Genetic influences on early word recognition
abilities and disabilities: a study of 7-year-old
Nicole Harlaar,1Frank M. Spinath,2Philip S. Dale,3and Robert Plomin1
1Institute of Psychiatry, UK;2University of Bielefeld, Germany;3University of Missouri-Columbia, USA
Background: A fundamental issue for child psychology concerns the origins of individual differences
in early reading development. Method: A measure of word recognition, the Test of Word Reading
Efficiency (TOWRE), was administered by telephone to a representative population sample of 3,909
same-sex and opposite-sex pairs of 7-year-old twins. Analyses allowing for sex differences in aetiology
were used to estimate the extent to which genetic and environmental influences contribute to normal
variation in word recognition and word recognition difficulties, defined by scores below the 5th and 10th
percentiles of the unselected sample. Results: Both normal variation in word recognition and impaired
word recognition abilities were substantially heritable (h2¼ .65–.67; h2
influences were primarily shared between twins, rather than specific to each individual, and small to
moderate in magnitude. There was evidence for qualitative sex differences. Quantitative sex differences
were also suggested at the extremes, with genetic influences being more important as a cause of reading
difficulties in boys than in girls. Conclusions: These findings indicate that early individual differences
and impairments in word recognition are principally influenced by genetic factors and may involve
partly distinct genetic or environmental effects in boys and girls. Crucially, they also provide evidence
that reading impairments are linked genetically to the normal distribution. Genetic risk for early
impairments in word recognition is continuous rather than discrete. Keywords: Individual differences,
reading difficulties, word recognition, genetics, twins, sex differences. Abbreviations: DZF: dizygotic
female; DZM: dizygotic male; DZO: dizygotic opposite-sex; MZF: monozygotic female; MZM: monozygotic
male; TEDS: Twins Early Development Study; TOWRE: Test of Word Reading Efficiency.
g¼ .37–.72). Environmental
Individual differences in reading ability are pervas-
ive, emerge early in development, and remain rel-
atively stable over time (Cunningham & Stanovich,
1997; Juel, 1988; Shaywitz et al., 1999). There has
been an increasing appreciation that genetic influ-
ences contribute to the emergence of these differ-
ences. Family and twin studies have provided
converging evidence that reading difficulties are
familial and substantially due to genetic factors (for
reviews, see Grigorenko, 2001; Raskind, 2001;
Schulte-Ko ¨rne, 2001). Normal variations in reading
ability are also known to be due, in part, to genetic
factors, with heritability (h2) estimates ranging from
18% to 87% across diverse measures of reading (e.g.,
Gaya ´n & Olson, 2003; Hohnen & Stevenson, 1999;
Petrill & Thompson, 1994; Reynolds et al., 1996;
1987). These findings have provided a firm basis for
molecular genetic attempts to approximately localise
genetic loci that may influence reading development
(for reviews, see Fisher & DeFries, 2002; Smith,
Kelley, & Brower, 1998).
Given the finding that both normal variations in
reading ability and reading difficulties are substan-
tially heritable, a question that arises is whether
genetic and environmental risk factors act across the
dimension of reading ability or alternatively only
influence reading disability. There is considerable
evidence to suggest that reading disability can be
considered dimensionally at a phenotypic level, and
this would be compatible with the hypothesis that
genetic (or environmental) risk factors for reading
difficulties are also continuous (Gilger, Borecki,
Smith, DeFries, & Pennington, 1996). Even when a
trait is normally distributed, however, this does not
exclude the possibility that the lower extreme of the
distribution may be aetiologically distinct. This
situation is illustrated by height. There is evidence
that height within the normal range is influenced by
multiple genetic and environmental factors acting
dimensionally, whereas extremes of height stature
are often influenced by specific pathological (and
often genetic) disorders (e.g., Silventoinen, 2003;
normal variation in reading ability similarly due to
distinct aetiological factors?
In the present twin study, we examined the links
between reading abilities and disabilities using
DeFries–Fulker (DF) extremes analysis (DeFries &
Fulker, 1985, 1988). This method addresses the
issue of why some children fall at the extreme of a
dimension, such as reading, as compared to the rest
of the population. If the mean difference between
probands and the unselected population (the pro-
band group deficit) on a quantitative measure of
reading is due to genetic factors, the co-twins of DZ
probands should regress further toward the popu-
lation mean, on average, than MZ co-twins. The
Journal of Child Psychology and Psychiatry 46:4 (2005), pp 373–384 doi: 10.1111/j.1469-7610.2004.00358.x
? Association for Child Psychology and Psychiatry, 2004.
