Content uploaded by Seung-Hee Claire Son
Author content
All content in this area was uploaded by Seung-Hee Claire Son
Content may be subject to copyright.
The Relationship of Young Children’s Motor Skills
to Later Reading and Math Achievement
Seung-Hee Son, Purdue University
Samuel J. Meisels, Erikson Institute
This study examined empirical evidence about the relationship between motor
skills at the beginning of kindergarten and reading and mathematics achieve-
ment at the end of first grade, using the Early Childhood Longitudinal Study—
Kindergarten cohort national dataset (N= 12,583). Results of hierarchical
regression analyses demonstrated that early kindergarten motor skills, especially
visual motor skills, add a small but unique amount of variance to achievement in
reading and mathematics at the end of first grade even after controlling for
initial skills and demographic information. Furthermore, Receiver-Operating-
Characteristic curve analyses showed that information from visual motor skills is
useful in identifying children at risk for academic underachievement. The results
suggest the importance of the role that motor skills can play in designing and
implementing an early school achievement battery.
The relationship between motor and cognitive skills has long been the sub-
ject of research in several fields. In child development, Piaget accorded sen-
sorimotor skills a central role in children’s early cognitive development.
According to Piaget’s (1952) developmental theory, motor skills contribute
to infants’ active exploration of the environment, and it is through such
actions that infants construct their knowledge of the world. Related studies
demonstrated that infants’ experiences of self-produced locomotion (e.g.,
crawling) are related to such cognitive skills as object permanence and the
755
M
ERRILL
-P
ALMER
Q
UARTERLY
, V
OL
. 52, N
O
. 4
Seung-Hee Son, Department of Child Development and Family Studies, School of Educa-
tion; Samuel J. Meisels, President, Erikson Institution.
The authors would like to thank Frederick J. Morrison and Joanne F. Carlisle for invaluable
feedback on earlier versions of this manuscript.
Correspondence should be addressed to Seung-Hee Son at the Department of Child Develop-
ment and Family Studies, Purdue University, 101 Gates Road, West Lafayette, IN 47907. Phone:
(765) 496-1687. Email: seunghee@purdue.edu.
Merrill-Palmer Quarterly, October 2006, Vol. 52, No. 4, pp. 755–778. Copyright © 2006 by Wayne
State University Press, Detroit, MI 48201.
050 seung (755-778) 12/1/06 11:06 AM Page 755
organization of spatial information (Bai & Bertenthal, 1992; Bertenthal,
Campos, & Kermoian, 1994; Campos et al., 2000).
Advances in neuropsychology also provide information about the rela-
tionship between motor skills and cognition based on brain activity and struc-
ture. In one of the classic theories of neuropsychology, it was suggested that
the same brain structure can participate in more than one functional system,
the same functional system can draw upon multiple local structures distrib-
uted throughout the brain, and the functional systems can reorganize through-
out development (Luria, 1973). Ellis (1985, 1987) applied a similar notion to
literacy acquisition. He proposed a cognitive neuropsychological model for
reading and writing acquisition in which reading and writing modules are not
preformed in the infant brain waiting to be elicited by a certain kind of envi-
ronmental stimulation at a particular time. Rather, the reading and writing
systems are constructed from other cognitive capabilities, such as the visual,
phonological, and semantic systems. Thus, developmental reading and writ-
ing disorders can be the consequence of disorders in other systems from
which reading and writing skills emerge.
Arelated review reported that intellectual and perceptual-motor skills
are acquired in fundamentally similar ways (Rosenbaum, Carlson, &
Gilmore, 2001). According to this study, learning rates, training effects, and
learning stages are highly similar for the two sets of skills. In addition, brain
sites serving thought and perceptual-motor processes are not as distinct as
once thought (Diamond, 2000). This may be explained by the theory that
motor and cognitive systems develop dynamically by interacting with each
other (Iverson & Thelen, 1999; Satz & Fletcher, 1988; Smith, Thelen,
Titzer, & McLin, 1999).
Overall, the weight of the theoretical evidence provides a conceptual
justification for examining motor skills in relationship to later cognitive
achievement. The literature suggests that development of motor skills is
associated with development of cognitive skills and motor skills can be an
indicator of cognitive skills development.
Predictability of Early Motor Assessments
Predictability is often regarded as fundamental to early childhood measure-
ment and is an important aspect of psychometric validity (Meisels, 1994).
Validity represents an overall evaluative judgment of the degree to which
empirical evidence and theoretical rationales support the adequacy and
appropriateness of conclusions drawn from some form of assessment (Mes-
sick, 1989). Predictive validity generally implies the degree to which an
instrument of interest provides accurate measurement by comparing scores
756 Merrill-Palmer Quarterly
050 seung (755-778) 12/1/06 11:06 AM Page 756
from the instrument with scores on a relevant criterion variable of later devel-
opment (Bryant, 2000). For example, screening tests that purport to identify
children at risk for future achievement problems should include appropriate
indicators presumed to show a predictive relationship to skills not yet
achieved. Information about predictability is thus an indispensable aspect of
validity for early childhood assessment (Meisels & Atkins-Burnett, 2000).
Reports concerning the predictive validity of motor skills assessments
for later cognitive achievement are mixed. In some studies, children’s
visual-perceptual or visual motor skills were among the best predictors of
reading achievement in first through third grades (Tramontana, Hooper, &
Selzer, 1988), with a mean correlation of .38 between visual-perceptual and
reading achievement (Kavale, 1982). In other studies, some of which con-
trolled for initial cognitive skills, correlations between motor and cognitive
achievement were generally <.35 or sometimes statistically nonsignificant
(Lesiak, 1984). Visual motor skills tend to have higher correlations with
reading and mathematics achievement than do gross motor skills (Payne &
Isaacs, 1999), although gross motor skills are also reported to have signifi-
cant correlations (Knight & Rizzuto, 1993).
The tenuous long-term predictability of motor skills based on data
obtained from measurements in early childhood is not, in fact, specific to
motor assessment. One of the reasons for weak predictability is the instabil-
ity of behaviors in young children. In addition to month-by-month changes,
some conditions identified as delayed or at risk in young children may not
be present as they grow older. For example, in the Kopp and McCall (1982)
and McCall (1981) research, instability in individual differences during the
first two years of life was demonstrated repeatedly in cognitive and motor
test performance. In still another study of gross motor development, 31 of
800 children demonstrated poor gross motor skills at age three, but only 10
of the 31 children continued to exhibit poor motor skills at age five (Silvia
& Ross, 1980). The remaining 21 children had caught up by age five. In
other words, age-to-age stability in mental performance increases at about
two years of age, increasing rapidly until approximately five years of age
for normally developing children (Meisels with Atkins-Burnette, 1994).
