for reductions in later crime and achievement.
They find that improvement in participants’ behavior does explain some of the program’s effects
on crime and income. However, most of the effects remain unexplained.
Our chapter sheds light on the Perry effect and other school-entry puzzles by turning to
theory as well as other empirical studies investigating links between young children’s skills and
behaviors and their later attainments.
The Nature of Early Achievement, Attention, and Behavior
Conceptual Model
We focus on three “skill” domains: achievement, attention, and behavior. Figure 3.1 presents our
theoretical model of how biology and environments interact to produce later school outcomes. It
draws from a Bronfenbrennerian perspective in which children are embedded in multiple
contexts and their development is shaped by their interactions within and across these contexts
(Bronfenbrenner and Morris 1998). As depicted in the figure, children enter school with a set of
skills determined by interactions between their own endowments (genetic and otherwise) as well
as the quality of their early experiences, including, for example, interactions in home and child
care settings. How school-entry skills develop is a vital question, has been studied extensively
(Shonkoff and Phillips 2000), and is not the focus of our chapter.
<!Fig. 3.1!>
Children’s skills at school entry facilitate the acquisition of later, more sophisticated
skills. But they also shape children’s environments, particularly interactions with teachers and
classmates, school experiences such as placement into ability groups, and interactions with
family members. These environments in turn can affect children’s learning and skill
development throughout the school years.
For example, strong letter identification skills at school entry may enhance a child’s
ability to map letters onto corresponding sounds, and thus provide a strong foundation for
developing reading skills. The presence of highly skilled children, if clustered in the same
classroom, may also enable a teacher to target and pace instruction to meet the needs of more
children with advanced skills. This instruction may lead the child to enjoy reading and to read
more during free-play time in the classroom and with family members. This in turn further builds
a child’s vocabulary, thus improving language and reading learning. Thus, by influencing both
the child and his or her social environment, early academic skills can be linked to subsequent
academic achievement because they provide the foundation for positive classroom adaptation
(Cunha et al. 2005; Entwisle, Alexander, and Olson 2007).
Negative feedback loops are also possible. A student’s early difficulties paying attention
or getting along with teachers and classmates can lead to fewer learning-related interactions with
teachers and social ostracism from classmates. Classroom disruptions can also interfere with
classmates’ opportunity to learn. In later grades, antisocial behaviors may lead to suspensions or
expulsions, with obvious detrimental consequences for student achievement. Such transactional
and recursive models of development are a staple of developmental theory (Sameroff and Fiese
A broader conception of the classroom environment box in figure 3.1 would include the
institutional practices of schoolsspecifically the ways in which children are sorted across
schools and also “tracked” within schools. For example, placement into more or less
academically challenging curricular tracks has been linked to students’ later outcomes. As the
sorting of students within schools is more central to later schooling experiences, it is addressed in
the chapter by Farkas.
Achievement, Attention, and Behavior Problems and Mental Health
Instead of “cognitive” and “noncognitive,” we find “achievement,” “attention,” and “behavior
problems and mental health” to be a productive way of categorizing the general domains of
children’s school-related functioning (table 3.1). By “achievement” we mean concrete academic
skills. “Attention” refers to the ability to control impulses and focus on tasks. “Behavior
problems and mental health” consists of two important dimensions—the ability to get along with
others and sound mental health.
<!Table 3.1!>
Notably absent from this schema are students’ own aspirations, goals, and attitudes. In
part, this omission is appropriate given our focus on younger children, self-constructs, goals, and
aspirations develop during the early school years, and as they become more differentiated and
complex, they also become more closely associated with children’s behavior and performance
(Davis-Kean et al. 2008). For this reason, a discussion of these constructs appears in the chapter
by Farkas in this volume.
Achievement Skills. “Achievement” in the preschool and middle-childhood years refers mainly to
a set of reading- and math-related skills. For preschoolers, reading-related skills encompass
identification of upper- and lowercase letters as well as decoding skills such as beginning to
associate sounds with letters at the beginning and end of words. Most early reading problems
reflect poor decoding skills and low levels of phonological and phonemic awareness, such as a
poor ability to break down words into component sounds. As children progress through
childhood, reading skills include recognizing words by sight, understanding words in context,
and making literal inferences from passages. By the end of elementary school, students are
developing reading comprehension and evaluation skills, which include identifying the main
points in a passage as well as understanding an author’s intentions and evaluating the adequacy
and logical consistency of supporting evidence. Writing skills, specifically a child’s ability to
express ideas in written form, develop in concert with reading skills.
Rudimentary math skills can be detected in children as young as six months (Posner and
Rothbart 2007). Concrete math skills begin with the ability to recognize numbers and shapes and
to compare relative sizes. Counting and sequencing skills are followed by the ability to perform
addition and subtraction tasks, as well as multiplication and division tasks. Understanding
numerical properties such as proportions, fractions, integers, and decimals also develops, as do
measurement skills and an understanding of geometry.
These pre-academic and academic skills develop as a result of learning opportunities
embedded in everyday activities and specific instruction, which is especially important for code-
related reading skills and computational mathematical skills. Achievement trajectories are
steepest in the early years of school, as children rapidly learn many new skills and improve
existing ones. Although learning continues into later school years, the rate of gaining new skills
declines over time as more focus is placed on elaborating and improving existing skills.
More general cognitive skills also play a role in skill development. For example, oral
language skills facilitate the acquisition of reading skills such as identifying letter sounds, and
they are increasingly important as children make the transition from “learning to read” to
“reading to learn.” Likewise, a strong foundation of basic number concepts such as one, two, and
three dimensions becomes increasingly important as children advance from basic computational
tasks to more complex mathematical problems that require flexible problem-solving techniques
(Baroody 2003; Hiebert and Wearne 1996).
Although many prior studies have focused on IQ as an important determinant of
scholastic skills, we do not discuss IQ per se for several reasons. Many IQ measures include
items that are related to the acquisition of basic early reading and math skills and thus overlap
with our “achievement” domain. Measures of IQ free of such “content” reflect the speed of
cognitive processing, for example, or the ability to recognize patterns. But these types of
assessments are rarely included in large studies, and although they may be influenced by
instruction, most intervention programs target achievement and behavior rather than IQ.
Attention Skills and Cognitive Self-Regulation. Self-regulation has been defined as the “processes
by which the human psyche exercises control over its functions, states, and inner processes”
(Baumeister and Vohs 2004, 1). It involves the ability to evaluate the steps and actions required
to meet a desired goal and to control behavior deliberately in order to reach that goal. Current
theory and research on young children’s self-regulation subdivides the construct in a variety of
ways, but almost all works in this area separate cognitive (cool) and emotional components (hot)
(Eisenberg, Sadovsky, and Spinrad 2005; Raver 2004; Raver et al. 2005). We too distinguish
between hot and cold self-regulation, placing cognitive self-regulation into our “attention”
category and emotional self-regulation into our “behavior problems and mental health” category.
Cognitive self-regulation is a broad construct including such overlapping subcomponents
as executive function, planning, sustaining attention, task persistence, and inhibition of impulsive
responses. We classify this collection of skills as “attention” but emphasize their diverse nature.
Research has shown that attention and impulsivity can be detected as early as age two and a half
but continue to develop until reaching relative stability between ages six and eight (Posner and
Rothbart 2000). It is widely accepted that some dimensions of executive functioning undergo
rapid development during adolescence.
Cognitive self-control can be measured by both direct assessments of particular
components and more general descriptions of children’s behaviors (especially in structured
classroom contexts).3 Parent and teacher reports of children’s cognitive self-regulation assess the
behavioral consequences of children’s self-regulatory skills. For example, items indicate the
extent to which children are able to sit still, concentrate on tasks, persist at a task despite minor
setbacks or frustrations, listen and follow directions, and work independently or, conversely,
whether they are easily distracted, overactive, or forgetful.
Attention skills and cognitive self-regulation are thought to be consequential to children’s
learning because they increase the time children are engaged and participating in academic
endeavors and increase children’s ability to solve problems. Studies have consistently found
positive associations between measures of children’s ability to control and sustain attention with
academic gains in the preschool and early elementary school years (Raver et al. 2005;
McClelland, Morrison, and Holmes 2000; Yen, Konold, and McDermott 2004; Brock et al.
Behavior Problems and Mental Health. Perhaps because these are easily identified by the
frequently used Child Behavior Checklist (CBCL; Achenbach 1991, 1992), developmentalists
often distinguish between two broad dimensions of behavior problems—externalizing and
internalizing. Externalizing behavior refers to a cluster of related behaviors including antisocial
behavior, conduct disorders, and more general aggression. Attention problems are also included
in most externalizing behavior scales, although we suggest that they should be separated from
other forms of behavior problems. Internalizing behavior refers to a similarly broad set of
constructs including anxiety and depression as well as somatic complaints and withdrawn
behavior. In terms of understanding how behavior shapes children’s schooling, greater attention
has been devoted to externalizing behavior than to internalizing behavior, likely because of its
obvious disruptive consequences in the classroom.
Although children’s behavior problems and mental health are predicted by their capacity
to regulate emotions, these constructs are not the same. Emotional regulation refers to the ability
to “modulate the experience and expression of positive and negative emotions” (Bridges,
Denham, and Ganiban, 2004, 340). It includes the ability to control anger, sadness, joy, and other
emotional reactions, which predict such behavior as aggression and internalizing problems (for
example, social withdrawal, anxiety) (Eisenberg, Sadovsky, and Spinrad 2005).4 Poor emotional
regulation is not the only reason for poor mental health or behavior problems. Indeed, children
differ in the strength of their emotional reactivity to experiences, including the underlying
physiological reactions. Children also differ along dimensions of emotional positivity and
negativity (Posner and Rothbart 2007). Furthermore, a large body of evidence points to the
importance of deviant social information processing, including hostile attribution and other
cognitive biases, as an important contributory factor to antisocial behavior (Crozier et al. 2008).
Among young children, externalizing behavior problems are assessed by asking parents
and teachers, for example, how often a child argues, fights, or throws tantrums; gets angry; acts
impulsively; and disturbs ongoing activities. Aggression refers to behaviors such as bragging,
teasing, fighting, and attacking and is closely related to antisocial behavior, which refers to
behavior that harms another person, whether by imposing physical or mental harm or by creating
property loss. Antisocial behaviors also encompass nonaggressive harmful behaviors such as
lying and cheating. In this chapter we focus on antisocial behavior, in particular, as a particularly
important dimension of externalizing behavior.
Externalizing behavior is quite common in young children. Reports of aggression and
other forms of externalizing behavior typically peak in the preschool and early school years, as
children use aggression as a way to assert control over their environment to compensate for their
own nascent communication skills. As children’s abilities to communicate, self-regulate, and
solve problems effectively increase, their aggressive and antisocial behavior typically decreases.
However, research suggests that for a small proportion of children, hostile, aggressive, and
antisocial behavior remains high throughout childhood and adolescence (Campbell, Shaw, and
Gilliom 2000; Moffitt, 1993). Boys are more likely to display these “life-course persistent”
patterns of behavior than girls.
Depressive behavior is measured by questions that ask how frequently children appear to
be in a sad or irritable mood and whether they demonstrate low self-esteem or low energy.
Anxiety captures a set of factors including children’s fears related to separation from caregivers,
obsessive/compulsive behavior, and social reticence. Social withdrawn behavior refers to a
child’s specifically social anxiety and avoidance of social interactions.
Internalizing behavior problems increase over the course of childhood. Research suggests
that anxiety may be relatively constant over time, although it takes different forms at different
ages. Depressive behaviors, however, increase over time, and do so more for girls than boys
(Bongers et al. 2003).
Children’s behavior problems are also expected to affect both individual learning and
classroom dynamics. Externalizing behaviors promote child-teacher conflict and social exclusion
(Newcomb, Bukowski, and Pattee 1993; Parker and Asher 1987), and these stressors may reduce
children’s participation in collaborative learning activities and adversely affect achievement
(Ladd, Birch, and Buhs 1999; Pianta and Stuhlman 2004). Likewise, depressive symptoms and
anxiety may also reduce children’s engagement in classroom group learning activities (Fantuzzo
et al. 2003; Fantuzzo et al. forthcoming). Evidence of this negative effect of behavior problems
on achievement, however, is mixed, with correlational evidence pointing to a detrimental effect
but more controlled models yielding no or much smaller associations. One possible explanation
is that teachers do not yet expect children to manage their emotional responses well and thus use
instructional approaches that minimize children’s need to do so independently (Brock et al.