Published by Blackwell Publishing, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA
magnitude of differential regression provides the
basic information for estimating group heritability
g), the extent to which the mean difference between
the proband group and the unselected population is
attributable to genetic factors. Crucially, evidence for
significant group heritability indicates that genetic
factors mediate the relationship between proband
status (i.e., low reading scores) and normal quanti-
tative variation in reading ability. As such, DF ex-
tremes analysis represents a way of determining, in
terms of aetiology, whether extreme scores on a
quantitative measure of reading reflect a qualitative
departure from the normal range of ability.
To date, DF extremes analysis has been employed
in just two independent twin studies of reading
ability and disability. In a sample of 285 pairs of
13-year-old twins, only proband group deficits on a
measure of homophone recognition were consist-
ently found to be significantly heritable (Stevenson,
1991). In contrast, recent analyses of data from over
500 twin pairs aged 8 to 20 years in the Colorado
Learning Disabilities Research Center (CLDRC) twin
project (DeFries et al., 1997) found that genetic fac-
tors accounted for between 50% and 85% of group
deficits across a broad range of component reading
skills (e.g., Gaya ´n & Olson, 2001), as well as to
combined factor scores based on a combination of
continuous measures (e.g., Davis et al., 2001). A
recent study indicated that high reading ability was
also related to normal variation in reading ability
(Boada et al., 2002). These findings suggest that
both normal variations and extremes of reading
ability are substantially heritable and are linked
genetically, pointing against a strict hypothesis of
Much of the current behavioural genetic research
on reading abilities and disabilities pertains to a
broad age range, from middle childhood through
adolescence. There has been little systematic re-
search into the reading development of young read-
ers (Olson & Gaya ´n, 2001). In this paper we examine
the aetiologies of impaired and non-impaired reading
ability, as assessed by a measure of word recogni-
tion, in a large, representative population-based
sample of 7-year-old twins in England and Wales.
Twins were beginning readers, having received, on
average, just over one year of formal reading in-
struction at the time of assessment. To date, the only
twin study that has assessed the reading abilities of
children at this developmental stage is a relatively
small population sample of 126 twin pairs, which
reported a heritability of 60% for individual differ-
ences in literacy among 7-year-old twins (Hohnen &
Stevenson, 1999). No information is currently avail-
able on the aetiology of early reading deficits. The
paucity of research on children in the early years of
elementary school is surprising, given the high and
increasing demands for literacy. The need for such
research is heightened by evidence for the persist-
ence of early reading difficulties (e.g., Jacobson,
1999; Shaywitz et al., 1999), and the deleterious
motivational, academic, and social consequences
that may ensue from early reading failure (e.g.,
Maughan, 1995; Stanovich, 1986).
The present study also includes a focus on
assessing possible sex differences in the aetiology of
reading abilities and difficulties. Both clinical and
epidemiological studies have provided evidence that
reading difficulties are more prevalent in boys than
girls (e.g., Flannery, Liederman, Daly, & Shultz,
2000; Meltzer, Gatward, Goodman, & Ford, 2000;
St. Sauver, Katusic,
Jacobsen, 2001). Boys also tend to show lower levels
of reading achievement than girls in unselected,
school-age populations (e.g., Hedges & Nowell, 1995;
Although evidence for mean sex differences does not
imply that sex differences exist in the underlying
causes of individual differences in reading ability,
these findings have nevertheless alerted researchers
to the possibility that gender may contribute to het-
erogeneity in the aetiologies of individual differences
and deficits in early reading ability. Two hypotheses
may be tested from a behavioural genetic perspect-
ive. First, the same genetic and environmental
influences may affect phenotypic variation in both
sexes, but the impact may be greater in one sex than
the other (quantitative sex differences). A second
possibility is that boys and girls may be influenced
by differentgenetic and
(qualitative sex differences, or sex-limitation effects).
To date, the evidence for sex differences in the
aetiology of reading ability is equivocal, both at the
level of individual differences and proband group
deficits. Virtually all possible outcomes have been
reported: higher heritability for boys (Knopik, Alar-
co ´n, & DeFries, 1998), higher heritability for girls
(Petrill & Thompson, 1994), no quantitative sex dif-
ferences (DeFries, Gillis, & Wadsworth, 1993; Gaya ´n
& Olson, 2003; Wadsworth, Knopik, & DeFries,
2000), qualitative sex differences (Alarco ´n, DeFries,
& Fulker, 1995; Knopik et al., 1998), and no quan-
titative or qualitative sex differences (Reynolds et al.,
1996; Stevenson, 1992). These conflicting results
might reflect methodological differences, such as the
age of participants and measures used. Further-
more, although most of the studies reviewed had
samples of more than 500 twin pairs, the power to
detect significant sex differences is low even with
these sample sizes (Eaves et al., 1997). The present
study used a large sample of same- and opposite-sex
twins, thereby providing the opportunity to investig-
ate both quantitative and qualitative sex differences
in the aetiology of individual differences and group
deficits in reading ability.