This suggests that measuring children’s skills when they are kindergartners
tends to have higher predictability for later achievement than testing when
they are younger.
Whatever the stability of motor skills, the problem of prediction may
lie as much with the analytic design of the investigation as with the phe-
nomena being measured. Amajor methodological problem with these stud-
ies is the use of correlation coefficients as evidence of the tests’ accuracy.
Despite the fact that the bivariate correlations provide information on the
Children’s Motor Skills 757
050 seung (755-778) 12/1/06 11:06 AM Page 757
similarity of the group’s performance on tests of motor and cognitive skills,
such studies provide no information about the predictability or specific
identification of children at risk for cognitive underachievement. One of the
most useful ways to evaluate a test’s predictive accuracy is through an
examination of individual classification decisions (Meisels, 1989). Another
way is to use multivariate statistical methods that can examine predictabil-
ity while controlling for such confounding variables as initial skills or
demographics (Bryant, 2000). Multivariate methods also make possible
examination of the combined as well as the independent effects of predic-
tors, a strategy that is often used in investigations of motor skills (Broad-
head & Church, 1988; Schmidt & Perino, 1985).
Using multivariate methods and individual classification decisions, the
present study focuses on exploring the relationship between motor skills
measured at kindergarten entry and cognitive achievement during early
schooling. Early years of formal schooling is the period when children are
instructed to develop basic academic skills in reading and mathematics,
which provides foundations for later school achievement. Considering the
importance of these basic skills and possible relations between them and
motor skills, we chose to examine the predictive relationship from the
beginning of kindergarten to the end of the first grade. Specifically, this
study uses motoric items adapted from the Early Screening Inventory-
Revised (ESI-R) (Meisels, Marsden, Wiske, & Henderson, 1997) adminis-
tered in fall kindergarten to predict achievement at the end of first grade
based on data from a national survey of early childhood learning. Astudy of
the predictive relationship between motor scores in early kindergarten and
cognitive achievement two schooling years later may shed light on the
value of focusing on motoric skills at the outset of school as well as on the
dynamic nature of child development among the developmental domains.
Since previous studies reported that visual motor skills have differen-
tial predictability for aspects of achievement (i.e., visual motor skills are
more predictive of mathematics than of reading) (Goldstein & Britt, 1994;
Kurdek & Sinclair, 2001), separate analyses were conducted on reading and
mathematics achievement. We hypothesized that we would discover a sig-
nificant relationship between early motor and achievement in reading and
mathematics in first grade and that early motor skills, especially visual
motor skills, would significantly predict later achievement. In addition, the
combined effects of motor and initial achievement were examined to test
the usefulness of adding motor items to assessments of school success.
Examining combined effects as well as unique effects of motor skills could
provide helpful information on the associations among motor and cognitive
domains of development.
758 Merrill-Palmer Quarterly
050 seung (755-778) 12/1/06 11:06 AM Page 758
Method
Data Source
To examine the relationship between kindergartners’ motor skills and cogni-
tive achievement in reading and mathematics two years of schooling after
initial testing, we used data from the Early Childhood Longitudinal Study-
Kindergarten Class of 1998–99 (ECLS-K). The ECLS-K is the first nation-
ally representative sample that focuses on children’s early school experiences
beginning from kindergarten. It provides information about children’s early
home, preschool, and kindergarten environments; their status at entry into
school; and their progression through the school year as measured in the fall
and spring of kindergarten and first grade (National Center for Education
Statistics, 2002).1
The ECLS-K uses complex survey designs. In the base year, the ECLS-
K data were collected using a multistage probability sample design to select
a nationally representative sample of children in the 1998–99 school year.
The primary sampling units consisted of 1,335 counties or groups of coun-
ties from which 1,280 public or private schools offering kindergarten pro-
grams were selected. A target sample of approximately 24 children from
each public school and 12 children from each private school was con-
structed, totaling 21,260 children attending kindergarten in the fall of 1998.
The ECLS-K longitudinal K–1 data file includes 17,212 children who were
assessed in the spring of the kindergarten year (spring 1999) and at least
once more during the first-grade year (fall 1999 or spring 2000).
Children’s motor skills were assessed in fall kindergarten, and reading
and mathematics achievement was assessed in the fall of kindergarten and
the spring of first grade. Fall kindergarten scores served as a measure of ini-
tial achievement and spring first-grade scores as an indicator of later
achievement. The time interval between initial and later achievement test-
ing was between 15.8 and 21.5 months, with an average of 18.4 months.
Children’s Motor Skills 759
1. By kindergarten year we mean the first year of data collection, the 1998–99 school year. By
first-grade year we mean the second year of data collection, school year 1999–2000. Although all
of the children in the kindergarten class of 1998–99 did not become first-graders by the second year
and some children repeated kindergarten or skipped the first grade, we continue to use the term
“first grade” or “first-grade year” for the second year of data collection throughout the paper for
convenience. The reason we did not include only children who were first-graders in 1999 in the
sample is that the current sample is more representative of the kindergarten children of the 1998–99
school year with more variability than is the first-time kindergarten and first-grade–only sample.
Additional analyses show that the results from the first-time kindergartners and first-grade–only
sample were very similar to those of the current sample.
050 seung (755-778) 12/1/06 11:06 AM Page 759
The Analytic Sample and Sample Weights
The sample used in this analysis was a subset of the ECLS-K, consisting of
first-time kindergartners and excluding kindergarten repeaters. First-time
kindergartners are children without any previous kindergarten experiences
when they entered kindergarten in the 1998–99 school year. Among the
first-time kindergartners, only children with information available on both
motor skills and achievement outcome measures were included in the ana-
lytic sample, consisting of approximately 12,500 children.
Analyses using ECLS-K data employ weights to compensate for the
unequal probabilities of selection and nonresponse rate. In this way, results
based on the weighted sample can be generalized to the children in the
United States who were kindergartners during the 1998–99 school year for
the first time. In using the weights we normalized the longitudinal child
weight variable by making it sum to the total sample size in order to avoid
inflating the sample size and miscalculating standard errors and degrees of
freedom. This enabled us to maintain the same sample size within the con-
text of a representative national sample.
The demographic characteristics of the analytic sample were examined
using the weighted sample (N= 12,583). The average age of the sample in
early September of the 1998–99 kindergarten year was 65 months. Although
the analytic sample included only first-time kindergartners, the oldest chil-
dren in the sample reached 83 months of age and the youngest children 49
months of age (SD = 4.07). In other words, the sample included some chil-
dren whose kindergarten entry was delayed by a year and some who were
promoted to kindergarten one year early.2Median family income was around
$40,000 (SD = $53,486.41). Approximately 20% of the sample reported that
their income was below the poverty level, and about 10% of the children used
a home language other than English. Gender of children was distributed
nearly evenly, with 50.4% males. The sample had a fair representation of
minority children, with 15.5% African American and 17.5% Hispanic.