Skills and Behaviors at School Entry and Beyond
The Early Childhood Longitudinal Study, Kindergarten Class of 1998–99 (ECLS-K) is a natural
choice for illustrating basic empirical properties of achievement, attention, and behavioral
measures owing to its large and representative national sample of kindergartners, its longitudinal
nature, and the quality of its measures. As detailed in the online appendix, the ECLS-K’s school-
entry reading measures assess skills such as identifying upper- and lowercase letters by name,
associating letters with sounds at the beginning and end of words, and recognizing common
words by sight. Its math measures reflect the ability to identify one- and two-digit numerals,
recognize geometric shapes, count up to ten objects, and recognize the next number in a
Attention and behavior problem measures are based on teacher reports. The attention and
cognitive self-regulation scale in the ECLS-K is called “approaches to learning” and includes
items that assess the child’s attentiveness, task persistence, eagerness to learn, learning
independence, flexibility, and organization. The externalizing behavior problems index rates the
frequency with which a child argues, fights, gets angry, acts impulsively, and disturbs ongoing
activities, while the internalizing behavior problem index covers anxiety, loneliness, low self-
esteem, and sadness.
Kindergarten and Cross-Time Correlations
Kindergarten correlations among the ECLS-K measures are shown in table 3.A1. At .69, reading
and math achievement have the highest correlation. But while both reading and math scores
correlate substantially with attention, the correlation between achievement and behavior
problems is much lower. Attention is moderately correlated with both achievement and behavior
problems—all four correlations are in or near the .3 to .5 range.
By fifth grade, virtually all the correlations have grown, some substantially (table 3.A1).
Most notably, correlations between the two achievement and the two behavior measures are all
above .2 (in absolute value). Although part of the increased correlations may come from better
measurement, the early school years may be a time in which children become somewhat more
differentiated into groups with higher achievement and good behavior and with lower
achievement and poor behavior.
Skill and Behavior Stability across Primary School
Although stability is the norm, some children do demonstrate both transitory fluctuations and
fundamental shifts in their achievement trajectories (Kowaleski-Jones and Duncan 1999;
Pungello et al. 1996). A look at the temporal persistence of the ECLS-K’s five achievement,
attention, and behavior measures shows a clear ranking of these correlations, with both time-
dependent math and reading test score correlations always above .6, externalizing behavior
problem and attention correlations falling to about .50 by first grade but then falling only
modestly after that, and internalizing behavior problem correlations dropping the most (tables
3.A2 to 3.A4). The pattern in the attention and behavior problem measures may reflect, in part,
the lower reliability of the internalizing behavior problem index (α = .80 in kindergarten)
compared with externalizing behavior problems (α = .90) and attention (α = .89 for the ECLS-
K’s “approaches to learning” scale).
Skill and Behavior Differences across Groups
Based on the detailed look provided in tables 3.A5 and 3.A6, figures 3.2 to 3.4 plot differences in
math scores,5 attention, and externalizing behavior problems across socioeconomic, racial/ethnic,
and gender groups in both kindergarten and fifth grade. These figures show simple differences
across groups; tables 3.A5 and 3.A6 also shows counterpart differences within classroom (that is,
adjusting for classroom fixed effects), which account for the way students are clustered within
schools and classrooms and, in the case of the attention and behavior measures, in the way
individual teachers respond to the scales.
<!Fig. 3.2!>
<!Fig. 3.3!>
<!Fig. 3.4!>
Overall, SES differences in skills and behaviors are larger than racial/ethnic differences.
In the case of math achievement, income gaps far exceed racial/ethnic and gender gaps. On
average, students in the bottom SES quintile (with average family income of about $15,500)
scored well over one standard deviation below children in the top SES quintile (average family
income of $100,000). This result mirrors those found in the Reardon chapter, which also uses
data from the ECLS-K. As shown in table 3.A5, SES gaps are roughly half as large for children
in the same schools as for children overall, suggesting that SES-based family selection into
schools accounts for some, but by no means all, of the achievement gaps.
The picture for attention and behavior problems is relatively favorable for Hispanics;
attention gaps between Hispanics and whites virtually disappear by the end of primary school,
and behavior problem differences between these two groups are very small through middle
childhood. But while achievement gaps do not increase, Hispanic fifth graders still lag far (half a
standard deviation) behind their white counterparts.
Most worrisome are the growing skill and behavior gaps between the SES groups and by
race. By fifth grade, non-Hispanic black children and children from low-SES families have
closed none of their achievement gap with children from white and more-advantaged families,
and have fallen further behind in terms of attention and behavior problems.
SES and Young Adult Outcomes
SES differences in early skills and behavior are worrisome because they may be an important
way in which SES is transmitted from parent to child. We turned to data from the children of the
National Longitudinal Survey of Youth (NLSY; details about the data are provided in the online
appendix) to examine the mediation role of early skills and behavior. Mother’s SES is measured
by mother’s highest grade of completed schooling when the child was age five or six. Outcomes
are measured around age twenty and include the probability of being arrested, completing high
school, and attending college.
Bivariate models suggest that, relative to children in the top SES quintile, children in the
bottom SES quintile have arrest rates 15 percentage points higher, high school completion rates
31 percentage points lower, and college attendance rates 40 percentage points lower (figure 3.5;
online appendix table 3.A7). Adding measures of children’s achievement and behavior at age six
explains about one-fourth of the arrest differences and one-eighth of the two sets of schooling
differences.6 Next we added the children’s average level of achievement and behavior at ages
eight, ten, and twelve. These more persistent skill and behavior measures accounted for more of
the SES differences, but in no case did they account for as much as half of them. This suggests
that mechanisms and pathways not involving early skills play an important role in the
intergenerational transmission of SES.
<!Fig. 3.5!>
School-Level Measures of Skill Distribution
As the description of achievement, attention, and behavior problem gaps suggests, low skill
levels are disproportionately concentrated among disadvantaged populations. Given the
geographic concentration of disadvantage, low-skilled children are more concentrated in schools
that serve disadvantaged children. This imparts a double disadvantage to many low-skilled
children—they have low skills and encounter classroom environments where concentrations of
achievement and behavior problems pose difficult classroom management challenges for
We examined the possible scope of problem-laden classrooms using ECLS-K data. We
defined math and attention problems as being in the most problematic 25 percent of the national
distribution on each of these measures. We tried to do the same for externalizing behavior, but
the discrete nature of the measure led us to draw the line at the eighteenth percentile of its
distribution. Taken as a whole, some 5 percent of kindergarteners exhibit problems in all three
dimensions (table 3A.8).
We then characterized schools by the percent of children qualifying for the federal free
school lunch program, the percent of racial or ethnic minority children, and population density
(urban vs. suburban). Income-based contrasts are striking, with four times as many triple-
problem children in poor (8 percent) as opposed to affluent (2 percent) schools. More generally,
the data suggest that schools with higher proportions of low-income or minority children have a
greater concentration of low math skills and significant behavior and attention problems.
Differences between urban and suburban schools are considerably smaller but still apparent.
With most peer-effect studies concentrating on the consequences of low- or high-achieving
classmates (for example, Betts and Zau 2004; Hanushek et al. 2003; Hoxby and Weingarth
2007), we know relatively little about possible tipping points surrounding the number of
multiple-problem classmates it takes for individual problems to become collective problems.
Nor do we know how the concentration of these problems affects the other half of the
sorting process—of teachers across schools. Schools serving more affluent children typically
have more economic resources and, it would appear from table 3.A8, more easily managed
classrooms. Little wonder they are able to attract and retain more highly qualified teachers than
poor schools (Phillips and Chin 2004). Even within a large urban school district, principals of
lower-achieving schools assign classroom management skills a much higher priority in looking
for new teachers than do principals of higher-achieving schools (Engel 2007).
Consequences of Skills and Behaviors for School Achievement
We turn now to the “so what?” question for early skills and behaviors: what difference do
they really make for later success in school and beyond? Here we review existing evidence
linking school-entry skills and behaviors to later school achievement and then generate new
evidence on links to early-adult school attainment and crime.
School-Entry Skills and Later Achievement
A number of experiments provide encouraging evidence that specially designed intervention
programs that target preschool children “at-risk” for school failure produce cognitive and
academic achievement gains; long-term reductions in referral for special education services,
grade retention, and dropping out; and increases in adult educational attainment (for a review,
see Blau and Currie 2006). But most of these programs had a broad curriculum designed to
enhance academic and social skills, so it is not possible to disentangle impacts of the self-
regulation, behavior, and academic components of the programs.
Another shortcoming of the experimental literature is that interventions that focus more
narrowly on just one aspect of skills or behaviors do not consider cross-domain effects.
Relatively few studies of behavioral interventions also estimate impacts on later academic
outcomes. Durlak et al.’s (forthcoming) review suggests that at least some of those do find
positive impacts on achievement.
Duncan et al. (2007) provide the most comprehensive nonexperimental assessment of the
associations between school-entry achievement, attention, and behavior and later school
achievement. Using six longitudinal data sets,7 they regressed reading and mathematics
achievement (from tests and, where available, teacher ratings) on kindergarten-entry measures of
reading and math achievement, attention, antisocial behavior, and internalizing behavior
problems. Importantly, controls for child IQ, behavior, and temperament and parent education
and income, all of which were measured prior to kindergarten entry, were included in the
regressions. To establish comparability across studies, dependent-variable measures of
achievement as well as school-entry skills and behaviors were standardized in all studies using
full-sample standard deviations. All postkindergarten reading and math achievement outcome
measures available in the six data sets were treated as dependent variables in separate
To summarize their results, they conducted a formal meta-analysis of the 236
standardized regression coefficients emerging from the individual study regressions. Average
coefficients from the regressions involving math and reading outcomes are presented in table 3.2.
A clear conclusion is that only three of the five school-entry skill categories predict subsequent
reading and math achievement: reading, math, and attention.8 Neither behavior problems nor
mental health problems were associated with later achievement, holding constant achievement as
well as child and family characteristics. Indeed, none had a standardized coefficient that
averaged more than .01 in absolute value.
<!Table 3.2!>
Not surprisingly, early reading skills were stronger predictors of later reading
achievement than later math achievement. Less expected was that early math skills (adjusting for
prior IQ in five of the six studies) were as predictive of later reading achievement as were early
reading skills. Children’s attention appeared equally important (and several dimensions of
socioemotional behaviors appeared uniformly unimportant) for reading and math achievement.9
These findings did not differ systematically by gender or family SES.
All in all, the Duncan et al. (2007) analysis provides a clear answer to one question
involving the relative role of school-entry skills and behavior: for later school achievement, early
academic skills correlate most strongly, even after adjusting for differences in the fact that early
achievers score higher on tests of cognitive ability and come from more-advantaged families. A
student’s school-entry ability to pay attention is modestly predictive of later achievement, while
early behavior problems and other dimensions of social skills and mental health problems are not
predictive, once the student’s initial levels of achievement are taken into account.10
Middle-Childhood Skills and High School Completion
It is far from clear whether early academic skills matter as much and early behaviors as little for
adolescent and early-adult school attainment as they do for middle-childhood reading and math
proficiency. Finishing high school likely requires a combination of achievement, engagement,
and perseverance. Antisocial behaviors in primary school may lead to inconsequential trips to the
principal’s office, while such behaviors in middle or high school may lead to suspension,
expulsion, or even criminal prosecution. Moreover, the far-from-perfect temporal correlations in
achievement and behaviors shown in appendix tables 3.A2 to 3.A4 mean that many children
perform and behave better and worse over time.
Magnuson et al. (2009) used the NLSY and Baltimore Beginning School Study (BSS) to
study links between middle-childhood skills and behavior problems and high school completion.
Here we reproduce and expand upon their NLSY-based results and note that the BSS data
produced very similar patterns of effects.
As detailed in the appendix, the NLSY measures reading and math achievement in its
biennial child survey. Attention/self-regulation in the NLSY is drawn from the hyperactivity
subscale of the parent-reported Behavior Problems Index (BPI). Antisocial behavior and anxiety
scales are drawn from the BPI as well.
Although the NLSY surveys children every other year, it provides concurrent
measurements on math, reading, and attention skills as well as internalizing and externalizing
behavior problems for school-age children of every age between five and fourteen. To
investigate when skills and behaviors begin to predict high school completion, we ran a series of
probit regressions that related high school completion to preschool measures of child cognitive
skills, temperament, and family background (see the online appendix for a complete list) and
middle-childhood skills and behaviors. The first regression measured these skills and behaviors
at age five, the second at age six, and so on, up to age fourteen. Regression results are presented
in online appendix table 3.A9.
In seeking to understand the role of early skills in determining later outcomes, we adopt a
regression method that includes our measures of concurrent skills as well as measures of child
and family characteristics from birth through age five. For comparative purposes we also provide
bivariate models that provide a sense of the magnitude of associations between each domain and
later outcomes. Such simple associations show uniformly significant prediction from all the
measures to later educational (and crime) outcomes, albeit larger associations in the later years
compared with the earlier years (columns 1 and 11 of online appendix tables 3.A9 and 3.A11).