In summary, there is a need for further research
into the aetiology of early reading difficulties and
reading abilities within the normal range, with cla-
rification of possible sex effects. Three related issues
were addressed in the present study. The first
Nicole Harlaar et al.
concerns the extent to which normal variation in
early reading ability is attributable to genetic and
whether genetic and environmental factors influence
group deficits in reading ability. Based on previous
studies, we hypothesised that both heritability and
group heritability would be substantial. Finally, we
examined the possibility that the aetiologies of
reading difficulties and normal variation of reading
ability differ for boys and girls. Given the inconsist-
ency of previous studies, we had no specific hypo-
However, the large sample and the focus on a single
age allow for a more rigorous test of sex differences
than previous studies.
The sample was drawn from the 1994 and 1995 cohorts
of the Twins Early Development Study (TEDS), a longi-
tudinal, population-based study of twins born in
England and Wales (for details, see Trouton, Spinath, &
Plomin, 2002). A total of 11,350 families identified by
the UK Office for National Statistics (ONS) as having
twins born in these years were contacted to take part in
TEDS when the twins were approximately 12 months
old. Informed consent was obtained from 10,932
(96.3%) families, of whom 9,442 subsequently com-
pleted a booklet requesting background family infor-
mation. Assessments of cognitive and behavioural
development have been carried out since the twins’
second birthday. The present study focuses on 4,338
families who returned the background booklet and
whose children took part in the 7-year assessment. The
mean age of the sample at the time of testing was
7.07 ± .22 years.
We assessed the representativeness of the sample by
comparing demographic characteristics of families who
participated in the 7-year assessments with UK popu-
lation data from the ONS (2002). As shown in Table 1,
children in the 7-year TEDS sample were slightly
more likely to be white and to have a mother who was
educated to at least A-level standard (analogous to
completing high-school in a university-oriented curri-
culum). We similarly evaluated the effects of sample
attrition by comparing demographic characteristics of
the families who participated in the 7-year assessment
with all 1994 and 1995 families originally enrolled in
TEDS and those that took part in the 4-year assess-
ments. As shown in Table 1, participation in the 7-year
assessment was associated with white ethnicity, higher
maternal education, and higher scores on an index of
socio-economic status (SES) (for details, see Petrill,
Pike, Price, & Plomin, submitted). Owing to the large
sample size, these differences were significant. How-
ever, when bias was tested by re-weighting available
cases to represent the demographic characteristics of
the entire sample and repeating all analyses, the results
were unaffected. Overall, these comparisons suggest
representative of the UK population in terms of ethnicity
and maternal education and additionally showed no
important deviations from TEDS families in the original
sample and those who completed the assessments at
A parent-rating instrument validated by polymorphic
DNA markers was used to determine the zygosity of
same-sex twin pairs with an error rate of £5% (Price
et al., 2000). Unclear cases were resolved through DNA
screening. Definitive zygosity diagnosis of a small group
of same-sex twins awaits genotyping, and these twins
were excluded from the analyses reported here (n ¼ 31
pairs). Twin pairs were also excluded if English was not
the first language spoken in the home, if either twin had
a hearing problem, specific medical or genetic condition
(e.g., cerebral palsy, Down’s Syndrome, chromosomal
abnormality), or was an outlier for birth weight, time
spent in hospital, special care after birth, gestational
age, or maternal alcohol consumption during preg-
nancy. We found no significant differences between the
original target sample and the sample after these
measure-specific exclusions in terms of the demogra-
phic indices summarised in Table 1. The final sample of
3909 families consisted of 1396 MZ twin pairs, 1298
same-sex DZ twin pairs, and 1215 opposite-sex DZ
(DZO) twin pairs.
Measures and procedure
We assessed reading ability by telephone using the Test
of Word Reading Efficiency (TOWRE; Torgesen, Wagner,
& Rashotte, 1999), Form B. The TOWRE comprises two
timed sub-tests: Sight Word Efficiency (SWE), which
assesses the ability to read aloud real words, and
Phonemic Decoding Efficiency (PDE), which assesses
the ability to read aloud pronounceable printed non-
words (e.g., ‘framble’, ‘tegwop’). Each subtest consists of
a list of graded words printed on a single page. The child
is given 45 seconds to read as many items as possible.