760 Merrill-Palmer Quarterly
2. A kindergarten class may include (1) young kindergartners who apparently were enrolled
in kindergarten early (younger than 59 months old, if September is assumed as the school cutoff),
(2) at-age kindergartners (60–71 months old), and (3) older kindergartners who apparently were
held out by their parents (>71 months old). The presence of younger or older kindergartners in this
sample is not unusual; rather, they are representative of the full kindergarten classroom spectrum in
the nation. The current study sample (49–83 months old) includes all three groups of children.
However, the age range of the sample is limited to a three-year span. Children who are two years
older or two years younger than those at age were not included. We assume that this allows for chil-
dren who were held out by their parents or enrolled early but retains enough homogeneity in the
sample that it is not overly influenced by children who are much older or much younger than com-
mon practice would normally enroll in kindergarten.
050 seung (755-778) 12/1/06 11:06 AM Page 760
Measures
Motor skills. Motor skills were assessed at the beginning of kinder-
garten using visual motor and gross motor scales derived primarily from a
developmental screening instrument, the ESI-R (Meisels, Marsden, Wiske,
& Henderson, 1997). The ESI-R is a well-standardized multidomain devel-
opmental screening test that has been widely used to identify preschool and
kindergarten children who may be at risk for school failure (Kimmel, 2001;
Paget, 2001).
The visual motor scale used in this study includes fine motor and eye-
hand coordination measured with seven tasks of building a gate, drawing a
person, and copying five simple figures: circle, cross, square, triangle, and
open square and circle. Gross motor skills consisted of balancing, hopping,
skipping, and walking backwards. Most items in the motor scales of the
current study used the ESI-R items except the open-square-and-circle fig-
ure and walking backward items, which were added to solve a potential
ceiling problem in the screening test that would preclude the use of certain
statistical procedures.
Descriptive analysis based on the current analytic sample indicated
that visual motor scores showed more variance than gross motor scores.
Visual motor scores had a mean of 5.75 (SD = 2.06) in a range of possible
values of 0–9; the average gross motor score was 6.34 (SD = 1.87) in a
range of values of 0–8. The one available reliability measure was interitem
reliability (alpha coefficient), and it was .57 for the visual motor and .51 for
the gross motor scale (National Center for Education Statistics, 2001). The
strength of these reliabilities is largely a function of the binary scoring sys-
tem of the items that comprise the scales.3
Cognitive achievement. Cognitive achievement was assessed in the
domains of reading and mathematics using direct cognitive assessments
developed by the National Center for Education Statistics (NCES) for the
ECLS-K. Reading achievement was assessed with five subscales: identify-
ing letters, letter-sound association at the beginning of words, letter-sound
association at the end of words, recognizing words, and reading words in
context. Measurement of mathematics achievement was designed to assess
five basic mathematics skill levels: number and shape, relative size, ordi-
nality and sequence, addition/subtraction, and multiplication/division.
Children’s Motor Skills 761
3. Since the 0–1 or 0–2 scoring system of the motor scale limits variance of scores within
each item, thus restricting the size of reliability coefficients, interitem reliability is not the best kind
of measure of reliability in a scale with this type of scoring. Nevertheless, alpha coefficients were
the only available reliability measure provided by NCES. This might have influenced the relatively
small effect size of motor skills in predicting cognitive achievement.
050 seung (755-778) 12/1/06 11:06 AM Page 761
We used Item Response Theory (IRT)-based composite scores of read-
ing and mathematics from spring first grade as outcome variables and those
of fall kindergarten as initial status. IRT-based assessment provides a rigor-
ous measure of overall achievement through statistical manipulation,
examining the pattern of correct, incorrect, and omitted responses of chil-
dren to the items and the difficulty, discriminability, and likelihood of
guessing for each item (Hambleton, Swaminathan, & Rogers, 1991). The
IRT makes it possible to establish a common scale of item difficulty and a
common scale of child ability or level of achievement. It is then possible to
estimate the score of the child based on the test items with differing diffi-
culty by placing each child on a continuous ability scale. In addition,
absence of a ceiling problem in the IRTmakes it possible to examine longi-
tudinal measurement of gain.
IRT scores in the data indicated that children improved their level of
reading and mathematics skills for two schooling years from fall kinder-
garten to spring first grade. Reading IRT scores had a mean of 22.81 (SD =
8.47) and 55.72 (SD = 13.77) for fall kindergarten and spring first grade,
respectively, with a range of values of 0–92. The average mathematics IRT
scores were 19.53 (SD = 7.09) in fall kindergarten and 43.42 (SD = 9.13) in
spring first grade with a range of values of 0–64. The reliability of the IRT-
based scores that is based on the variance of repeated estimates of theta (the
overall ability estimate) ranged from .93 to .95 for reading achievement and
from .92 to .94 for mathematics achievement.
Missing Data
As noted previously, only children with complete data were included in the
study. To rule out systematic bias in our conclusions due to missing data,
we compared the final total analytic sample used for our analyses with the
children who missed at least one of the key variables in the analysis and
were thus removed from the analyses.
First, we compared whether there were systematic differences between
the final total sample used in the reading analysis (N= 12,386) and the
missing group excluded from the analysis due to the absence of the motor
assessment or spring first-grade reading scores (N= 358). The missing
group had lower family income, was younger, and was much more likely to
be minority than the study group children, with all the differences statisti-
cally significant. The proportion of Hispanic and Asian children in the two
groups especially contrasted sharply. Hispanic children comprised 16.3%
of the sample group and 59.8% of the missing group, and Asian children
762 Merrill-Palmer Quarterly
050 seung (755-778) 12/1/06 11:06 AM Page 762
comprised 1.9% versus 13.8% of the total sample and missing group,
respectively. These differences might be explained as attrition due to lim-
ited English proficiency. Indeed, 71.1% of the missing group children had a
home language other than English, which was extremely high compared to
8.7% for the final total group.
Second, characteristics of the total sample included in the math analy-
sis (N= 12,579) were compared with those of the missing group excluded
from math analysis due to not having the motor assessment or spring first-
grade math scores (N= 165). The missing group in the math analysis had
characteristics similar to the missing group in reading, but the differences
between groups were not as extreme. More children in the missing group in
math had a home language other than English (40.6%) as compared with
the math sample group children (10.1%). The missing group was also more
likely to be Asian students (30.1%) than the total sample (1.9%). In addi-
tion, missing group children had lower scores on the later reading test.