This finding is not surprising, and it confirms the common observation that early skill deficits
across a range of domains are linked to later outcomes. Such bivariate associations, however,
may be simply proxying for other skills or family circumstances that are the true cause of later
outcomes. For this reason, we focus on results from regression models that hold constant not
only other important domains but also family and child characteristics.
In models with a full set of controls, math and reading skills have uniformly positive but
often statistically insignificant effects on high school completion, with neither being consistently
more predictive than the other (top panel of online appendix table 3.A9). When combined into a
single, standardized composite, however, achievement effects became uniformly significant
(figure 3.6 and bottom panel of online appendix table 3.A9).11 Standard deviation increases in
the achievement composite are generally associated with smaller increases in the probability of
high school completion before age 10 than after.
<!Fig. 3.6!>
For the attention and behavior problem measures, only the measure of antisocial behavior
is consistently predictive of high school completion (in-text figure 3.7 and online appendix table
3.A10). Once antisocial behavior is taken into account, attention skills and anxiety/depression do
not predict high school completion. As with the achievement composite, behavior problems
become more predictive around age ten. Increases of one standard deviation in externalizing
behavior problems for ages ten to fourteen are associated with reductions of five to ten
percentage points in high school completion rates.
<!Fig. 3.7!>
Persistent Problems and High School Completion
Prior research has suggested that a student’s trajectory of behavior problems may be more
important than the level of behavior problems at any single age in predicting later educational
attainment (Kokko et al. 2006). This might also be true for achievement trajectories. To test
whether the persistence of academic, attention, and behavior problems is a stronger predictor of
later attainment than early behavior, we categorized children according to their pattern of scores
during the early school years (ages six, eight, and ten in the NLSY). In light of prior empirical
work, we chose the seventy-fifth percentile to demark a “high” level of behavior problems and
the twenty-fifth percentile as the threshold for low achievement.
We then sorted children in the NLSY data into three groups—never, intermittent, and
persistent—depending on whether the child fell into the worst quarter of a given measure’s
distribution on zero, one or two, or all three measurement occasions. Bivariate associations
between high school completion and all five of the skill and behavior measures are very strong
(first column of online appendix table 3.A10), with the contrasts between the “persistent” and
“never” groups associated with drops of twenty to thirty percentage points in high school
completion rates. As with the single-year measures, regression adjustments left only the
achievement and antisocial behavior problem measures to be predictive of high school
completion (table 3.3). Persistent early math achievement and antisocial behavior problems are
associated with drops of thirteen to sixteen percentage points in high school completion rates.
Surprisingly, persistent early reading problems are not predictive of high school completion, nor
are persistent attention or anxiety problems. Extending the outcomes to college attendance
produces similar patterns, with persistent math problems associated with a 34 percentage point
drop in the probability of college attendance.
<!Table 3.3!>
We considered whether the association between both levels and patterns of achievement,
attention, and behavior problems differed across several relevant subgroups defined by SES,
race, and gender. We found some variation but little systematic differences by SES and race.
Associations did, however, differ by gender. In particular, antisocial behavior was more
predictive of schooling attainment for boys than for girls.
Although educational attainment is an important measure of young adults’ successful transition
into adulthood, it is not the only one. To broaden the scope of our study of adolescent and early-
adult outcomes, we repeated the NLSY-based analyses using reports of whether a child had ever
been arrested by age twenty. Duncan et al. (2009) show that results from NLSY parallel those for
the Beginning School Study sample and its measure of incarceration by age twenty or twenty-
one and the Infant Health and Development sample and its measure of arrest by age eighteen.
As with high school completion, we ran a series of probit regressions, all of which related
high school completion to preschool measures of child IQ, temperament, and family background
as well as middle-childhood skills and behaviors. Regression results are presented in table 3.A10.
Again, results from the bivariate models uniformly indicated that the achievement, attention, and
behavior problem domain measures all predicted later arrests. Turning to the fully controlled
models, only the antisocial behavior reports were predictive of later crime. Year-by-year patterns
are shown in figure 3.8. Coefficient sizes are generally modest (although statistically significant)
until age ten, at which point they roughly double. The sample mean is about 22 percent, so a
three percentage point coefficient amounts to about a 15 percent increase relative to the base rate,
and a six percentage point coefficient increases the base rate by about 30 percent.
<!Fig. 3.8!>
Although the individual-year effects of behavior problems from ages five to ten are only
modestly predictive of later crime, persistent early antisocial behavior is very predictive. As in
the high school completion analysis, we sorted the NLSY data into never, intermittent, and
persistent groups depending on whether the child fell into the worst quarter of a given measure’s
distribution on zero, one or two, or all three measurement occasions (table 3.A12). Children
exhibiting persistent early antisocial behavior had nearly double the chance of being arrested. As
shown in table 3.A12, this effect is somewhat larger for males than for females.12
In sum, most of the action in predicting early adult crime is within the domain of
antisocial behavior. Persistent antisocial behavior problems in primary school are quite
predictive; persistent achievement, attention, or anxiety problems are not.
We motivated our chapter with the Perry puzzle: if not cognitive skills, what other skills or
positive behaviors might the Perry preschool intervention have promoted that kept Perry children
on track in school, in good jobs, and out of jail? Our bivariate NLSY-based analyses do little to
narrow the field of important skills; virtually all of our skill and behavior problem measures have
significant correlations with the later outcomes. Holding constant family background and
concurrent skills produces a much more selective picture.
In the case of early-adult crime, our guess is that Perry reduced antisocial behavior
problems in the intervention group. Our longitudinal analyses consistently point to early
antisocial behavior problems, but not early achievement, attention, or internalizing behavior
problems, as being strong predictors of arrests and incarceration, especially among boys.
Children who persistently display such problems between ages six and ten had double the chance
(roughly 40 percent rather than 20 percent) of ever having been arrested or incarcerateda result
replicated in three data sets in Magnuson et al. (2009). The impressive explanatory power of
early antisocial behavior problems for later crime stood in marked contrast to the inability of
even persistent early reading, math, attention, or mental health problems to predict later criminal
arrest, once such behavior and family background are taken into account.13
Speculating about the early-skills antecedents behind Perry’s success in promoting school
attainment is more tenuous. Here our longitudinal analyses suggest that both early achievement
and positive behaviors help children negotiate their way through successful completion of high
school and that both may be even more important in distinguishing those who enroll in
postsecondary education.
We close with a number of observations. First, although school-entry skills proved quite
predictive of later school achievement, the persistence of early achievement and behavior
problems mattered most for later attainment and crime. Single assessments of primary school
children are, at best, relatively weakly predictive of where children will end up in late
adolescence or early adulthood. Repeating these assessments over a number of years boosts the
explanatory power of at least some of them considerably.
Second, we were somewhat surprised that early attention did not matter more than it did
for long-run outcomes. Much has been written recently about the importance of a child’s ability
to regulate attention, plan tasks, and engage in the demands of a school curriculum (Baumeister
and Vohs 2004). While our measures of attention leave much to be desired, they appear about as
reliably measured as antisocial behaviors, which proved to be predictors of later outcomes. One
possibility is that our attainment measures (high school completion and on-time college
attendance) focus on the lower end of the attainment distribution and attention skills may be
more consequential for persistence and attainment at the higher end. It may also be that attention
skills developed by the early grades matter much less than higher-level attention skills that
emerge during the transition to adolescence.
Third, we noted, but were unable to test for, how one child’s achievement or behavior
problems might prove detrimental to his or her classmates. We found that high-poverty
classrooms have four times the concentrations of academic, attention, and behavior problems as
low-poverty classrooms. Although prior research has produced mixed evidence on spillover
effects for low achievers, we know much less about the classroom implications of substantial
numbers of children with behavior problems.
Finally, none of the links between middle childhood skills and adult success appeared to
be all-determining. Associations between skills and outcomes were generally stronger after age
ten than before. And even when we judged persistent early skill problems to have strong effects
on our outcomes, there were still many exceptions to the rule.
An optimistic interpretation is that teachers and parents are somehow able to prevent
most early skill and behavior problems from translating into long-run attainment problems.
Alternatively, perhaps the course of children’s development is sufficiently variable, and
subjected to so many positive and negative shocks, that cross-time skill/attainment correlations
fall quickly to modest levels.
But low correlations do not necessarily mean that early interventions designed to boost
skills, attention, or behavior are ill considered. The appropriate policy test involves costs and
benefits rather than correlation size. High-quality, intensive interventions like the Perry
preschool program have proven their worth. Whether larger-scale early interventions can do so
remains a vital policy question.
Achenbach, Thomas. 1991. Manual for the Child Behavior Checklist/4-18 and 1991 Profile.
Burlington, Vt.: University of Vermont Department of Psychiatry.
——. 1992. Manual for the Child Behavior Checklist/2-3 and 1992 Profile. Burlington, Vt.:
University of Vermont Department of Psychiatry.
Baker, Paula C., Canada K. Keck, Frank L. Mott, and Stephen V. Quilan. 1993. NLSY Child
Handbook, Revised Edition: A Guide to the 1986–1990 NLSY Child Data. Columbus,
Ohio: Ohio State University, Center for Human Resource Research.
Baroody, Arthur J. 2003. “The Development of Adaptive Expertise and Flexibility: The
Integration of Conceptual and Procedural Knowledge.” In The Development of Arithmetic
Concepts and Skills: Constructing Adaptive Expertise Studies, edited by Arthur J.
Baroody and Ann Dowker. Mahwah, N.J.: Lawrence Erlbaum.
Baumeister, Roy, and Kathleen Vohs. 2004. Handbook of Self-Regulation: Research, Theory,
and Applications. New York: Guilford Press.
Betts, Julian R., and Andrew Zau. 2004. “Peer Groups and Academic Achievement: Panel
Evidence from Administrative Data.” Working paper. La Jolla, Calif.: University of
California, San Diego.
Blau, David, and Janet Currie. 2006. “Pre-school, Day Care, and After-school Care: Who’s
Minding the Kids?” Handbook of the Economics of Education. Amsterdam: North-
Bongers, Ilja L., Hans M. Koot, Jan van der Ende, and Frank C. Verhulst. 2003. “The Normative
Development of Child and Adolescent Problem Behavior.” Journal of Abnormal
Psychology 112(5): 179–92.
Bridges, Lisa J., Susanne A. Denham, and Jody M. Ganiban. 2004. “Definitional Issues in
Emotion Regulation Research.” Child Development 75(2): 340–45.
Brock, Laura, Sara E. Rimm-Kaufman, Lori Nathanson, and Kevin J. Grimm. 2009. “The
Contributions of Hot and Cool Executive Function to Children’s Academic Achievement,
Learning-Related Behaviors, and Engagement in Kindergarten.” Early Childhood
Research Quarterly 24(3): 337–49.
Bronfenbrenner, Urie, and Pamela A. Morris. 1998. “The Ecology of Developmental Processes.”
In Handbook of Child Psychology: Theoretical Models of Human Development. 5th ed.
Vol. 1, edited by Richard M. Lerner. New York: Wiley.
Campbell, Susan B., Daniel S. Shaw, and Miles Gilliom. 2000. “Early Externalizing Behavior
Problems: Toddlers and Preschoolers at Risk for Later Adjustment.” Development and
Psychopathology 12(3): 467–88.
Currie, Janet, and Mark Stabile. 2007. “Child Mental Health and Human Capital Accumulation:
The Case of ADHD.” Journal of Health Economics 25(6): 1094–1118.
Cunha, Flavio, and James J. Heckman. 2009. “The Economics and Psychology of Inequality and
Human Development.” Journal of the European Economic Association 7(2–3): 320–64.
Cunha, Flavio, James Heckman, Lance Lochner, and Dimitri Masterov. 2005. “Interpreting the
Evidence on Life Cycle Skill Formation.” In Handbook of the Economics of Education,
edited by Eric A. Hanushek and Finis Welch. Amsterdam: North-Holland.
Crozier, Joseph C., Kenneth A. Dodge, Reid Griffith Fontaine, Jennifer E. Lansford, John E.
Bates, Gregory S. Pettit, and Robert W. Levenson. 2008. “Social Information Processing
and Cardiac Predictors of Adolescent Antisocial Behavior.” Journal of Abnormal
Psychology 117(2): 253–67.
Davenport, Betty M. 1976. “A Comparison of the Peabody Individual Achievement Test, the
Metropolitan Achievement Test, and the Otis-Lennon Mental Ability Test.” Psychology
in the Schools 13(3): 291–97.