TOWRE composite scores, obtained by standardising
and summing the sub-test scores, were used in our
Table 1 Comparison of demographic characteristics of UK
population and TEDS families (1994 and 1995 cohorts)
TEDS 4 yr3
TEDS 7 yr4
1Data from the General Household Survey (GHS: Office for
National Statistics, 2002).
2TEDS families who completed the background booklet
(N ¼ 9442).
3TEDS families who took part in the 4-year assessment (N ¼
4TEDS families who took part in the 7-year assessment (N ¼
5‘A-levels’ in the UK involve staying on in the UK equivalent of
high school and successfully completing college entrance
6z-scores on SES scale derived from socio-demographic infor-
mation in background booklet: fathers’ and mothers’ highest
educational level, fathers’ and mothers’ occupational status,
and age of mother at birth of eldest child.
Genetic influences on early reading
The item lists were mailed to families in a sealed
package prior to the test sessions, with separate
instructions that they should not be opened until the
time of testing. Twins in each pair were tested sepa-
rately within the same test session and by the same
tester, who was blind to zygosity classification. Parents
were asked to ensure that twins were not in the same
room during the test session. Although no information
is available on the correspondence between telephone
administration and standard in-person administration
of the TOWRE, we have shown that the SWE and PDE
subtests of the TOWRE correlate substantially (r ¼ .83;
Dale et al., submitted; r ¼ .82 in the present study),
replicating previous findings based on in-person test-
ing. Furthermore, a recent follow-up telephone assess-
ment of 54 twin pairs showed that the correlation
between TOWRE Form B at 7 years and TOWRE Form A
at 9 years was r ¼ .83. This finding is consistent with
previous research demonstrating the longitudinal sta-
bility of word-level reading skills (Juel, 1988; Torgesen
& Burgess, 1998), and can be seen as a lower-limit
estimate of reliability.
Although the use of a single test is a limitation of the
present study, there is strong evidence that accurate
and fluent word recognition is central to the reading
process (Share & Stanovich, 1995), and correlates
strongly with the ultimate goal of reading, comprehen-
sion of text (Shankweiler et al., 1999). Moreover, diffi-
culties in word recognition have been identified as the
proximal impediment to reading in reading-disabled
children (Adams & Bruck, 1993).
Individual differences analysis
Intraclass correlations (Shrout & Fleiss, 1979) for age-
and sex-corrected TOWRE scores (McGue & Bouchard,
1984) were compared by zygosity to obtain an initial
approximation of the aetiology of individual differences
in TOWRE scores. MZ twins share all of their genes
identical-by-descent, whereas DZ twins share, on
average, 50% of their segregating genes. Thus, inde-
pendent of sex, the extent to which the MZ correlation
exceeds the DZ correlation indexes the contribution of
additive genetic influences to within-pair resemblance.
Heritability (h2), the proportion of phenotypic variance
that can be ascribed to heritable genetic influences, can
be estimated as twice the MZ–DZ correlation difference.
Shared environmental influences, referring to all envir-
onmental influences that make siblings more similar to
one another, can be estimated by subtracting the estim-
ate of heritability from the MZ correlation. The variance
that remains is ascribed to unique (non-shared) envi-
ronmental influences specific to each twin within a pair
and measurement error.
When correlations are compared by sex as well as
zygosity, it is possible to assess sex differences in aeti-
ology. Quantitative sex differences in the magnitude of
genetic and environmental influences are suggested
when the difference between the MZ and DZ correla-
tions in male twin pairs differs from the corresponding
difference between the MZ and DZ correlations in
female twin pairs. Qualitative sex differences are
suggested when the correlation for dizygotic opposite-
sex (DZO) twins is significantly less than the correla-
tions of same-sex DZ pairs, based on the assumption
that genetic or environmental influences on reading
difficulties that are specific to one sex will reduce
within-pair similarity. It should be noted that regress-
ing out the mean effects of sex from TOWRE scores has
no bearing on these analyses, which are concerned with
differences in genetic and environmental variance
between the sexes rather than phenotypic mean differ-
To formally test the observations derived from the
intraclass correlations, variance-covariance matrices
for each of the five zygosity-sex group were analysed in
a series of models using the structural equation pro-
gram Mx (Neale, Boker, Xie, & Maes, 2002). These
models are based on the standard univariate twin
model, modified with the inclusion of DZO twin pairs
(Neale & Maes, 1999). This model is shown in Figure 1
for DZO twins. Variance in TOWRE scores was modelled
as a function of latent additive genetic (A), shared
environmental (C), and non-shared environmental (E)
influences. Within same-sex twin pairs, the correlation
between additive genetic influences (rA) on Twin 1 and
Twin 2 was fixed at 1.0 for MZ and 0.5 for DZ twin pairs.