However, there was no significant difference in age, initial math scores, or
family income.
The final sample used in the analyses was less likely to be minority and
more likely to be from higher SES families, to have higher achievement
scores, and to speak English as a first language than the groups that were
excluded from analyses. While the predictive validity of the motor assess-
ment might be restricted by unintentionally excluding many potential at-
risk children, the missing sample comprised less than 3% of the sample
used in reading analyses and only 1.3% of the math sample.
Results
Correlations between Motor Skills and Cognitive Achievement
Table 1 displays correlations between reading and mathematics scores of
fall kindergarten and spring first grade on the one hand and students’visual
motor and gross motor skills of fall kindergarten on the other. Correlations
revealed that visual motor skills had significantly higher correlations with
cognitive achievement than did gross motor skills (p< .001). Overall, all
correlations with math scores were significantly higher than those with
reading (p< .01), and correlations with spring first-grade achievement were
significantly higher than those with fall kindergarten achievement (p<
.001) except for the comparison of two correlations, one between gross
motor and fall kindergarten math and another between gross motor and
spring first-grade math, which were not statistically significant.
Children’s Motor Skills 763
050 seung (755-778) 12/1/06 11:06 AM Page 763
Hierarchical Regressions of Motor Skills on Later Cognitive
Achievement
Four-step hierarchical regression analyses were performed separately for
reading and mathematics achievement to examine whether predictive
validity of motor assessment is statistically significant even after control-
ling for differences in fall kindergarten achievement and demographic
information including gender, ethnicity, age, home language, and socioeco-
nomic status.4Using spring first-grade achievement as a dependent vari-
able, motor assessment was examined to learn if it added any unique
variance above and beyond demographic variables and children’s initial
competencies. To examine the unique influence of motor skills, the demo-
graphic variables were entered as the first step. Then, children’s initial
achievement and the time gap between fall kindergarten and spring first-
grade assessments were entered.5Finally, visual motor and gross motor
skills were added as the third and the fourth step, respectively. All continu-
ous variables in the regression models were given z-scores before being
entered; thus, unstandardized regression coefficients could be interpreted
as effect sizes. Categorical variables in the regression models were modi-
fied into dummy variables, so their unstandardized coefficients could be
interpreted as effects of group differences. Tables 2 and 3 show unstandard-
764 Merrill-Palmer Quarterly
Table 1. Correlations of Motor Skills and Cognitive Skills
Subtests Visual Motor Skills Gross Motor Skills
Reading
Fall kindergarten .35 .15
Spring first grade .40 .19
Mathematics
Fall kindergarten .44 .20
Spring first grade .48 .22
Note: All significance levels are based on 2-tail tests using a normalized child weight.
All the coefficients were statistically significant at the level of p< .001.
4. Socioeconomic status was represented as the mean of the five standardized variables of
income, parental education, and parental job prestige index scores.
5. Since the assessment timing in fall kindergarten and spring first grade was different across
classes and schools and since an earlier assessment might result in lower achievement scores than a
later assessment, we included a timing variable—the gap between fall kindergarten and spring first-
grade assessments—as another control in order to be conservative in examining the influence of
motor skills after controlling for initial cognitive scores.
050 seung (755-778) 12/1/06 11:06 AM Page 764
ized coefficients of the predictors of spring first-grade reading and mathe-
matics achievement respectively.
In Reading Regression 1, significant associations between visual motor
and reading were found even after controlling for the effects of demograph-
ics and initial reading scores. Inclusion of the variable of gross motor skills
at Step 4 explained an additional 0.3% of the variance, and the predictability
of visual motor skills diminished slightly. When all the predictors were
entered, the model explained 48.1% of the variance in spring first-grade
reading. Results indicated that before entering motor skills and initial read-
ing scores into the analysis, SES was the strongest predictor for reading
among the demographics. However, fall kindergarten reading scores were
the best predictor at Step 2 and thereafter. Here, the model explained 45.1%
of the variance of spring first-grade reading scores by adding an additional
27% variance explained by fall kindergarten reading (p< .001). Since both
the initial and later reading scores were based on the same measure of
ECLS-K direct cognitive assessments, it was expected that the two scores
would be highly correlated and the predictability of motor skills would not
be as strong as that of initial reading. Inclusion of visual motor skills at Step
3 explained an additional 2.8% variance in later reading (p< .001). Although
visual motor skills explained a statistically significant and unique amount of
variance in reading, the effect size was small. This could be due to high cor-
relations among motor skills, fall kindergarten reading, and demographics.
Additional analysis showed that including fall kindergarten reading
scores and visual and gross motor skills as the first three steps in the model
explained almost 46% of variance in spring first-grade reading (Reading
Regression 2). In other words, the combined contribution of motor skills
and reading achievement was significant and had an effect size of .68. In
another model, we tested the combined variance explained by motor skills
and fall kindergarten reading scores in the residuals of first-grade reading
scores, after the effects of demographics variables were partialed out. Here,
the combined effect size was .53. These additional analyses indicate that
motor skills combined with initial reading significantly predicted spring
first-grade reading achievement.
The same procedure was conducted for relating motor skills to spring
first-grade mathematics achievement, after controlling for fall kindergarten
math and demographics. In Table 3, Mathematics Regression 1 indicates
that fall kindergarten math scores were the strongest predictors of first-
grade math at Step 2 and thereafter. At Step 3, the visual motor skills
explained an additional 3.4% of the variance in first-grade math after demo-
graphics and initial math scores explained 50.3% at Step 2. Inclusion of
gross motor skills at Step 4 did not substantially change the model. All the
Children’s Motor Skills 765
050 seung (755-778) 12/1/06 11:06 AM Page 765
Table 2. Hierarchical Regression Analyses of Spring First-Grade Reading Achievement (N= 11,803)
Reading Regression 1 Reading Regression 2
Variable Step 1 Step 2 Step 3 Step 4 Variable Step 1 Step 2 Step 3 Step 4
Female .20*** .11*** .09*** .08*** Fall K reading .65*** .58*** .58*** .53***
Asian .29*** .16** .14** .14** Assessment time gap .07*** .07*** .07*** .07***
Hispanic –.12*** –.04 –.05* –.05* Visual motor skills .20*** .19*** .17***
Black –.27*** –.18*** –.13*** –.14*** Gross motor skills .05*** .06***
Other race –.31*** –.15*** –.16*** –.17*** Female .08***
Age of assessment .15*** .04*** .01 .01 Asian .14**
Home language other –.13*** –.17*** –.20*** –.20*** Hispanic –.05*
than English
SES .32*** .11*** .09*** .09*** Black –.14***
Fall K reading .58*** .53*** .53*** Other race –.17***
Assessment time gap .06*** .07*** .07*** Age of assessment .01
Visual motor skills .18*** .17*** Home language other –.20***
than English
Gross motor skills .06*** SES .09***
R2.18*** .45*** .48*** .48*** R2.42*** .46*** .46*** .48***
R2change .18*** .27*** .03*** .00*** R2change .42*** .04*** .00*** .02***
Note: All regression coefficients are unstandardized b’s.