Davis-Kean, Pamela E., L. Rowell Huesmann, Justin Jager, W. Andrew Collins, John E. Bates,
and Jennifer Lansford. 2008. “Changes in the Relation of Beliefs and Behaviors during
Middle Childhood.” Child Development 79(5): 1257–69.
Deming, David. 2009. “Early Childhood Intervention and Life-Cycle Skill Development:
Evidence from Head Start.” American Economic Journal: Applied Economics 1(3): 111–
Diamond, Adele, W. Steven Barnett, Jessica Thomas, and Sarah Munro. 2007. “Preschool
Program Improves Cognitive Control.” Science 318(5855): 1387–88.
Duncan, Greg, Chantelle Dowsett, Amy Classens, Katherine Magnuson, Aletha Huston, Pamela
Klebanov, Linda Pagani, et al. 2007. “School Readiness and Later Achievement.
Developmental Psychology 43(6): 1428–46.
Duncan, Greg, You-geon Lee, Katherine Magnuson, and Molly Metzger. 2009. “Early School
Skills and Behaviors: Precursors to Young Adult Crime?” Paper presented at the biennial
meeting of the Society for Research in Child Development. Denver, Colo. (April 2,
Dunn, Loyd M., and Frederick C. Markwardt Jr. 1970. Peabody Individual Achievement Test
Manual. Circle Pines, Minn.: American Guidance System.
Durlak, Joseph, Roger Weissberg, Allison Dymnicki, Rebecca Taylor, and Kriston Schellinger.
Forthcoming. “The Impact of Enhancing Students’ Social and Emotional Learning: A
Meta-analysis of School-Based Universal Interventions.” Child Development.
Eisenberg, Nancy, Adiran Sadovsky, and Tracy L. Spinrad. 2005. “Associations of Emotion-
Related Regulation with Language Skills, Emotion Knowledge, and Academic
Outcomes.” New Directions for Child and Adolescent Development 2005(109): 109–18.
Engel, Miriam 2007. “What Characteristics Do Principals in the Chicago Public Schools Look
For in Teachers? A Mixed Methods Study.” Paper presented at the annual meeting of the
Association for Public Policy Analysis and Management. Washington, D.C. (November
9, 2007).
Entwisle, Doris R., Karl L. Alexander, and Linda S. Olson. 2007. “Early Schooling: The
Handicap of Being Poor and Male.” Sociology of Education 80(2): 114–38.
Fantuzzo, John, Rebecca Bulotsky, Paul McDermott, Samuel Mosca, and Megan N. Lutz. 2003.
“A Multivariate Analysis of Emotional and Behavioral Adjustment and Preschool
Educational Outcomes.” School Psychology Review 32(2): 185–203.
Fantuzzo, John, Rebecca Shearer, Douglas Frye, P. McDermott, Christine McWayne, and Stacy
Perlman. Forthcoming.Investigation of Dimensions of Social-Emotional Classroom
Behavior and School Readiness for Low-Income Urban Preschool Children.” School
Psychology Review.
Hanushek, Eric A., John F. Kain, Jacob M. Markman, and Steven G. Rivkin. 2003. “Does Peer
Ability Affect Student Achievement?Journal of Applied Econometrics 18(5): 527–44.
Heckman, James, Seong H. Moon, Rodrogo Pinto, Peter A. Savleyev, and Adam Yavitz. 2009.
“The Rate of Return to the High Scope/Perry Preschool Program.” NBER working paper
15471. Cambridge, Mass.: National Bureau of Economic Research.
Hiebert, James, and Diana Wearne. 1996. “Instruction, Understanding, and Skill in Multidigit
Addition and Subtraction.” Cognition and Instruction 14(3): 251–83.
Hoxby, Caroline M., and Gretchen Weingarth. 2007. “Taking Race Out of the Equation: School
Reassignment and the Structure of Peer Effects.” Working paper. Cambridge, Mass.:
Harvard University.
Kokko, Katja, Richard E. Tremblay, Eric LaCourse, Daniel Nagin, and Frank Vitaro. 2006.
“Trajectories of Prosocial Behavior and Physical Aggression in Middle Childhood: Links
to Adolescent School Dropout and Physical Violence.” Journal of Research on
Adolescence 16(3): 404–28.
Kowaleski-Jones, Lori, and Greg J. Duncan. 1999. “The Structure of Achievement and Behavior
across Middle Childhood.” Child Development 70(4): 930–43.
Ladd, Gary W., Sondra H. Birch, and Eric Buhs. 1999. “Children’s Social and Scholastic Lives
in Kindergarten: Related Spheres of Influence?” Child Development 70(6): 1373–1400.
Leschied, Alan, Debbie Chiodo, Elizabeth Nowicki, and Susan Rodger. 2008. “Childhood
Predictors of Adult Criminality: A Meta-analysis Drawn from the Prospective
Longitudinal Literature.” Canadian Journal of Criminology and Criminal Justice 50(4):
Magnuson, Katherine, Greg Duncan, Molly Metzger, and You-geon Lee. 2009. “Early School
Adjustment and High School Dropout.” Paper presented at the biennial meeting of the
Society for Research in Child Development. Denver, Colo. (April 2, 2009).
McClelland, Megan, Frederick J. Morrison, and Deborah L. Holmes. 2000. “Children at Risk for
Early Academic Problems: The Role of Learning-Related Social Skills.” Early Childhood
Research Quarterly 15(3): 307–29.
Mischel, Walter, Yuichi Shoda, and Monica Rodriguez. 1989. “Delay of Gratification in
Children.” Science 244(4907): 933–38.
Moffitt, Terrie E. 1993. “Adolescence-Limited and Life-Course-Persistent Antisocial Behavior:
A Developmental Taxonomy.” Psychological Review 100: 674–701.
Newcomb, Andrew F., William M. Bukowski, and Linda Pattee. 1993. “Children’s Peer
Relations: A Meta-analytic Review of Popular, Rejected, Neglected, Controversial, and
Average Sociometric Status.” Psychological Bulletin 113(1): 99–128.
Parker, Jeffrey G., and Steven R. Asher. 1987. “Peer Relations and Later Personal Adjustment:
Are Low-Accepted Children at Risk?Psychological Bulletin 86(3): 357–89.
Phillips, Meredith, and Tiffany Chin. 2004. “School Inequality: What Do We Know?” In Social
Inequality, edited by Kathryn Neckerman. New York: Russell Sage Foundation.
Pianta, Robert, and Megan Stuhlman. 2004. “Teacher-Child Relationships and Children’s
Success in the First Years of School.” School Psychology Review 33(3): 444–58.
Posner, Michael, and Mary K. Rothbart. 2000. “Developing Mechanisms of Self-Regulation.”
Development and Psychopathology 12(3): 427–41.
——. 2007. Educating the Human Mind. Washington, D.C.: American Psychological
Pungello, Elizabeth P., Janis B. Kupersmidt, Margaret R. Burchinal, and Charlotte Patterson.
1996. “Environmental Risk Factors and Children’s Achievement from Middle Childhood
to Adolescence.” Developmental Psychology 32(4): 755–67.
Raver, C. Cybele. 2004. “Placing Emotional Self-Regulation in Sociocultural and
Socioeconomic Contexts.” Child Development 75(2): 346–53.
Raver, C. Cybele, Radiah Smith-Donald, Tiffany Hayes, and Stephanie M. Jones. 2005. “Self-
Regulation across Differing Risk and Sociocultural Contexts: Preliminary Findings from
the Chicago School Readiness Project.” Paper presented at the biennial meeting of the
Society for Research in Child Development. Atlanta, Ga. (April 8, 2005).
Sameroff, Arnold J., and Barbara H. Fiese. 2000. “Transactional Regulation: The Developmental
Ecology of Early Intervention.” In Handbook of Early Childhood Intervention, edited by
Jack P. Shonkoff and Samuel J. Meisels. 2nd ed. New York: Cambridge University Press.
Schweinhart, Lawrence, Jeanne Monte, Zongping Xiang, W. Steven Barnett, Clive R. Belfield,
and Milagros Nores. 2005. Lifetime Effects: The High/Scope Perry Preschool Study
through Age 40. Ypsilanti, Mich.: High/Scope Press.
Shonkoff, Jack, and Deborah Phillips, eds. 2000. From Neurons to Neighborhoods: The Science
of Early Childhood Development. Washington, D.C.: National Academy Press.
Wikoff, Richard L. 1978. “Correlational and Factor Analysis of the Peabody Individual
Achievement Test and the WISC-R.” Journal of Consulting and Clinical Psychology
46(2): 322–25.
Yen, Cherng-Jyh, Timothy R. Konold, and Paul A. McDermott. 2004. “Does Learning Behavior
Augment Cognitive Ability as an Indicator of Academic Achievement?” Journal of
School Psychology 42(2): 157–69.
Table 3.1: A Taxonomy of Skills and Behaviors
Achievement Attention Problem
Mental health
Concrete academic
Ability to control
impulses and focus
on tasks
Ability to get along
with others
Sound mental health
test areas or
Knowing letters and
numbers; beginning
word sounds, word
Can’t sit still; can’t
concentrate; score
from a computer
test of impulse
Cheats or tells lies,
bullies, is
disobedient at
Is sad, depressed,
used index
IRT (in ECLS-K) or
composite reading
and math scores
Approaches to
Learning index (in
ECLSK) and
Attention Problems
behavior problems
(in ECLS-K and
Internalizing behavior
problems (in ECLSK
and NLSY)
Table 3.2: Effect sizes of School-entry Skills and
Behaviors on Later Achievement; Meta-analysis of 236
Grades 1 to 8:
School-entry: Math achievement Reading achievement
Reading .09* .24*
Math .41* .26*
Attention .10* .08*
(- expected)
.01 ns .01 ns
(- expected)
.01 ns -.01 ns
* p<.05; n=236 estimated coefficients; Source: Duncan et al. (2007). Meta-analytic estimates
control for time to test, test/teacher outcome and study fixed effects; coefficients are weighted by
inverse of their variances.
Figure 3.1: Skills, behaviors and attainment across childhood
Prenatal and
attention and
Home and
child care
Classroom and
school and
labor market
Child’s K-12
attention and
Contributes to
Prenatal to school entry K-12 Adult
Focal paths of influence Other paths of influence
Gap in standard deviation units
Figure 3.2: Math gaps in kindergarten and fifth
First grade
Fifth grade
Gap in standard deviation units
Figure 3.3: Attention/engagement gaps in
kindergarten and fifth grade
First grade
Fifth grade
Gap in standard deviation units
Figure 3.4: Anti-social behavior differences in
kindergarten and fifth grade
First grade
Fifth grade
Figure 3.5: Accounting for the association between
bottom and top SES quintiles in early-adult outcomes
HS Completion
Attend college
No adj.
Adj. for age 6 skills and
Adj. for age 8-12 skills and
Figure 3.6: Effect of a 1 sd Increase in Composite
Achievement at Various Ages on the Probability of
High School Graduation, Full Controls
Source: NLSY
Figure 3.7: Effect of a 1 sd Increase in Anti-social
Behavior at Various Ages on the Probability of
High School Graduation, Full Controls
Source: NLSY
Figure 3.8: Effect of a 1 sd Increase in Anti-social
Behavior at Various Ages on the Probability of Ever
Arrested, Full Controls
Age 6,
8 & 10
Source: NLSY
Appendix on ECLS-K
The Early Childhood Longitudinal Study-Kindergarten Cohort (ECLS-K) has followed a
nationally representative sample of 21,260 children who were in kindergarten in the 1998-99
school year. The ECLS-K uses a multistage probability design. The primary sampling units were
counties or groups of counties. The second sampling stage was public and private schools with
kindergartens and the third stage sampled children of kindergarten age within each school. On
average at baseline, there were six children per classroom. The study thus far has released six
waves of data: fall of kindergarten and spring of kindergarten, first, third, fifth, and eighth
grades. Data were collected from multiple sources, including direct cognitive assessments of
children, interviews with parents and surveys of teachers and school administrators
Achievement. Achievement tests were administered in all study waves. The battery of
achievement tests given in kindergarten covered language and literacy as well as early math
skills. The children pointed to answers or gave verbal responses and were not asked to write or
explain their reasoning. The tests were administered using a computer-assisted interviewing
methodology. Not all children were given the same items. A set of “routing” items were used to
assess whether children should subsequently receive more or less difficult items. For this reason,
the cognitive assessment scores provided in the data are item response theory (IRT) scores. We
reports results of analyses using standardized values of these latent ability scores. Reliabilities
reported for the overall IRT scores in reading and mathematics are over .9.
In the fall of kindergarten the reading assessment evaluated children’s ability to identify
upper- and lower-case letters of the alphabet by name, associate letters with sounds at the
beginning and end of words, and recognize common words by sight. The math skills measured
include the ability to identify one and two digit numerals, recognize geometric shapes, count up
to ten objects and recognize the next number in a sequence.