The correlation between shared environmental influ-
ences (rC) was fixed at 1.0 for both zygosity groups.
Within DZO pairs, in contrast, rAand rCmay be less
than the expected values of 0.5 and 1.0, respectively, if
there are significant sex-specific genetic or environ-
mental influences. It should be noted that it is not
possible to estimate rAand rCsimultaneously, hence it
cannot be ascertained whether qualitative sex differ-
ences are genetic or environmental in origin.
As a test of qualitative sex differences, we compared
models in which either rAor rCbetween DZO twins was
freely estimated (full sex-limitation model) with models
in which the rAand rCbetween DZO twins were fixed at
the expected values of. 5 and 1.0, respectively (common
effects model). Fitting a further model that constrained
all parameter estimates to be equal for males and
females (null model) provided a test of quantitative sex
differences. Because these alternative models are hier-
archically related, the relative fit of each alternative
model was determined using standard v2difference
tests, with degrees of freedom (df ) equal to the differ-
ence in degrees of freedom between the two models
(Neale & Maes, 1999). More parsimonious models were
Figure 1 Full sex-limitation model in DZO twin pairs
Nicole Harlaar et al.
considered preferable unless a significant deterioration
in fit was observed. Proportions of variance attributable
to additive genetic, shared environment, and non-
shared environment were estimated from the best-
equality of shared environmental effects across MZ and
DZ twin pairs, the absence of assortative mating, and
independence and additivity of the A, C, and E compo-
nents (see Boomsma, Busjahn, & Peltonen, 2002;
Martin, Boomsma, & Machin, 1997
McClearn, & McGuffin, 2001).
; Plomin, DeFries,
The 10th and 5th centiles of the distribution of stand-
ardised TOWRE scores were used to select children with
word recognition impairments, with the 5th centile
representing a more severe degree of impairment. These
are arbitrary but commonly used cut-off points for
identifying reading disability (Olson, 1999), corres-
ponding to scores 1.33 and 1.63 standard deviations,
respectively, below the mean in the unselected sample
in the present study. The selection of low-scoring pro-
bands was based on TOWRE scores uncorrected for
gender to allow representative gender proportions in the
residual scores were used in the analyses. In view of the
substantial evidence that IQ information adds little to
the external validity of the identification of reading dif-
ficulties, we did not reference low reading scores to
either a discrepancy between achievement and IQ test
scores or to a minimum IQ criterion (see Fletcher et al.,
1998; Siegel, 1989). Because IQ-based definitions of
reading difficulties remain widely used, however, the
analyses described below were also conducted sepa-
rately when reading difficulties were identified on the
basis of a regression discrepancy method defined in
Reynolds (1985) and Fletcher et al. (1994). Virtually
identical results were obtained, hence the findings
based on the low achievement-only criterion are
For the purpose of the genetic analyses, each twin
was entered once as the proband and once as a co-twin.
This double-entry procedure facilitates the most valid
estimates of h2
gwhen a sample has been ascertained
using truncate selection (Thompson & Thompson,
1986). To obtain a preliminary impression of the aeti-
ology of reading difficulties, we compared probandwise
concordance rates for the five zygosity-sex groups. The
probandwise rate directly estimates risk to the co-twin
of an affected proband, and is the most appropriate
concordance rate under truncate selection (McGue,
1992). Because concordance rates are insensitive to the
variability of reading scores within and between affected
vs. unaffected groups, we also compared the trans-
formed co-twin mean scores for MZ and DZ probands.
The transformed scores were obtained by dividing the
standardised scores for the selected MZ and DZ twin
samples by the difference between the zygosity- and
sex-specific proband and population means (for details,
see Viding et al., 2004). This transformation results in a
proband mean of 1.0, as compared with the unselected
population mean of 0, and ensures that the MZ and DZ
probands are equally divergent from the mean proband
scores. The transformed co-twin means will fall between
Age- and sex-corrected
these values and are indicative of the extent to which
MZ and DZ co-twins regress toward the mean of the
unselected population. As noted in the introduction, DZ
co-twins are expected to regress further toward the
mean of the unselected population than are MZ co-
twins to the extent that the proband group deficit is due
to genetic influence.