*p< .05; **p< .01; ***p< .001.
050 seung (755-778) 12/1/06 11:06 AM Page 766
Table 3. Hierarchical Regression Analyses of Spring First-Grade Mathematics Achievement (N= 12,459)
Mathematics Regression 1 Mathematics Regression 2
Variable Step 1 Step 2 Step 3 Step 4 Variable Step 1 Step 2 Step 3 Step 4
Female –.03* –.02~ –.06*** –.07*** Fall K math .67*** .60*** .60*** .54***
Asian .03 –.13** –.15*** –.15*** Assessment time gap .08*** .09*** .09*** .08***
Hispanic –.23*** –.06** –.08*** –.09*** Visual motor skills .22*** .21*** .20***
Black –.49*** –.29*** –.25*** –.27*** Gross motor skills .04*** .06***
Other race –.35*** –.16*** –.17*** –.19*** Female –.07***
Age of assessment .20*** .04*** .01* .01 Asian –.15**
Home language other –.09** .02 –.03 –.03 Hispanic –.09***
than English
SES .32*** .09*** .07*** .07*** Black –.27***
Fall K math .62*** .54*** .54*** Other race –.19***
Assessment time gap .08*** .08*** .08*** Age of assessment .01
Visual motor skills .21*** .20*** Home language other .03
than English
Gross motor skills .06*** SES .07***
R2.23*** .50*** .54*** .54*** R2.48*** .52*** .52*** .54***
R2change .23*** .28*** .03*** .00*** R2change .48*** .04*** .00*** .02***
Note: All regression coefficients are unstandardized b’s.
*p< .05; **p< .01; ***p< .001.
050 seung (755-778) 12/1/06 11:06 AM Page 767
predictors in the model together explained 54% of the variance in spring
first-grade math. With initial math scores and the assessment time-gap con-
trolled, visual motor skills uniquely predicted spring first-grade math. The
effect sizes of visual and gross motor skills for predicting math achieve-
ment were higher than those for reading achievement.
We ran additional hierarchical regression analyses to determine the
extent to which visual motor skills combined with fall kindergarten math
scores can predict spring first-grade math. Inclusion of fall kindergarten
math and visual and gross motor skills as the first three steps in the model
explained almost 52% of variance in first-grade mathematics (Mathematics
Regression 2). This represents an overall effect size of .72. In another
model, we tested the combined variance explained by motor skills and
kindergarten mathematics scores in the residuals of first-grade mathematics
scores after the effects of demographics were partialed out. Here, the com-
bined effect size was .53. It appears that visual motor skills together with
initial mathematics scores predict later mathematics scores with substantial
success. The character of the longitudinal relationship of motor assessment
with cognitive achievement becomes clearer when one appreciates that fall
kindergarten and spring first-grade cognitive achievements are highly cor-
related (r= .65 for reading, r= .69 for mathematics).
Receiver-Operating-Characteristic Curves
To test whether the motor assessment could identify children specifically at risk
or not at risk in spring first-grade reading and math achievement independ-
ently and jointly with fall kindergarten achievement, Receiver-Operating-
Characteristic (ROC) curve analyses were conducted. Since previous
regression analyses demonstrated a higher predictability of visual motor skills
than gross motor skills, we focused only on the visual motor skills.
ROC curve analysis is a component of logistic regression that permits a
computation of sensitivity (i.e., the probability that the motor screening test
is positive given that the person has cognitive difficulties) and specificity
(i.e., the probability that a motor screening test is negative given that the
person does not have cognitive difficulties) at a number of cutoff points in
the motor assessment (Hasselblad & Hedges, 1995; Sackett, Haynes, &
Tugwell, 1985; Toteson & Begg, 1988). It defines an optimal cutoff based
on a favorable ratio of sensitivity and specificity. In other words, the opti-
mal cutoff point will produce a favorable ratio of overreferrals to underre-
ferrals while maximizing correct identifications.
In the analysis of how well visual motor skills identify reading difficul-
ties, we used the visual motor skills as a test variable and spring first-grade
reading achievement as an outcome state variable. Since the reading score
768 Merrill-Palmer Quarterly
050 seung (755-778) 12/1/06 11:06 AM Page 768
was not a dichotomous variable, 1.5 SDs below the mean (the correspon-
ding reading IRT score is approximately 37) was used as a cutoff for con-
sidering children at risk for reading difficulties. Due to the fact that the
distribution of scores was somewhat negatively skewed, 1 SD below the
mean captured only about the lowest 1.4% of children. But 1.5 SDs below
the mean included approximately 7.1% (unweighted N= 874) of the chil-
dren in the database as at risk for reading problems. The results of ROC
curve analysis showed that the area under the ROC curve was .74, repre-
senting the probability of students performing poorly or well on both motor
and reading assessments (Figure 1a). That is, students with poor visual
motor skills who were chosen randomly had a much higher probability of
being ranked lower on reading than students who had at or above average
visual motor skills. ROC analysis showed that motor assessment had a sen-
sitivity of .75 and specificity of .63 for reading achievement when 5.5 is
considered as a cutoff score for visual motor skills problems.
The ROC curve of fall kindergarten reading achievement as a test vari-
able had a similar but higher probability (the area under the curve was .87,
sensitivity was .80, and specificity was .79 (Figure 1b). This suggests that
motor skills and fall kindergarten reading achievement shared considerable
variance.
To test the predictability of motor assessment in addition to cognitive
achievement, additional ROC analysis was performed by combining infor-
mation of motor and kindergarten reading achievement. We used a combined
score of visual motor skills and fall kindergarten reading as a test variable and
spring first-grade reading as an outcome state variable dichotomized at 1.5
SDs below the mean. The combined scores were calculated by adding stan-
dardized scores of visual motor skills and kindergarten reading after each
standardized score was weighted by the relative effect size found in the
regression analysis (the ratio of relative effect sizes of motor to reading is 1:3;
see Table 2). The area under the ROC curve was .88, representing the proba-
bility of students performing poorly or well on both combined scores and
spring first-grade reading (Figure 1c). The probability was much better than
that of independent ROC analysis only with visual motor skills and some-
what better than that of independent ROC analysis only with fall kindergarten
reading. The sensitivity and the specificity of the combined ROC (.80 and
.80, respectively) were higher than those of the independent ROCs (.75 and
.63 for motor and .80 and .79 for initial reading, respectively). These results
demonstrated that visual motor skills tended to classify children in a similar
way as the spring first-grade reading difficulty variable and even more simi-
larly when motor and reading scores were considered simultaneously.