In fifth grade, children were again assessed on their mathematics and reading skills.
These fifth grade assessments required students to complete workbooks and open-ended
mathematics problems. Reading passages and questions were provided to children so that they
could reference the passages when answering questions. However, all questions were read to the
students in both reading and math. In math, all answer choices were read to the students; in
reading, the students read the answer options.
The fifth grade mathematics assessment included items tapping the following areas:
simple multiplication and division and recognizing complex number patterns; demonstrating an
understanding of place value in integers to hundreds place; using knowledge of measurement and
rate to solve word problems; solving problems using fractions; and solving word problems
involving area and volume. The fifth grade reading assessment included the following skill areas:
making literal inferences, extrapolation, understanding homonyms, and evaluation. Skills
measured exclusively in fifth grade tested students ability to evaluate nonfiction.
The ECLS-K also asked teachers to complete academic rating scales (ARS) on student
reading and mathematics achievement in all survey waves. Teacher’s rated children’s proficiency
in particular skills on a scale that ranges from “not yet (1)” to “proficient (5).” In kindergarten,
the reading scale combined ratings of student’s speaking, listening, early reading, writing, and
computer literacy. The kindergarten math assessment asked about student’s proficiency with
five skills: number concepts, solving number problems, using math strategies, data analysis
(graphing), and measurement.
In fifth grade, teacher ratings of proficiency in expressing ideas, use of strategies to gain
information, reading on grade level, and writing were combined to measure reading skills. In
mathematics, teachers’ rating of student’s understanding of number concepts (place value,
fractions, and estimation), measurement, operations, geometry, application of mathematical
strategies, and beginning algebraic thinking were combined.1 At all time points, these measures
had high levels of reliability (internal consistency).
Attention and Behavior Problems. Measures of children’s attention and problem behavior
were constructed from teacher responses to self-administered questionnaires. The responses
categories for all items range 1 “never” to 4 “very often”.
The ELCS-K’s “Approaches to Learning” scale, which we use as the measure of attention
skills, includes six items that measure the child’s attentiveness, task persistence, eagerness to
learn, learning independence, flexibility and organization. This measure has a reliability of .89
in the fall of kindergarten.
The teacher-reported measure of externalizing problem behaviors consists of five items
that rate the frequency with which a child argues, fights, gets angry, acts impulsively, and
disturbs ongoing activities. The four items that make up the measure of internalizing behaviors
ask about the apparent presence of anxiety, loneliness, low self-esteem, and sadness. The
reliabilities for externalizing and internalizing problem behaviors are .90 and .80, respectively.
SES. The ECLS-K measured family SES by a combination of parents’ education and
occupation prestige, as well as household income. Each of the five measures were standardized
to have a mean of 0 and standard deviation of 1. For families in which two parents were present,
the composite SES variable was constructed by averaging of five measures (two measures of
parental education and occupational prestige and one measure of household income). In cases
where only one parent is present, an average of three measures was constructed (parent’s
education, occupational.
Missing Data. Although baseline data were collected from over 21,000 children, missing
data reduced our analysis samples to approximately 17,600 in kindergarten fall and 11,265
children in the spring of fifth grade. Some of the missing data are deliberate, since the ECLS-K
study randomly sampled half of children who changed schools and compensated for the losses
with adjustments to the sampling weights. We use pair-wise deletion in calculating the
correlations in appendix tables 1-4. All analyses use appropriate weights to account for non-
response and attrition.
Appendix on NLSY
The National Longitudinal Survey of Youth is a multi-stage stratified random sample of
12,686 individuals aged fourteen to twenty-one in 1979. Black, Hispanic, and low-income youth
were over-represented in the sample. Annual (through 1994) and biennial (between 1994 and
1 Reading ARS scores are available for the full sample, but only half of the teachers were asked to rate students in
2000) interviews with sample members, and very low cumulative attrition in the study,
contribute to the quality of the study’s data.
Beginning in 1986, the children born to NLSY female participants were tracked through
biennial mother interview supplements and direct child assessments. Given the nature of the
sample, it is important to note that early cohorts of the child sample were born disproportionately
to young mothers. Our target sample consists of 3,893 children who were age 5 or 6 in 1986
(n=921), 1988 (n=1,160), 1990 (n=951) or 1992 (n=861). These children were ages 19 or 20 in
2000, 2002, 2004, and 2006 respectively. With its biennial measurement interval, the NLSY
yields two independent samples of children (i.e., those observed at approximately 5, 7, 9, etc. and
those observed at approximately 6, 8, 10, etc.).
Dependent variables. In our analyses, we use both measures of educational attainment
and criminal activities as outcomes. Our primary measure of educational attainment is a
dichotomous indicator of whether a child completed high school at age 19 or 20. We characterize
students who are still enrolled in regular school at this age as having completed high school. We
make this exception for students who because of the timing of the interview may be a few
months shy of graduating. The rate of high school completion is between 77-79%. For the
NLSY’s three oldest cohorts, we used data collected at age 20 or 21 to measure whether the
participant has ever attended college. Since it is taken early in adulthood, this is a dichotomous
indicator of “on time” college attendance, and available for only three of the four cohorts for
which we have high school completion data. About 45-48% of the sample had attended college
by this age.
To measure criminal activity we use a self-report indicator, taken at age 19 or 20, of
whether the youth had ever been arrested for a crime. Some 22-24% of the NLSY sample
reported that they had been arrested.
Key predictors. We use as key independent variables the assessments of academic skills,
specifically reading and math achievement, as well as three dimensions of behavior – inattention
and two aspects of problems behavior anxiety/depression and antisocial behavior. These are
measured every two years in the NLSY data (ages 5/6, 7/8, 9/10, 11/12).
Reading and math achievement. Children’s early academic skills are measured by
standardized Peabody Individual Achievement Tests (PIAT, reading recognition and math). For
the purposes of analysis, scores are standardized to have a mean of 0 and standard deviation of 1
(based on the full NLSY sample distribution).
Interviewers verbally administered the PIATs. Children were first given an age
appropriate item, and a basal score was established when a child answered five consecutive
questions correctly. Once a basal was established, interviewers continued to ask the child
questions until the child answered 5 out of 7 consecutive items incorrectly. Subtracting the
number of incorrect scores between the basal and the ceiling score from the ceiling score
produced a raw test score.
The reading recognition test consists of 84 items that measure word recognition and
pronunciation ability. It tests children’s skills at matching letters, naming names, and reading
single words out loud. Dunn and Markwardt (1970) reported the one-month temporal reliability
of a national sample, and the test-retest correlations ranged from a low of .81 for kindergarteners
to a high of .94 for third grade students. Overall the test had an average temporal reliability of
.89. Studies of the tests concurrent validity find that the test was moderately correlated with
other tests of intelligence (e.g.,Wechsler Intelligence Scale for Children-Revised) and reading
vocabulary (e.g., Metropolitan Achievement Test) (Davenport, 1976; Wikoff, 1978).
The PIAT math subscale consists of 84 multiple-choice items designed to measure
mathematic concepts taught in mainstream classrooms. The problems were designed so that
children are required to apply math concepts to questions rather than conduct increasingly
complicated computations. The test starts with basic skills such as number recognition and
counting. The test increases in difficulty to problems involving division, multiplication, and
fractions. The most difficult questions involve advanced concepts from algebra and geometry.
Dunn and Markwardt (1970) reported one-month test-retest reliabilities from a national sample.
The reliabilities ranged from a low of .52 for kindergarteners to a high of .84 for high school
seniors. On average the test-retest reliability was .74. Studies of the PIAT math test’s concurrent
validity found that the test correlated moderately with other tests of intelligence and math
achievement (Davenport, 1976; Wikoff, 1978). The PIAT reading recognition and math test
scores are highly correlated (r ranges from .36 at age 13 to .60 at age 8/9).
Antisocial behavior, inattention, and anxiety/depression. In the NLSY, behavior
problems were assessed by mothers’ responses to 28 items that asked how true statements were
about a child’s behavior during the past 3 months. These questions were created specifically for
the NLSY, and consist of items derived from the Achenbach Behavior Problems Checklist as
well as other established measures (Baker et al., 1993). The single item questions were recoded
so that a response of “not true” corresponded to a score of 0, and “sometimes true” and “often”
corresponded to a score of 1.
Six subscales were created by the NLSY staff based on a factor analysis of the items. The
process for creating these subscales and the reliability of each is reported in Baker et al. (1993).
Three of the 6 behavior problem subscales are used in this study—attention problems
(hyperactivity), antisocial, and depression/anxiety. However, for the purposes of the analyses,
the raw scores are translated into standardized scores with a mean of 0, and standard deviation of
The attention problem scale is comprised of 5 items that ask about the following child
behaviors: being restless and overactive, having difficulty concentrating or paying attention,
being easily confused or in a fog, and having trouble with obsessions. The NLSY reports that
this subscale has adequate reliability (alpha of .69).
The antisocial subscale is created from 6 items that measure whether the child cheats or
tells lies, bullies or is cruel to others, does not feel sorry after misbehaving, breaks things
deliberately, is disobedient at school, and has trouble getting along with teachers. The anti-social
subscale has adequate reliability (alpha of .67).
The anxious/depressed scale consists of 5 items that indicate how often the child: has
sudden changes in mood or feeling, feels or complains that no one loves him/her, is too fearful or
anxious, feels worthless or inferior, and is unhappy, sad or depressed. The reliability of this scale
is also adequate (alpha of .65). The attention and antisocial subscales are highly correlated, with
correlations in the .45 -.55 range.2
2 The antisocial and inattention/hyperactivity scale are both part of the larger externalizing scale created by NLSY
staff. When we use the externalizing measure in analyses results parallel those found for the antisocial measure.
The anxiety/depression scale is part of the larger internalizing scale.
Covariates. An important strength of the NLSY is the depth and range of longitudinal
information collected about families. We take advantage of these data to construct a
comprehensive set of covariates that capture potentially important confounds that may be
correlated both with early skills and behavior as well as later attainment and crime.
Maternal and interviewer reports of two relevant dimensions of children’s temperament –
sociability and compliance – are drawn from the children’s age 3 or 4 interviews.3 The Peabody
Picture Vocabulary Test- Revised (PPVT) is used to measure children’s early receptive
vocabulary at age 3/4. The PPVT consists of 175 vocabulary items which increase in difficulty.
Nationally standardized scores are used in our analyses.
Data on children’s family environments were coded to correspond to two intervals—
between birth and age 5 and at age 5/6. Measures available at both times include: family income,
family structure, and urban residence. Some information was only measured when children were
age 5 or 6 including children’s HOME environment and two measures of family structure
(blended family and cohabitation). The highest grade a mother completed when the child was
age 5/6 is also used as a control.
The NLSY measures an array of child and mother background characteristics, which are
used as covariates in analyses. These variables include, for example, measures of the child’s race
(Black, Hispanic, or non-Hispanic white) and mothers’ percentile scores on the Armed Forces
Qualifying Test (AFQT, a measure of mothers’ academic aptitude assessed in 1980). In addition,
several variables that measure mothers’ risk-taking behaviors (drug and alcohol use) and her
adolescent experiences are also included as covariates.4
Missing data. The longitudinal nature of data collection results in missing data. In the
NLSY, between a quarter and a third of a particular age cohort of children is missing information
on key outcome variable (ever arrested). Missing data on key predictors (achievement and
behavior problems) is quite low during the early school years, with no more than 10% missing
data on achievement or behavior at ages 5 or 6. Yet, as expected rates of missing data increase
over time so that by age 13, about 30% of the sample has missing data on the predictor variables.
We handle this missing data by using multiple imputation techniques to create and analyze five
datasets in STATA. However, our estimation results are similar if we use only cases with
3 The compliance measure was created by summing maternal ratings of the frequency of children’s behavior on a
five-point scale from almost never (1) to almost always (5). Taken together, the seven items capture how well the
child follows directions. For example, questions include how often “the child obeys when told to go to bed” and
“turns off the TV when asked.” This measure has adequate reliability, with NLSY reporting the alpha of .59 for
children of all ages (Baker et al., 1993). Summing 3 interviewer ratings of the child’s cooperation during the
assessment created the sociability scale. Children were rated on a scale of poor (1) to excellent (5). Items include,
for example, the observer’s rating of how cooperative the child was in completing the assessment and of the child’s
attitude toward being tested. This measure has a high reliability; the NLSY reports an alpha of .93 (Baker et al.,
1993). Children who were age 5 or 6 in 1986 do not have early childhood measures of PPVT or temperament
because the maternal and child interview was not conducted at an earlier age for these children. In addition, NLSY’s
restriction of the measurement of sociability to children over age 4 in 1990, resulted in a large number of missing
data on this measure for children in cohort 4 that were age 3 in 1990. These data are imputed for children with
missing observations.