The differential regression of co-twin means in MZ
and DZ twin pairs was tested more formally using a
model-fitting extension of the DF regression equation
that allows DZO twins to be incorporated within a sex-
limitation model (Purcell & Sham, 2003). Similar to the
individual difference analyses, the fit of a full sex-lim-
itation model was compared with that of nested com-
mon effects and null models in order to test for
aetiological sex differences in proband group deficits.
Components of variance were estimated from the
results of the best-fitting model.
Individual differences analysis
Table 2 presents means and intraclass twin corre-
lations for standardised TOWRE scores, uncorrected
for the mean effects of sex and age, for the five zyg-
osity-sex groups in the full sample. There was a
significant mean sex difference, with males scoring
lower than females. However, the effect size was
negligible (g2¼ .002), indicating that the overlap
in the distribution of TOWRE scores for males and
females was substantial.
Intraclass correlations were consistently higher for
MZ twins, with DZ twin correlations differing from
MZ correlations by .53–.59 units. This pattern indic-
ates that the variance in each measure is partly
attributable to both additive genetic and shared
environmental influences. The importance of non-
shared environmental influences is indicated by the
discrepancy between the MZ correlations and unity.
The MZ–DZ correlation difference was identical for
males and females in same-sex twin pairs. However,
correlations for same-sex DZ pairs were greater than
those for DZO pairs, implicating some qualitative sex
differences in genetic or environmental effects.
Model-fitting analyses confirmed the results sug-
gested by the intraclass correlations. There was a
significant deterioration in fit from the full sex-limi-
tation model to the common effects model, indicating
that the genetic correlation (rA) or shared environ-
ment correlation (rC) between males and females
Table 2 Means and intraclass (ICC) correlations for stand-
ardised TOWRE scores
Genetic influences on early reading
could not be fixed in the DZO group. Figure 2 pre-
sents the parameter estimates derived from the full
model for the genetic and environmental compon-
ents of variance for males and females and the
coefficients of genetic and shared environmental
relatedness for DZO twins.
parameter estimates indicate that additive genetic
factors accounted for approximately two-thirds of
variance in TOWRE scores in both sexes (.65 for
boys, .67 for girls). Shared and non-shared envir-
onmental factors each accounted for about one-fifth
of the variance and were also comparable for boys
and girls (.19 and. 17, respectively). The confidence
intervals around the genetic and shared environ-
mental estimates did not overlap, indicating that
additive genetic factors explained significantly more
individual variance than did shared environmental
Genetic influences and environmental influences
on individual differences in reading performance
were of similar magnitude for boys and girls, provid-
ing little evidence for quantitative sex differences.
However, the estimate of rA(.38) between boys and
girls in DZO pairs was less than .50, and the estim-
ate of rC(.57) was less than 1.0. Thus, the model-
fitting results support the conclusion that genetic or
shared environmental influences differ somewhat for
boys and girls, as suggested by the lower intraclass
correlation for DZO twins compared with same-sex
DZ twins (Table 2).
Table 3 presents standardised proband and co-twin
means, probandwise concordances,
formed co-twin means for each of the groups entered
into the extremes analyses. Similar to the finding for
significant mean sex differences in reading ability
across the normal range, proband mean scores were
significantly lower for males than females; however,
the effect size was neglible (g2¼ .015). The ratio of
male to female probands was 1.3:1 when the 10%
cut-off was used as a threshold to determine reading
difficulties, and 1.4:1 when the more stringent 5%
cut-off was applied. These findings indicate that a
marginally greater proportion of boys had reading
Probandwise concordances for twins in MZ pairs
were uniformly higher than those for twins in DZ
pairs, indicating that the risk that co-twins of pro-
bands are themselves affected is partly genetically
influenced. The difference between MZ and DZ con-
cordance rates was comparable for males and fe-
males at the 10% cut-off, but greater for males at the
5% cut-off. This finding indicates that the degree to
which early reading difficulties are influenced by
genetic factors may differ by sex, at least when more
severe cut-offs are used to define reading difficulties.
At both cut-offs, the DZO concordance rate was
somewhat lower than the concordances for males
and females in same-sex DZ twin pairs. This finding
indicates that the risk for reading difficulties in the
co-twins of probands in DZO pairs is less than that of
Variance in TOWRE scores
Figure 2 Individual differences: Additive genetic (A),
Shared environmental (C) and Non-shared environ-
mental (E) influences on variance in TOWRE scores,
with 95% confidence intervals (below figure). Estimates
are from full sex-limitation model, where DZO rA¼ .38
(.28–.47) and DZO rC¼ .57 (.30–.84)
Table 3 Standardardised proband and co-twin means, probandwise concordances and transformed co-twin means for 10th and
P(bar) (SD) C(bar) (SD)PCTco(bar) (SD)
P(bar) (SD)C(bar) (SD)PC Tco(bar) (SD)
Note: PC ¼ probandwise concordances, P(bar) ¼ proband mean, TCo(bar) ¼ transformed co-twin mean, N ¼ number of probands.