The ROC analysis of mathematics achievement showed similar results.
With the fall kindergarten visual motor skill scores as a test variable and
Children’s Motor Skills 769
050 seung (755-778) 12/1/06 11:06 AM Page 769
spring first-grade mathematics scores as an outcome variable that was
dichotomized at 1.5 SD below the mean (corresponding to a math IRTscore
of about 31) as a cutoff for considering children at risk for mathematics dif-
ficulties (6.87% of the sample; unweighted N= 847), we found that 77.5%
of students were performing poorly or well on both motor and first-grade
mathematics assessments (Figure 2a). With 5.5 as a cutoff point for visual
motor skills, motor assessment had a sensitivity of .78 and a specificity of
.63 for spring first-grade mathematics. The ROC curve of fall kindergarten
mathematics achievement as a test variable had a similar but higher proba-
bility (the area under the curve was .89, sensitivity was .82, and specificity
was .82; see Figure 2b).
To test the unique predictability of motor assessment in addition to the
mathematics assessment, additional ROC analyses were performed with
information of both motor and kindergarten mathematics achievement
combined. As a test variable, we used a combined standardized score of
visual motor skills and fall kindergarten mathematics achievement,
weighted by the relative effect size; the ratio of relative effect sizes of
motor to kindergarten mathematics is also 1:3 (see Table 3). The area under
the ROC curve was .90, which is better than that of the independent ROC
analyses (see Figure 2c). Results showed that motor and kindergarten math-
770 Merrill-Palmer Quarterly
Figure 1. ROC Curve for Spring First-Grade Reading
(a) ROC with Visual Motor Skills
Area under the curve = .74; Sensitivity = .75; Specificity = .63.
(b) ROC with Initial Reading
Area under the curve = .87; Sensitivity = .80; Specificity = .79.
(c) Joint ROC with Visual Motor and Initial Reading
Area under the curve = .88; Sensitivity = .80; Specificity = .80.
050 seung (755-778) 12/1/06 11:06 AM Page 770
ematics jointly had a sensitivity of .82 and specificity of .82 for spring first-
grade mathematics, both of which were higher than those of the independ-
ent ROCs (.78 and .63 for motor and .82 and .82 for initial mathematics,
respectively).
The ROC curve analyses demonstrate that assessment of reading and
math achievement in spring first grade and that of visual motor skills in
kindergarten assigned students to roughly the same categories of risk. If
assessment results of fall kindergarten achievement are added to those of
motor skills, then the combined information has a greater probability of
classifying children accurately.
Discussion
Using a nationally representative sample with more than 12,000 children,
the current study intended to evince the predictive validity of motor assess-
ment, information that was not available in previous small sample studies.
Although relatively few data were missing, the final sample used in our
analyses was heavily weighted toward English-speaking children. Thus,
results from this study can only be generalized with caution to all children
in the nation.
Children’s Motor Skills 771
Figure 2. ROC Curve for Spring First-Grade Mathematics
(a) ROC with Visual Motor Skills
Area under the curve = .78; Sensitivity = .78; Specificity = .63.
(b) ROC with Initial Math
Area under the curve = .89; Sensitivity = .82; Specificity = .82.
(c) Joint ROC with Visual Motor and Initial Math
Area under the curve = .90; Sensitivity = .82; Specificity = .82.
050 seung (755-778) 12/1/06 11:06 AM Page 771
Longitudinal Associations between Motor Skills
and Cognitive Achievement
Overall, our results support the hypothesis that motor skills, specifically
visual motor skills, are related to later cognitive achievement and are able
to successfully identify children at risk for academic underachievement at
the outset of schooling, especially when combined with initial achievement
information. Correlational analyses demonstrated a significant relation
between motor skills and cognitive achievement. Coefficients were similar
to but higher than those reported in previous studies, whose average coeffi-
cient was .30 for visual motor skills. Interestingly, correlations of kinder-
garten motor skills with spring first-grade achievement were significantly
higher than those with fall kindergarten achievement, suggesting longitudi-
nal relations between early motor skills and the kinds of skills represented
in later cognitive achievement.
Unique and Combined Predictability of Motor Skills
for Later Cognitive Achievement
Hierarchical regression analyses that directly examined the predictability
of early kindergarten motor skills for cognitive achievement during early
schooling showed that visual motor skills at kindergarten entry were a
unique, significant predictor of spring first-grade reading and mathematics
achievement. Gross motor skills were also significant, but their effect sizes
were negligible. Although the unique predictability of visual motor skills
was statistically significant, the effect size was not large. The results sug-
gest that while motor skills are highly associated with later cognitive
achievement, combined assessment of visual motor and initial cognitive
achievement can provide better predictability than using motor information
alone as an indicator.
At the time of kindergarten entry, early cognitive and language abilities
such as narrative skills, vocabulary, or phonological awareness may also be
strong predictors of later academic achievement (Catts, Fey, Zhang, &
Tomblin, 1999; Hart & Risley, 1995; Snow, 1985). Unfortunately, the
ECLS-K study does not include any variables concerning initial level of
general cognitive abilities, IQ, or language skills other than early kinder-
garten reading and mathematics achievement that we controlled for in our
analyses. Future studies should consider including assessment of broadly
based cognitive and language development in their model in order to deter-
mine the best predictors of later school achievement.
772 Merrill-Palmer Quarterly
050 seung (755-778) 12/1/06 11:06 AM Page 772
Identification of At-Risk Children Based on Motor Skills
The visual motor assessment also provided reliable classification informa-
tion for children at risk for later academic achievement difficulties, inde-
pendently and jointly with initial cognitive achievement assessments. ROC
curve analyses demonstrated that visual motor screening yielded high sen-
sitivity and specificity for identifying later academic difficulties in both
reading and mathematics, although the unique predictability of motor skills
assessment over and above the predictability of kindergarten cognitive
achievement was not substantial.
The results suggest that academic skills develop partly by interacting
with visual motor skills; in practice, the results imply the use of visual
motor assessment for a kindergarten developmental screening as one ele-
ment of a larger profile or battery (see Meisels with Atkins-Burnett, 1994).
As shown by the regression and ROC results, the effect size of visual motor
skills was not large, although it was statistically significant. This means that
the information from the motor assessment is not strong enough to be used
alone, although motor information adds some predictability to the informa-
tion obtained about initial cognitive achievement.