4 Currie and Stabile’s (2007) analysis takes advantage of the fact that the NLSY provides observations on siblings
by estimating fixed-effect sibling models. They find very similar coefficients on early attention and anti-social
behavior in their models of school enrollment. Given our lengthy time period between early-grade measurement of
skills and behavior and eventual attainment, sibling models are not possible for our analyses.
complete data. This approach assumes that data were missing at random (conditional on
observed characteristics).
Appendix Table 3.A1: Bivariate Correlations Among Achievement,
Attention and Behavior in Kindergarten and Fifth Grade, ECLS-K
1. Reading
2. Math
3. Approaches to Learning
4. Externalizing Behavior
5. Internalizing Behavior
Note. Correlations below the diagonal are for kindergarten and above
the diagonal are for fifth grade. **p<.01.
Appendix Table 3.A2: Bivariate Correlations Among Math and Reading
Scores, Kindergarten through Fifth Grade, ECLS-K
1. Kindergarten-Fall
2. Kindergarten-Spring
3. First Grade-Spring
4. Third Grade-Spring
5. Fifth Grade-Spring
Note. Correlations below the diagonal are for reading and above the
are for math. **p<.01.
Appendix Table 3.A3: Bivariate Correlations Among Externalizing and
Internalizing Behavior Scores, Kindergarten through Fifth Grade, ECLS-K
1. Kindergarten-Fall
2. Kindergarten-Spring
3. First Grade-Spring
4. Third Grade-Spring
5. Fifth Grade-Spring
Note. Correlations below the diagonal are for externalizing behavior and above
the diagonal are for internalizing behavior. **p<.01.
Appendix Table 3.A4: Bivariate Correlations Among Approaches to Learning
Scores, Kindergarten through Fifth Grade, ECLS-K
1. Kindergarten-Fall
2. Kindergarten-Spring
3. First Grade-Spring
4. Third Grade-Spring
5. Fifth Grade-Spring
Note. **p<.01.
Appendix Table 3.A5: Gaps in Children's Academic and Behavior Skills in the Fall of Kindergarten, ECLS-K
Approaches to
Lack of
Lack of
SES: 1st quintile/5th
SES: 1st quintile/3rd
SES: 3rd quintile/5th
Note: All positive numbers represent gaps in reference to the advantaged group indicated on the right hand side of the first column (e.g., girls, on
average, score 0.17sd higher than boys in reading). Negative numbers indicate that the left hand group has better scores, on average.
aFor both externalizing and internalizing behaviors, a positive gap indicates better behavior (i.e., less externalizing and internalizing) for the advantaged
Appendix Table 3.A6: Gaps in Children's Academic and Behavior Skills in the Spring of 5th Grade, ECLS-K
Approaches to
SES: 1st quintile/5th quintile
SES: 1st quintile/3rd quintile
SES: 3rd quintile/5th quintile
Note: In this table, all positive numbers represent gaps in reference to the advantaged group indicated on the right hand side of the first column (e.g., girls, on
average, score 0.13 sd higher than boys in reading). Negative numbers indicate that the left hand group has better scores, on average. aFor both externalizing
and internalizing behaviors, a positive gap indicates better behavior (i.e., less externalizing and internalizing) for the advantaged group.
Appendix Table 3.A7. Summary of Probit
Regressions of Ever Arrested, HS completion, and Attending College on Patterns of Childhood Antisocial Behavior,
NLSY, Ages 8, 10, and 12
Ever Arrested
HS Completion
Attending College
(1) (2) (3) (1) (2) (3) (1) (2) (3)
SES (ref: lowest quintile)
SES quintile 2 .003 .005 .007 .058 .050 .042 .085† .075 .060
(.040) (.040) (.041) (.037) (.037) (.037) (.050) (.052) (.054)
SES quintile 3 -.075* -.058 -.049 .153*** .126*** .106** .210*** .164** .131*
(.034) (.036) (.038) (.031) (.034) (.034) (.048) (.052) (.053)
SES quintile 4 -.125*** -.098** -.080* .235*** .198*** .170*** .314*** .252*** .215***
(.035) (.038) (.040) (.031) (.034) (.035) (.041) (.046) (.048)
SES quintile 5 -.150*** -.112** -.090* .312*** .268*** .236*** .398*** .337*** .285***
(.033) (.037) (.039) (.025) (.028) (.032) (.036) (.042) (.046)
At age 6,
Antisocial Problem .054*** .011 -.040* .001 -.044† .000
(.016) (.018) (.017) (.020) (.024) (.026)
Attention Problem .015 .014 -.026 -.001 -.032 .019
(.017) (.018) (.018) (.019) (.028) (.030)
Anxious Problem -.006 -.003 -.006 -.012 -.019 -.033
(.014) (.015) (.018) (.020) (.020) (.024)
Achievement Composite -.023 -.028 .067*** .022 .125*** .028
(.016) (.019) (.017) (.021) (.022) (.030)
Age 8-10-12
Antisocial Problems
Average age 8-10-12
.124*** -.095*** -.104***
(.023) (.024) (.030)
Attention Problems
Average age 8-10-12
(.023) (.028) (.041)
Anxious Problems
Average age 8-10-12
(.020) (.026) (.041)
Achievement Composite
Average age 8-10-12
(.020) (.024) (.036)
Family & Child Characteristics
Notes: *** p<.001; ** p<.01; * p<.05; † p<.1
Probit model coefficients and standard errors are "marginal effects" -- percentage point changes in the probability of ever being arrested associated with unit
changes in the given independent variable.
Column 1 coefficients represent simple bivariate relationships between ses composite and dependent variable.
Column 2 adds antisocial, attention, anxiety, and achievement composite at age 6.
Column 3 adds averages of each measure at age 8, 10, and 12.
Appendix Table 3.A8: School-level Concentrations of Kindergarten Achievement, Attention and Behavior
School Characteristics
Percent of children with …
Free Lunch
Free Lunch
Population ≥
50% Minority
Low math skills
Significant attention problems
Significant behavior problems
All three problems
Percent of full sample
Notes:“Low math skills” are scoring in the bottom 25% of the math IRT distribution.
“Significant attention problems” are scoring in the bottom 25% of the attention scale
“Significant behavior problems” are scoring in the top 18% of the externalizing behavior problem scale
Appendix Table 3.A9. Summary of Results from Probit Regressions of High School Completion on Achievement and Behavior Problems across Middle Childhood, NLSY
Age 5
Age 6
Age 7
Age 8
Age 9
Age 10
Age 11
Age 12
Age 13
Age 14
Age 14
Antisocial -.063*** -.030* -.037* -.021 -.048*** -.039** -.058*** -.053*** -.061*** -.080*** -.158*** -.104***
(.011) (.014) (.016) (.014) (.015) (.015) (.016) (.015) (.016) (.016) (.016) (.024)
Inattention -.050*** .005 -.001 .006 -.013 .017 -.007 -.005 -.037† .014 -.111*** -.007
(.011) (.014) (.016) (.017) (.019) (.015) (.017) (.017) (.020) (.019) (.013) (.017)
Anxious -.028* .004 -.003 -.011 .003 -.014 .014 .009 .026 -.017 -.092*** -.001
(.011) (.014) (.017) (.015) (.015) (.015) (.017) (.015) (.017) (.017) (.013) (.021)
Reading .090*** .027* .032† .046*** .065*** -.001 .032 .029* .047† .017 .036** .044†
(.011) (.014) (.017) (.014) (.018) (.014) (.021) (.014) (.026) (.021) (.012) (.023)
Math .080*** .019 .011 .026† -.005 .056*** .056** .033* .030 .056*** .118*** .046*
(.012) (.015) (.019) (.015) (.019) (.015) (.019) (.015) (.018) (.015) (.014) (.019)
Antisocial -.063*** -.030* -.037* -.021 -.049*** -.039* -.058*** -.053*** -.061*** -.080*** -.158*** -.104***
(.011) (.014) (.016) (.014) (.015) (.015) (.016) (.016) (.016) (.016) (.016) (.024)
Inattention -.050*** .005 -.001 .006 -.014 .017 -.007 -.005 -.037† .015 -.111*** -.007
(.011) (.014) (.016) (.017) (.019) (.015) (.017) (.017) (.020) (.019) (.013) (.017)
Anxious -.028* .004 -.003 -.011 .005 -.013 .015 .008 .026 -.018 -.092*** -.001
(.011) (.014) (.017) (.015) (.015) (.015) (.017) (.015) (.017) (.017) (.013) (.021)
Composite .114*** .047** .042** .072*** .061*** .054*** .088*** .063*** .076*** .080*** .108*** .091***
(.013) (.016) (.016) (.017) (.018) (.015) (.019) (.017) (.023) (.017) (.016) (.026)
Family &
Child Char.
no yes yes yes yes yes yes yes yes yes no yes
Sample Size 2005 2005 1888 1764 1832 1774 1756 1828 1658 1803 1667 1667
Notes: *** p<.001; **p<.01; *p<.05; †p<.1
Probit model coefficients and standard errors are "marginal effects" -- percentage point changes in the probability of the high school completion associated with unit changes
in the given independent variable
Results in columns (1) and (11) are based on bivariate probit regressions
Results in columns (2)-(10) and (12) are from a model with full controls and both behavior measures.
Results in the top panel are based on regressions that include separate measures of reading and math achievement. Results in the bottom panel are based on regressions that
include a single composition measure of achievement.
Family and child controls include race; Hispanic ethnicity; gender; age 0-5: % years in poverty,% years with middle income, % years with middle high income, % urban
residence, % years mother never married, % years mother divorced, % years resided with grandmother, ave # children; child: age 3/4: PPVT standardized score, age 4/5:
compliance, age 4/5: sociability; household age5/6: urban residence, number of children, mother's education, poverty, child's father present in household, mother never
married, mother divorced , mother cohabiting with partner, mother married to partner, total home; mother: age at birth, mother academic aptitude (AFQT) , ever use
alcohol, mother fight, mother steal, age mother first tried smoking, mother never smoke, marijuana use: occasional, marijuana use: moderate, drug use: occasional, drug
use: high, mother lived with two parents at age, mother us born, mother drank alcohol during pregnancy, used prenatal care , mother smoked during pregnancy.
Appendix Table 3.A10. Summary of Probit Regressions of High School Completion on Patterns of Childhood
Antisocial Behavior, NLSY Ages 6, 8, and 10 (N=1,437 for high school completion and 1,081 for college
(1) (2) (3)
High School Completion
Ever Attended College by
age 21/22
Bivariate Adjusted Bivariate
Antisocial Problems
-.164*** -.056 -.189*** -.048
(.032) (.036) (.041) (.053)
-.346*** -.162* -.354*** -.165†
(.055) (.068) (.061) (.098)
Attention Problems
-.159*** -.023 -.185*** -.045
(.033) (.032) (.043) (.054)
-.268*** .033 -.313*** -.011
(.060) (.054) (.057) (.091)
Anxiety Problems
-.089** -.023 -.109** -.047
(.030) (.034) (.039) (.051)
-.229*** -.075 -.232*** -.114
(.055) (.070) (.063) (.088)
Reading Problems
-.206*** -.077* -.264*** -.114*
(.031) (.035) (.037) (.052)
-.319*** -.076 -.358*** -.092
(.055) (.066) (.063) (.099)
Math Problems
-.168*** -.058† -.211*** -.101*
(.032) (.034) (.042) (.051)
-.314*** -.133† -.438*** -.338***
(.060) (.073) (.047) (.076)
Family & Child Characteristics
Notes: *** p<.001; ** p<.01; * p<.05; † p<.1
Probit model coefficients and standard errors are expressed as "marginal effects" -- percentage point changes in the
probability of high school completion or college attendance associated with unit changes in the given independent
Columns 1 and 3 show bivariate coefficients between intermittent and persistent behavior problem groups and the
no problem reference group
Columns 2 and 4 include all listed variables, plus child and family controls simultaneously.
"Persistent" reflects cases above the 75th percentile at Ages 6, 8, and 10.