P(bar) and Tco(bar) are based on age- and sex-regressed scores.
Nicole Harlaar et al.
co-twins of probands in same-sex DZ pairs, and
implicates qualitative sex differences in the aetiology
of reading difficulties.
The pattern of transformed MZ and DZ co-twin
means closely mirrors the probandwise concordance
rates. At both cut-offs, the DZ co-twin means regress
further toward the population mean than the MZ co-
twin means, indicating that the quantitative trait
score difference between probands and the popula-
tion was partly due to additive genetic factors. In
addition, both quantitative and qualitative sex dif-
ferences in aetiology were also suggested. The dif-
ference between MZ and DZ co-twin means was
slightly greater for males than in females. In addi-
tion, co-twins of opposite-sex DZ probands regressed
further to the population mean than co-twins of
same-sex DZ probands.
Analyses based on the DF model incorporating
DZO twins corroborated the trends suggested by the
concordance rates and transformed co-twin means.
At the 10% cut-off, maximum-likelihood testing
yielded a significant reduction in fit from the full
model to the common effects model, suggestive of
significant qualitative sex differences. At the 5% cut-
off, the common effects model provided a more par-
simonious fit over the full model, but the null model
fit significantly worse than the common effects
model, indicating that parameter estimates could
not be equated for boys and girls. Overall, these
findings provide evidence for significant qualitative
sex differences at the 10% cut-off and significant
quantitative sex differences at the 5% cut-off.
Figure 3 shows the parameter estimates for the DF
model-fitting analyses. The results provide evidence
for substantial group heritability estimates that were
somewhat greater for boys than girls at the 10% cut-
off (.67 and .50, respectively) and considerably
greater for boys at the more stringent 5% cut (.72
and .37, respectively). Group shared environment
effects were correspondingly higher for girls (.40 for
girls and .21 for boys at the 10% cut-off; .37 for girls
and .16 for boys at the 5% cut-off). The trend toward
lower heritability and greater shared environmental
effects on deficits in reading at the lowest 5%, al-
though perhaps indicative of the possibility that
environmental factors play a larger role in severe
reading impairment, was not significant. Non-shared
environmental effects accounted for approximately
10% of the group deficits at each cut-off and for both
sexes. Congruent with the findings from the indi-
vidual differences analyses, the genetic and shared
environmental correlations between boys and girls
suggested qualitative sex differences in the aetiology
of reading deficits.
Because this is the first report of a DF extremes
analysis for reading difficulties using DZO twins, we
re-ran all the analyses separately for male and fe-
male twin pairs in the MZ and same-sex DZ samples
using the DF multiple regression model (DeFries &
Fulker, 1985, 1988). For the purpose of these ana-
lyses, standard errors and significance levels were
corrected in order to take into account the artificial
inflation of sample size (e.g., Stevenson, Pennington,
Gilger, DeFries, & Gillis, 1993). The results, shown
Boys GirlsBoys Girls
Variance in TOWRE scores
Figure 3 Proband group deficits: Additive genetic (A), Shared environmental (C) and Non-shared environmental (E)
influences on proband group deficits (10% and 5% cut-offs) in TOWRE scores, with 95% confidence intervals (below
figure). Estimates for 10% cut-off are from full sex-limitation model, where DZO rA¼ .23 (.00–.41) and DZO rC¼ .52
(.24–.98). Estimates for 5% cut-off are from common effects model, where DZO rA¼ .50 and rC¼ 1.0
Genetic influences on early reading
in Table 4, closely match those obtained from the
model-fitting extension of the DF regression equa-
tion. For example, genetic influences on group defi-
cits were substantial and higher for boys than girls,
particularly at the 5% cut-off. Overall, these findings
provide converging evidence that early reading diffi-
culties are substantially heritable.