Implications and Limitations
The current study is not designed to verify the developmental mechanism
through which visual motor skills may influence cognitive achievement. In
fact, a proven causal mechanism has not yet been discovered. The present
study suggests that more than one model of the development of academic
skills is valid. In addition to cognitive models (e.g., phonics approach to read-
ing), there can be other pathways through visual or spatial processing that
may influence achievement. Recent findings from neuropsychology demon-
strate that visual and spatial processing is needed to fulfill complex cognitive
tasks in reading and mathematics (Goel & Dolan, 2000; Zago et al., 2001).
Studies of dyslexia also show that phonological, visual, and motor processing
systems all play a role in this disorder (Stein & Walsh, 1997). These studies
suggest that several specific aspects of visual motor skills (e.g., visual, spa-
tial, or motor processing; visuospatial working memory; motor speed or auto-
maticity; etc.) are related to cognitive achievement.
Moreover, motor skills may influence achievement indirectly through
social and emotional skills. For example, visual motor skills may be associ-
ated with self-regulatory functions that influence academic achievement
(McClelland, Morrison, & Holmes, 2000). Holding a pencil and drawing a
Children’s Motor Skills 773
050 seung (755-778) 12/1/06 11:06 AM Page 773
line, as assessed in the current visual motor scale, may be heavily influenced
by self-control. In addition, gross motor skills can be indirectly related to aca-
demic achievement through peer relations (Gruber, 1985), which may
explain the lower effect size of gross motor skills on cognitive achievement.
Future research may address which aspects of visual motor skills best
predict which subskills in reading or mathematics. Although the current
study was unable to examine the specificity of motor predictability, it may
be the case that specific aspects of motor skills predict different subskills in
reading or mathematics. In addition, future studies may expand the study
by addressing predictability of motor skills for achievement beyond first
grade. The significant relationship between motor skills and cognitive
achievement may be present in other developmental periods of time, con-
sidering the evidence that cognitive achievement of children is stable
across grade levels (Cunningham & Stanovich, 1997). However, the degree
of association between motor skills and achievement can change depending
on the grade levels assessed, and the relationship between motor and cogni-
tive achievement can vary depending on the skills measured. For instance,
one study (Kurdek & Sinclair, 2001) reported a significant relationship
between motor skills and achievement in fourth grade, although the corre-
lation coefficients were smaller (.26 with reading and .35 with mathemat-
ics) than those in the current study of first-graders. Future studies may
employ more in-depth assessment of motor skills and cognitive achieve-
ment, including diverse aspects of motor skills and cognitive processing, to
better understand how particular aspects of visual motor skills are associ-
ated with aspects of ongoing achievement.
Several implications follow from our analyses. First, motor assessment
is associated with later school achievement and can be used as one of the
indicators of future school achievement of young children. This study
showed that motor performance, especially visual motor performance, can
contribute to examining children’s cognitive preparedness for school.
Motor scores are not strong enough to be used in isolation to predict later
achievement or identify at-risk children, compared to the strength of initial
cognitive achievement. But including motor skills in an early school
assessment may increase the predictability of later achievement and the
probability of identifying children at risk for school failure. This may be
even more the case when early assessment or screening occurs prior to
school entry, when adequate assessment of an early cognitive achievement
level is not relatively easy.
Second, our results support the position that developmental domains—
specifically, motor, reading, and mathematics skills—are intercorrelated
aspects of a young child’s developing skills. As recent research on early
774 Merrill-Palmer Quarterly
050 seung (755-778) 12/1/06 11:06 AM Page 774
development suggests, children’s development cannot be separated easily
into disconnected developmental domains (Smith et al., 1999). Rather,
development is a complex process and is a result of interactions among the
domains; thus, it is the whole child that needs our attention (Bowman,
Burns, & Donovan, 2001). Indeed, effective early childhood assessment
tools usually consist of multiple indicators or predictors of development
(Meisels, 1994; Meisels & Atkins-Burnett, 2000). Teaching and research
with young children should be based on an appreciation of this dynamic
and intricate characteristic of development.
References
Bai, D. L., & Bertenthal, B. I. (1992). Locomotor status and the development of
spatial search skills. Child Development, 63, 215–226.
Bertenthal, B. I., Campos, J. J., & Kermoian, R. (1994). An epigenetic perspective
on the development of self-produced locomotion and its consequences. Cur-
rent Directions in Psychological Science, 3(5), 140–145.
Bowman, B. T., Burns, M. S., & Donovan, M. S. (Eds.). (2001). Eager to learn:
Educating our preschoolers. Committee on Early Childhood Pedagogy. Wash-
ington, DC: National Academy Press.
Broadhead, G. D., & Church, G. E. (1988). Contributions of motor, concept, and
communication skills to preschool learning. In J. E. Clark & J. H. Humphrey
(Eds.), Advances in motor development research (Vol. 2, pp. 115–126). New
York: AMS Press.
Bryant, F. B. (2000). Assessing the validity of measurement. In L. G. Grimm &
P. R. Yarnold (Eds.), Reading and understanding more multivariate statistics
(pp. 99–146). Washington, DC: American Psychological Association.
Campos, J. J., Anderson, D. I., Barbu-Roth, M. A., Hubbard, E. M., Hertenstein, M.
J., & Witherington, D. (2000). Travel broadens the mind. Infancy, 1(2),
149–219.
Catts, H. W., Fey, M. E., Zhang, X., & Tomblin, J. B. (1999). Language basis of
reading and reading disabilities: Evidence from a longitudinal investigation.
Scientific Studies of Reading, 3(4), 331–361.
Cunningham, A. E., & Stanovich, K. E. (1997). Early reading acquisition and its
relation to reading experience and ability 10 years later. Developmental Psy-
chology, 33(6), 934–945.
Diamond, A. (2000). Close interrelation of motor development and cognitive
development and of the cerebullum and prefrontal cortex. Child Development,
71(1), 44–56.
Ellis, A. (1985). The cognitive neuropsychology of developmental (and acquired)
dyslexia: A critical survey. Cognitive Neuropsychology, 2, 169–205.
Children’s Motor Skills 775
050 seung (755-778) 12/1/06 11:06 AM Page 775
Ellis, A. (1987). Review of problems in developing cognitively transmitted cogni-
tive modules. Mind and Language, 2, 242–251.
Goel, V., & Dolan, R. J. (2000). Anatomical segregation of component processes in
an inductive inference task. Journal of Cognitive Neuroscience, 12(1),
110–119.
Goldstein, D. J., & Britt, T. W., Jr. (1994). Visual-motor coordination and intelli-
gence as predictors of reading, mathematics, and written language ability. Per-
ceptual and Motor Skills, 78, 819–823.