"Intermittent" reflects cases above the 75th percentile for at least 1 but not all 3 time points
Controls are listed in Appendix Table 9
Appendix Table 3.A11. Summary of Results from Probit Regressions of "Ever Arrested" on Achievement and Behavior Problems across Middle Childhood, NLSY
Age 5
Age 6
Age 7
Age 8
Age 9
Age 10
Age 11
Age 12
Age 13
Age 14
.050*** .033** .032* .032* .032* .033* .065*** .051*** .063*** .095*** .115*** .078***
(.011) (.012) (.015) (.015) (.016) (.014) (.019) (.013) (.014) (.013) (.012) (.015)
.042*** .006 .007 .009 -.003 .011 -.016 .005 -.008 -.008 .082*** .007
(.011) (.014) (.016) (.015) (.015) (.015) (.015) (.015) (.018) (.016) (.013) (.017)
.019† -.005 -.011 -.008 -.008 -.009 -.008 -.002 -.003 -.018 .055*** -.006
(.010) (.013) (.014) (.014) (.013) (.015) (.016) (.014) (.016) (.016) (.014) (.015)
-.036*** -.008 -.016 -.004 -.005 -.003 -.007 .001 .007 -.023 -.049*** -.035
(.011) (.014) (.015) (.016) (.017) (.016) (.017) (.015) (.019) (.023) (.012) (.028)
-.028* -.010 .018 -.008 .003 -.001 .022 -.006 -.010 .010 -.039*** .014
(.011) (.014) (.015) (.016) (.017) (.015) (.015) (.015) (.017) (.014) (.012) (.017)
.050*** .033** .032* .032* .032* .033* .065*** .051*** .062*** .095*** .115*** .079***
(.011) (.012) (.015) (.015) (.016) (.014) (.019) (.013) (.014) (.013) (.012) (.016)
.042*** .006 .007 .009 -.003 .011 -.016 .005 -.008 -.007 .082*** .008
(.011) (.014) (.016) (.015) (.015) (.015) (.015) (.015) (.018) (.016) (.013) (.017)
.019† -.005 -.011 -.008 -.008 -.009 -.007 -.002 -.003 -.019 .055*** -.007
(.010) (.013) (.014) (.014) (.013) (.015) (.016) (.014) (.016) (.016) (.014) (.015)
Achievement Composite
-.043*** -.019 .003 -.011 -.002 -.004 .015 -.006 -.003 -.007 -.065*** -.011
(.013) (.017) (.017) (.016) (.017) (.015) (.016) (.018) (.015) (.018) (.015) (.024)
Family & Child Char.
Sample Size
Notes: *** p<.001; **p<.01; *p<.05; †p<.1
Probit model coefficients and standard errors are "marginal effects" -- percentage point changes in the probability of ever being arrested associated with unit changes in the given
independent variable.
Results in columns (1) and (11) are based on bivariate probit regressions
Results in columns (2)-(10) and (12) are from a model with full controls and both behavior
Results in the top panel are based on regressions that include separate measures of reading and math achievement. Results in the bottom panel are based on regressions that include a
single composition measure of achievement.
Appendix Table 3.A12. Summary of Probit Regressions of Ever Arrested on Patterns of Childhood Achievement,
Attention and Behavior Problems at age 6, 8 and 10
NLSY (N=1,437 for All; 699 for Males only)
(1) (2) (3) (4)
All Males only
Bivariate Adjusted Bivariate Adjusted
Antisocial Problems
-.164*** -.056 -.212*** -.107†
(.032) (.036) (.048) (.057)
-.346*** -.162* -.395*** -.235**
(.055) (.068) (.059) (.083)
Attention Problems
-.159*** -.023 -.180*** -.004
(.033) (.032) (.042) (.050)
-.268*** .033 -.343*** -.007
(.060) (.054) (.066) (.088)
Anxious Problems
-.089** -.023 -.151*** -.061
(.030) (.034) (.045) (.059)
-.229*** -.075 -.319*** -.115
(.055) (.070) (.069) (.103)
Reading Problems
-.206*** -.077* -.225*** -.079
(.031) (.035) (.044) (.058)
-.319*** -.076 -.360*** -.111
(.055) (.066) (.072) (.104)
Math Problems
-.168*** -.058† -.198*** -.070
(.032) (.034) (.045) (.057)
-.314*** -.133† -.333*** -.097
(.060) (.073) (.080) (.114)
Family & Child Characteristics
Notes: *** p<.001; ** p<.01; * p<.05; † p<.1
Probit model coefficients and standard errors are expressed as "marginal effects" -- percentage point changes in the
probability of ever being arrested associated with unit changes in the given independent variable
Columns 1 and 3 show bivariate coefficients between intermittent and persistent behavior problem groups and the
no problem reference group
Columns 2 and 4 include all listed variables, plus child and family controls simultaneously.
"Persistent" reflects cases above the 75th percentile at Ages 6, 8, and 10.
"Intermittent" reflects cases above the 75th percentile for at least 1 but not all 3 timepoints
Controls are listed in Appendix Table 9.
1. These data regarding the Perry program are taken from Schweinhart et al. (2005).
2. To be sure, not all outcomes differed significantly between Perry and control children, but the
long-run impacts are impressive, as reflected both in the evaluation reports written by the
organization that ran the Perry study and in an independent reanalysis of the Perry data
(Heckman et al. 2009).
3. Direct assessments of young children’s inhibition require children to suppress a dominant or
congruent response, yet measures differ in the extent to which the tasks also include an
emotional component. A measure of cognitive self-regulation involves suppressions but little
emotional work.
4. Emotional self-regulation is measured by tasks that require children to control (typically de-
escalate) their emotions, usually their excitement. Most often these tasks for young children
involve delaying the gratification of a desired reward—candy or a gift. For example, in one task
a child is told not to peek as the assessor noisily wraps a present in front of the child.
5. We chose math over reading owing to second-language complications with early reading
scores; the appendix tables show broadly similar patterns for math and reading.
6. For example, the unadjusted 13.8 percent difference between low and high SES and arrest
rates falls to 10.2 percent, or by about one quarter.
7. The data sets included the Children of the National Longitudinal Survey of Youth (NLSY), the
National Institute of Child Health and Human Development Study of Early Child Care and
Youth Development (NICHD SECCYD), the 1970 British Birth Cohort (BCS), the Early
Childhood Longitudinal Study Kindergarten Cohort (ECLS-K), the Infant Health and
Development Program (IHDP), and the Montreal Longitudinal-Experimental Preschool Study
8. This conclusion held both across studies as well as within each of the six data sets they
examined. Their analysis included a sixth category—school-entry social skills—which also
proved to be completely unpredictive of later school achievement.
9. Key results from the meta-analysis appeared robust to a host of potential problems related to
measurement and modeling, including the inclusion of controls (see Duncan et al. 2007).
10. The Duncan et al. (2007) analysis was of population-based data sets that provided little to no
ability to identify children with diagnosed conduct disorder, attention deficit disorder, or other
behavioral conditions. It is best to think of their analyses as focusing on children with relatively
high, but not clinical levels of learning, attention, and behavior problems.
11. The shading on the bars in figure 3.6 indicates levels of statistical significance, with light
shading indicating p < .05 and darker shading indicating p < .01.
12. No consistently significant differences by family SES were found in either the attainment or
crime analyses.
13. Two words of caution to this conclusion. First, arrest is an imperfect and incomplete way to
measure criminal behavior, and it does not distinguish between types of criminal behavior
(violent vs. nonviolent). Second, although Head Start programs rarely match the intensity of
model programs such as the Perry one, Deming’s (2009) sibling-based analysis of Head Start
showed long-run impacts on arrests but not shorter-run impacts on behavior problems.
... Understanding the association between the pandemic and preschool children's problem behaviors is important, as maladjustment in early childhood can cause adjustment problems persisting throughout their adulthood. Extant literature suggests that skill and behavioral problems in early childhood are associated with later educational attainment such as high school graduation and college attendance (Ansari, 2018;Duncan & Magnuson, 2011;Duncan et al., 2010;Magnuson & Duncan, 2016;Sylva et al., 2010). ...
... Preschool children's behavioral, emotional, and social problems are often measured according to two broad dimensions, namely, internalizing and externalizing problem behaviors (Achenbach et al., 2016;Duncan & Magnuson, 2011). Internalizing problem behaviors refer to depression, anxiety, withdrawal, and somatic complaints. ...
... Externalizing problem behaviors refer to aggressive and rule-breaking behavior (Achenbach, 1999). Internalizing and externalizing problem behaviors are common among preschool children and are significant as both problem behaviors are persistent through adulthood affecting life outcomes (Duncan & Magnuson, 2011). Therefore, this study focused on mothers raising preschool children and aimed to understand the mechanisms of the impact of mother's COVID-19 stress on preschool children's behavioral problems. ...
Full-text available
The COVID-19 pandemic is affecting families and children worldwide. Experiencing the pandemic leads to stress in families resulting from fear of infection and social isolation derived from social distancing. For families raising preschoolers, the prolonged closure of childcare centers puts additional childcare burden on family members, especially mothers. Due to the limited research exploring the impact of COVID-19 on preschool children’s problem behaviors, this study examines the association between stress due to COVID-19 and preschool children’s internalizing and externalizing problem behaviors related to mother’s depression and parenting behavior. The study sample included data collected from 316 South Korean mothers raising preschool-aged children aged 3 to 5. The study findings suggest that mother’s COVID-19 stress was indirectly associated with preschool children’s internalizing and externalizing problem behaviors resulting from the mother’s depression and parenting behaviors, although the direct effect of COVID-19 stress on preschool children’s outcomes was not statistically significant. Increase in mother’s COVID-19 stress was associated with increase in depression, and sequentially decreased positive parenting behaviors, which in turn resulted in preschool children’s internalizing and externalizing problem behaviors. The study findings highlight the need to focus on enhancing mental health of mothers and preschool children’s adjustment by implementing supportive interventions to reduce the adverse impacts of the prolonged COVID-19 pandemic.
... Geary (2011a) confirmed the relation between poor math skills and unemployment, low chances to get promotion and low SES. Another study (N = 21260) revealed that children with math problems had less chance to end their secondary school with a diploma and to enter higher education (Duncan & Magnuson, 2011). addition, regarding mathematics, children di er as far as their motivation and a ect are concerned. ...
... Mathematics is important in our society (Duncan & Magnuson, 2011;Geary, 2011a & b;Ojose, 2011). The Opportunity-Propensity (O-P) model Byrnes & Miller, 2007;Wang et al., 2013) integrates predictors of learning, and helps gaining insight into how predictors are interrelated, and whether some are more important than others. ...
Several factors seem important to understand the nature of mathematical learning. Byrnes and Miller combined these factors into the Opportunity-Propensity model. In this study the model was used to predict the number-processing factor and the arithmetic fluency in grade 4 (n = 195) and grade 5 (n = 213). Gender, intelligence and affect (positive affect for arithmetic fluency and negative affect for calculation accuracy) predicted math learning, and pointed to the importance of the propensity factors. We have to be careful not to interpret gender differences, since this is a social construct, our analyses pointed to the relevance of including antecedent factors in the model as well . The Implications of the study for math learning will be discussed below.
... Extensive research from the sociology of education, the economics of education, and developmental psychology has documented how cultural, economic, and social resources in the family shape children's skill development early in life through monetary investments and parenting practices (Francesconi & Heckman, 2016;Duncan & Magnuson, 2011;Farkas, 2003;Bourdieu, 1986). ...
Full-text available
Information and communications technology (ICT) skills are crucial for labour market success and full participation in society. Socioeconomic status (SES) inequality in the development of ICT skills would prevent disadvantaged children from reaping the benefits of the digital age. Besides, the digital divide in ICT literacy might add to the already well-documented large and persistent SES inequality in ‘hard’ skills—like math, reading, and science. This article studies the roots, evolution, and drivers of SES inequality in ICT literacy from age 8 to 15 in Germany. Drawing from the German National Educational Panel Study (NEPS), we highlight five main findings: (1) SES gaps in ICT literacy exist as early as age 8 (grade 3) and are similar in size compared to SES gaps in hard skills; (2) like hard skills, SES gaps in ICT literacy remain stable over primary and tracked lower secondary schooling; (3) ICT access and use at home and school do not substantially explain SES gaps in ICT literacy at any age; (4) selection into school tracks seems a critical pathway, although not necessarily a causal one, leading to SES inequality in secondary school; (5) SES gaps in ICT literacy are not observed among children with similar levels of hard skills. We discuss the implications of these findings for the interdisciplinary literature on social stratification, skill formation, and the digital divide.
... This increase in the achievement gap translates to scores 1.25 standard deviations higher on standardized tests, on average, for wealthier students compared to their lower SES peers (Reardon, 2011; U.S. Department of Education). Similar gaps favoring higher SES students are found in other academic measures including grade point averages (Sirin, 2005;White, 1982), high school completion rates (Brooks-Gunn & Duncan, 1997;Duncan & Magnuson, 2011), and college entry and completion (Bailey & Dynarski, 2011). ...