Gruber, J. J. (1985). Physical activity and self-esteem development in children: A
meta-analysis. Academy Papers, 19, 30–48.
Hambleton, R. K., Swaminathan, H., & Rogers, H. J. (1991). Fundamentals of item
response theory. Newbury Park, CA: Sage.
Hart, B., & Risley, T. R. (1995). Meaningful differences in the everyday experience
of young American children. Baltimore, MD: Paul Brookes.
Hasselblad, V., & Hedges, L. V. (1995). Meta-analysis of screening and diagnostic
tests. Psychological Bulletin, 117(1), 167–178.
Iverson, J. M., & Thelen, E. (1999). Hand, mouth and brain: The dynamic emer-
gence of speech and gesture. Journal of Consciousness Studies, 6(11–12),
19–40.
Kavale, K. (1982). Meta-analysis of the relationship between visual perception
skills and reading achievement. Journal of Learning Disabilities, 15, 42–51.
Kimmel, E. (2001). Review of the Early Screening Inventory-Revised. In B. S.
Plake & J. C. Impara (Eds.), The fourteenth mental measurements yearbook
(pp. 450–451). Lincoln: Buros Institute of Mental Measurements of the Uni-
versity of Nebraska-Lincoln.
Knight, D., & Rizzuto, T. (1993). Relations for children in grades 2, 3, and 4
between balance skills and academic achievement. Perceptual and Motor
Skills, 76, 1296–1298.
Kopp, C. B., & McCall, R. B. (1982). Predicting later mental performance for nor-
mal, at-risk, and handicapped infants. In P. B. Baltes & O. G. Brim (Eds.), Life-
span development and behavior (Vol. 4, pp. 33–61). New York: Academic Press.
Kurdek, L. A., & Sinclair, R. J. (2001). Predicting reading and mathematics
achievement in fourth-grade children from kindergarten readiness scores.
Journal of Educational Psychology, 93(3), 451–455.
Lesiak, J. (1984). The Bender Visual Motor Gestalt Test: Implications for the diag-
nosis and prediction of reading achievement. Journal of School Psychology,
22, 391–405.
Luria, A. R. (1973). The working brain. New York: Basic Books.
McCall, R. B. (1981). Early predictors of later IQ: The search continues. Intelli-
gence, 5, 141–147.
776 Merrill-Palmer Quarterly
050 seung (755-778) 12/1/06 11:06 AM Page 776
McClelland, M. M., Morrison, F. J., & Holmes, D. L. (2000). Children at risk for
early academic problems: The role of learning-related social skills. Early
Childhood Research Quarterly, 15(3), 307–329.
Meisels, S. J. (1989). Can developmental screening tests identify children who are
developmentally at risk? Pediatrics, 83, 578–585.
Meisels, S. J. (1994). Designing meaningful measurements for early childhood. In
B. L. Mallory & R. S. New (Eds.), Diversity and developmentally appropriate
practices: Challenges for early childhood education (pp. 202–222). New
York: Teachers College Press.
Meisels, S. J., with Atkins-Burnett, S. (1994). Developmental screening in early
childhood: A guide (4th ed.). Washington, DC: National Association for the
Education of Young Children.
Meisels, S. J., & Atkins-Burnett, S. (2000). The elements of early childhood assess-
ment. In J. P. Shonkoff & S. J. Meisels (Eds.), Handbook of early childhood
intervention (2nd ed., pp. 231–257). New York: Cambridge University Press.
Meisels, S. J., Marsden, D. B., Wiske, M. S., & Henderson, L. W. (1997). Early
Screening Inventory-Revised. New York: Pearson Early Learning.
Messick, S. (1989). Validity. In R. L. Linn (Ed.), Educational measurement (3rd
ed., pp. 13–103). New York: Macmillan.
National Center for Education Statistics. (2001). ECLS-K base year public-use
data files and electronic codebook. Washington, DC: National Center for Edu-
cation Statistics.
National Center for Education Statistics. (2002). User’s manual for the ECLS-K
first grade public-use data files and electronic code book. Washington, DC:
National Center for Education Statistics.
Paget, K. D. (2001). Review of the Early Screening Inventory-Revised. In B. S.
Plake & J. C. Impara (Eds.), The fourteenth mental measurements yearbook
(pp. 451–453). Lincoln: Buros Institute of Mental Measurements of the Uni-
versity of Nebraska-Lincoln.
Payne, V. G., & Isaacs, L. D. (1999). Human motor development: A lifespan
approach (4th ed.). Mountain View, CA: Mayfield.
Piaget, J. (1952). The origins of intelligence in children. New York: International
Universities Press.
Rosenbaum, D. A., Carlson, R. A., & Gilmore, R. O. (2001). Acquisition of intel-
lectual and perceptual-motor skills. Annual Review of Psychology, 52,
453–470.
Sackett, D. L., Haynes, R. B., & Tugwell, P. (1985). Clinical epidemiology: A basic
science for clinical medicine. Boston: Little, Brown.
Satz, P., & Fletcher, J. M. (1988). Early identification of learning disabled children:
An old problem revisited. Journal of Consulting and Clinical Psychology,
56(6), 824–829.
Children’s Motor Skills 777
050 seung (755-778) 12/1/06 11:06 AM Page 777
Schmidt, S., & Perino, J. (1985). Kindergarten screening results as predictors of
academic achievement, potential, and placement in second grade. Psychology
in the Schools, 22(2), 146–151.
Silvia, P. A., & Ross, B. (1980). Gross motor development and delays in develop-
ment in early childhood: Assessing significance. Journal of Human Movement
Studies, 6, 211–226.
Smith, L. B., Thelen, E., Titzer, R., & McLin, D. (1999). Knowing in the context of
acting: The task dynamics of the A-not-B error. Psychological Review, 106(2),
235–260.
Snow, C. E. (1985). Literacy and language relationships during the preschool years.
Harvard Educational Review, 53(2), 165–189.
Stein, J., & Walsh, V. (1997). To see but not to read: The magnocellular theory of
dyslexia. Trends in Neuroscience, 20(4), 147–151.
Toteson, A. N. A., & Begg, C. B. (1988). A general regression methodology for
ROC curve estimation. Medical Decision Making, 8, 204–215.
Tramontana, M. C., Hooper, S. R., & Selzer, S. C. (1988). Research on the pre-
school prediction of later academic achievement: A review. Developmental
Review, 8, 89–146.
Zago, L., Pesenti, M., Mellet, M., Crivello, F., Mazoyer, B., & Tzourio-Mazoyer,
N. (2001). Neural correlates of simple and complex mental calculation. Neu-
roImage, 13, 314–327.
778 Merrill-Palmer Quarterly
050 seung (755-778) 12/1/06 11:06 AM Page 778