... Importantly, some of the ways the challenges of poverty manifest themselves are in the expression of anti-social behaviors that spill over into the school experiences of peers. Lower-familyincome students are much more likely to have classroom peers who have experienced a higher frequency of childhood traumatic events and who are more likely to exhibit inappropriate classroom behavior (Duncan and Magnuson, 2011). Further, lower-family-income students who have traumatized children in their classrooms are also more likely to misbehave in class, as a product of the presence of traumatized children (Carrell and Hoekstra, 2010). ...
... This troubling finding that certain groups of children are experiencing considerable educational gaps has been confirmed by decades of research (Fryer & Levitt, 2006;Phillips, Crouse, & Ralph, 1998). Even more alarming, research suggests that these educational gaps are evident early in children's educational lives and persist or become larger over time (Duncan & Magnuson, 2011). ...
... Department of Education, 2014). Preschool age is a particularly important developmental period where children exhibit rapid growth across the different domains of school readiness such as self-regulation skills (Pianta, 2007), which are important for later academic achievement (Panlilio et al., 2018;Panlilio & Corr, 2020;Duncan & Magnuson, 2011). It is also a period in which young children are reliant upon the quality of their early childhood education (ECE) experience, especially one in which their ECE teachers are able to model and scaffold socioemotional competence such as appropriate regulation of emotions (Jennings & Greenberg, 2009). ...
Full-text available
Despite the known relationship between trauma and academic outcomes, including expulsion risk, for preschoolers, little is known about the role that teachers may play in addressing the effects of childhood trauma within preschool settings. The current study examined the relationship between a teacher’s overall stress, trauma-informed attitudes, and indicators of children’s expulsion decision risk using a sample of preschool lead and assistant teachers (n = 129) recruited from Head Start classrooms in the Mountain West. Multivariate multiple regression was used to determine whether teachers stress and trauma-informed attitudes (trauma-informed knowledge, self-efficacy, and reactions) were related to three indicators of expulsion decision risk using subscales of the Preschool Expulsion Risk Measure (classroom disruption, fear of accountability, and child-related stress) for the most disruptive child in the teacher’s classroom. Higher overall stress significantly predicted higher fear of accountability (β = 0.26, 95% CI = 0.07, .45, p = 0.007). Higher trauma-informed knowledge was significantly related to lower child-related stress (β = −0.40, 95% CI = −0.63, −.17, p = 0.001). Higher trauma-informed self-efficacy was significantly related to lower classroom disruption (β = −0.45, 95% CI = −0.66, −.25, p < 0.001). Multigroup models revealed significantly different pathways for children of color (Black, Latinx, and American Indian children) compared to White children; teacher stress predicted higher expulsion decision risk for children of color and trauma-informed attitudes predicted lower expulsion decision risk for White children. Implications for development and evaluation of trauma-informed approaches for early childhood settings are discussed.
... The effects and symptoms in different behavior aspects include the child's developmental rate index [1], [2]. Due to global health, care, and awareness, more children survive that supporting their health and quality of life is an absolute priority. ...
Full-text available
Keywords: ABSTRACT Social determinants of health; Social factors; Development of children; Children. Children's development is a complex issue in which social factors and their rising environment influence their behavior and practice changes. The present study aimed to identify and prioritize social factors and strategies affecting children's developmental health. The statistical population consisted of 12 physician's experts and experts related to the discussion of children's development in Tehran for the qualitative section, among which the experts' interview and a questionnaire were distributed, the subjects were selected using the snowball method, and seven of them completed the questionnaire for DEMATEL questionnaire and 5 for TOPSIS questionnaire. The study's main purposes include five criteria of individual factors related to children, family factors, environmental factors, governance factors, and global factors. The research indicators were identified using the Delphi method. Data analysis was performed using the DEMATEL-TOPSIS approach. It was proved that the "family factors" criterion was the priority among the criteria and had the most significant influence and interaction with other criteria, and the criteria of governance factors had the most significant impact among the criteria. The best solution is "improving society's health status and correcting the health behaviors of families." The strategy of "creating and developing academic disciplines or trends in senior, doctoral and specialized medical levels, related to the growth and development of the child" is in the last place. Social factors play an essential role in children's developmental health, and paying attention to them can lead to an improvement in children's developmental health process. This work is licensed under a Creative Commons Attribution Non-Commercial 4.0 International License. Sadrkhanlou,, 2021 Teikyo Medical Journal 2368 1. INTRODUCTION Changing attitudes towards children, childhood, and childhood life is a new subject and has been placed on university institutions, research, and human rights defenders in the last period of human and legal demands. This trend has many historical ups and downs; for example, before the Renaissance, children were the tools of adult hands to meet emotional needs, power, property, culture transfer, hereditary authority, and order preservation. Meanwhile, childhood was a neglected concept and only a specific biological period that had to be rapidly traveled on the path of growing up. Among the various historical, social, economic, and political developments that have influenced childhood attitudes. Therefore, the child's development is done regularly and continuously in a specific context and plan. Growth is visible, evaluation, and measurement reflected in the nervous system through physiological symptoms and behavior.
Although positive effects of future thinking have been demonstrated, the effects of future thinking on children’s academic achievement are less known. We examined the effects of three forms of thinking about the future or alternative outcomes on math performance in 9- to 12-year-olds (N = 127). After a math pre-assessment, participants were asked to think about math success according to a between-subjects condition: episodic prospection (episodically simulating a personal future event), semantic prospection (thinking about the future in a non-personal, general sense), or episodic counterfactual thinking (episodically simulating an alternative past event). Results show that semantic prospection promoted gains in mean math accuracy and a greater proportion of 3rd-person visual perspective. A 3rd-person visual perspective also related to gains in mean math accuracy across conditions. Semantic prospection may be a more beneficial form of future thinking in some contexts, perhaps because it supports greater psychological distancing. Academic achievement interventions may benefit from targeting specific forms of future thinking.
This chapter, from a student’s perspective, reports on a descriptive and exploratory analysis of a Hong Kong (HK) undergraduate’s transformative growth through her 4-year residential experience in the United Kingdom, Canada and HK respectively, using the “Attitude-Skills-Knowledge (ASK)” educational model which proposes that a person’s ability is the synergy among the three domains. The chapter will first highlight the undergraduate’s use of the ASK model in decomposing her experience into three learning categories. Then, it will include her personal reflection on how her learning synergistically equips her with transferable learning outcomes. In conclusion, plans for how stakeholders at different levels can help promote the ASK model to facilitate student residents’ growth will be presented.
Full-text available
The study examined the multivariate relationship between dimensions of preschool emotional and behavioral adjustment assessed at the beginning of the year by the Adjustment Scales for Preschool Intervention (ASPI) and multiple learning and social competencies at the end of the year with an urban Head Start sample. This study also examined the differential pattern of emotional and behavioral problems for children identified to receive services by Head Start staff. Results indicated that overactive dimensions at the beginning of the year predicted socially disruptive problems in the classroom at the end of the year. Underactive problem dimensions were associated with disengagement in play and poor emergent literacy and classroom learning outcomes. Findings indicated that Head Start staff under identified children with emotional/behavioral problems as a group, with a pattern toward identifying children with overactive needs. Children with underactive needs were least likely to be detected by the staff and were most at-risk for poor school readiness outcomes. Implications for policy, practice, and future research are discussed
Sufficient research now exists in the psychology of criminal conduct literature to address the long-term impact of early childhood and adolescent experiences on later adult outcomes. In the present meta-analysis, selected studies were prospective and longitudinal, tracking a variety of early childhood and family factors that could potentially predict later involvement in the adult criminal justice system. Thirty-eight studies met the selection criteria. Major findings indicate that dynamic versus static predictors are related to later adult criminal justice involvement. The older the child was at the time the predictor was measured, the stronger was the relationship to adult offending. Within the set of dynamic predictors, childhood and adolescent factors that rate most highly include a variety of behavioural concerns including early identification of aggression, attentional problems, motor restlessness, and attention seeking. Emotional concerns consistent with depression including withdrawal, anxiety, self-deprecation, and social alienation are also represented. Predictors also included family descriptors such as a variety of negative parenting strategies including coerciveness, authoritarian behaviours, lack of child supervision, and family structure variables such as witnessing violence, inter-parental conflict, family stressors, and poor communication. Results are discussed in relation to prevention strategies for targeted services that influence the probability of antisocial outcomes for children into adulthood.
As we enter the twenty-first century, poor and non-Asian minority students lag considerably behind their nonpoor, Asian, and white counterparts on many dimensions of academic performance. Although scholars have long known that these academic disparities stem from many causes, commentators on both sides of the political spectrum often attribute these gaps to disparities in school quality. Thus, President George W. Bush has promoted his "No Child Left Behind" education reform legislation as a crusade against lowquality schools. "We don't want schools languishing in mediocrity and excuse-making," Bush said in 2002. "We want the best for every child. . . . And that starts with making sure that every child gets a good education." But just how unequal is the U.S. educational system? Do schools that serve disadvantaged students "languish in mediocrity"? It is certainly true that dilapidated facilities staffed by inexperienced teachers haunt journalists' depictions of the schools that serve disadvantaged students (see, for example, Kozol 1991). But to what extent do these portraits accurately describe the typical schools that disadvantaged students attend? This chapter uses national data to examine the prevalence of "savage inequalities" at the turn of the twentyfirst century. We assess not only the extent to which poor and nonwhite students attend "worse" schools than their nonpoor and white counterparts, but also whether these inequalities have widened or narrowed since the late 1980s. In addition, we discuss whether reducing disparities in any particular dimension of school quality is likely to reduce disparities in students' academic achievement. We begin by describing trends in academic performance among students from different ethnic and socioeconomic backgrounds. In the next section, we briefly examine the extent of ethnic and socioeconomic segregation across schools. The third and most important section describes ethnic and socioeconomic disparities in public school quality and, when possible, whether these disparities widened or narrowed over the 1990s. We first examine inequities in teacher quality, including differential access to well-educated, credentialed, experienced, and academically skilled teachers. Then we describe disparities in access to instructional attention, as reflected both in the amount of time students spend in school and in class sizes. Next we look at inequalities in instructional resources, as measured by the availability of "gifted" or "advanced placement" programs, instructional materials, computer technology, visual and performing arts offerings, and exposure to academically oriented peers. We conclude this section by describing disparities in access to school services, comfortable facilities, and a safe school environment. The next section briefly discusses whether our focus on public schools understates the extent of education inequality in the United States, and finally, we summarize our main conclusions and offer suggestions for future research.
In the last and current decade, the W ake County school district reassigned numerous students to schools, moving up to five percent of the enrolled population in any given year. Before 2000, the explicit goal was balancing schools'racial composition; after 2000, it was balancing schools'income composition. Throughout, finding space for the area's rapidly expanding student population was the most important concern. The reassignments generate a very large number of natural experiments in which students experience new peers in the classroom. Using panel data on students before and after they experience policy-induced changes in peers, we explore which models of peer effects explain the data. We also review common models and econometric identification of peer effects. Our results reject the popular linear-in-means and single-crossing models as stand-alone models of peer effects. We find support for the Boutique and Focus models of peer effects, as well as for a monotonicity property by which a higher achieving peer is better for a student's own achievement all else equal. Our results indicate that, when we properly account for the effects of peers'achievement, peers'race, ethnicity, income, and parental education have no or at most very slight effects. W e compute that switching from race-based to income-based desegregation has at most very slight effects, so that W ake County's numerous reassignments mainly affected achievement through the redistribution of lower and higher-achieving peers.
This work examines associations between closeness and conflict in teacher-child relationships and children's social and academic skills in first grade in a sample of 490 children. Assessments of teacher-child relationships were obtained in preschool, kindergarten, and first grade. Results demonstrate moderate correlations among teachers' ratings of conflict and slightly lower correlations among teachers' ratings of closeness across years. Hierarchical regression analyses predicted children's skills in first grade from teacher-child relationship quality. Child gender, socioeconomic status, and preschool estimates of outcomes of interest were controlled statistically. Although preschool assessments of social and academic skills were closely associated with individual skill differences at first grade, teacher-child relationship quality also was associated with changes in skill levels. Findings generally confirm that teacher-child relationships play a role in children's ability to acquire the skills necessary for success in school.
We traced the emerging relations between children's understanding of multidigit numbers and their computational skill and investigated how instruction influenced these relations. We followed about 70 children over the first 3 years of school while they were learning about place value and multidigit addition and subtraction in 2 different instructional environments. By interviewing the students several times each year, we found that understanding and skill were closely related on tasks for which students had not yet received instruction as well as on more difficult tasks even after instruction. Students appeared to apply specific understandings to invent new procedures and modify old ones. The alternative instruction, which encouraged students to develop their own procedures and to make sense of procedures presented by others, appeared to facilitate higher levels of understanding and closer connections between understanding and skill.