The present study, the first to investigate the aetiol-
ogy of both individual differences and deficits in
reading ability in a large community sample of chil-
dren in the early school years, yielded three main
findings. First, individual differences in TOWRE
scores were substantially heritable. Additive genetic
influences accounted for about two-thirds of indi-
vidual differences in TOWRE performance in both
boys and girls. Lower but statistically significant
estimates of shared environmental influences were
also obtained, accounting for approximately one-
fifth of the variance in word reading ability. These
findings are consistent with the results obtained for
the sample of 6- to 7-year-old twins reported by
Hohnen and Stevenson (1999), as well as previous
twin studies of children in older or mixed age groups
(e.g., Davis et al., 2001; Reynolds et al., 1996), in
showing that individual differences in reading ability
are primarily due to genetic influences rather than
shared environmental factors.
The second finding is that a similar pattern
emerged for deficits in reading ability for both 10%
and 5% cut-offs. Results from the DF extremes
analysis incorporating DZO twins indicated that
additive genetic factors accounted for up to 72% of
the mean difference between 7-year-old children
with reading difficulties and the unselected sample.
Shared environmental factors, in contrast, accoun-
ted for 21–50% of group deficits. These findings
corroborate, with a younger and larger sample,
similar results from the CLDRC study (DeFries &
Alarco ´n, 1996; Gaya ´n & Olson, 2001). The con-
vergence of findings indicates that the group deficits
in word recognition in the early school years are at
least as heritable as group deficits in later childhood
A third set of findings concerns the issue of aetio-
logical sex differences. Evidence for sex-specific in-
environmental origin, was found for individual
differences across the full sample. The present study
also provides the evidence that deficits in reading
ability may include sex-specific influences. The pat-
tern of probandwise concordances and transformed
co-twin means were indicative of qualitative sex dif-
ferences on deficits in TOWRE scores, although the
results of the DF model incorporating DZO twins
indicated that these were only significant at the 10%
cut-off. In addition, although there were no quanti-
tative sex differences for the individual differences
analysis, for both 10% and 5% cut-offs, heritability
was greater for boys and shared environment was
greater for girls, with the differences reaching signi-
ficance for the 5% cut-off. Overall, these findings
point to the conclusion that heterogeneity in the
aetiology of individual differences and group deficits
in reading ability may occur as a function of gender,
at least among children in the beginning years of
The most important finding of the present study
is the significant group heritability at both the 5th
and 10th percentiles, indicating that the aetiologies
of word recognition difficulties and normal variation
in word recognition are not qualitatively distinct. In
this regard, the present
implications for understanding the genetic basis of
reading abilities and disabilities. Instead of a single
‘disease’ gene, which is rare in the population and
which is by itself necessary and sufficient to pro-
duce reading difficulties, the results suggest that
genetic influences on reading may be best concep-
tualised as multiple susceptibility loci, so-called
quantitative trait loci (QTLs), that modify the distri-
bution of reading abilities in the population at large
and create a continuum of genetic risk for reading
difficulties (Pennington, 1999; Plomin, Owen, &
McGuffin, 1994). The strongest test of these hypo-
theses will be to determine, via molecular and
functional genomic research, whether specific gen-
etic loci associated with reading difficulties also
contribute to normal variation in reading (Plomin,
DeFries, Craig, & McGuffin, 2003). The findings
from the sex-limitation analyses in the current
study further suggest the possibility that some
genetic loci conferring susceptibility for reading
difficulties may have effects specific to one sex, and,
perhaps, that some genetic factors associated with
severe reading difficulties will show stronger effects
in boys than in girls.
Several limitations of this study should be noted.
First, assessment was restricted to a single measure
of reading skill, namely word recognition. Although
word recognition is widely recognised as a crucial
component of both reading ability and reading dis-
ability, reading encompasses a broad range of cog-
nitive processes. The importance of considering the
multifaceted nature of reading has been highlighted
by evidence for at least partial independence in the
aetiologies of individual differences and deficits in
different component reading skills underlying word
Table 4 Group genetic and shared environment estimates
with 95% CI from DF regression model (DeFries & Fulker,
10% .68 (.50–.86) .21 (.03–.39) .50 (.32–.68) .37 (.19–.55)
5% .60 (.37–.83) .26 (.08–.44) .40 (.14–.66) .47 (.19–.55)
Nicole Harlaar et al.
recognition (see Fisher & DeFries, 2002), as well as
by evidence for differential genetic aetiology across
subgroups of poor readers who differ in their reading
profiles (Castles, Datta, Gayan, & Olson, 1999). The
use of multiple measures of reading and related
skills is needed in future studies of the aetiology of
early reading difficulties.
A second limitation concerns the assessment of
reading difficulties. Two issues may be considered.
First, although studies have shown that the TOWRE
is a useful and valid tool for identifying children
with poor reading difficulties (e.g., Raskind, Hsu,
Berninger, Thomson, & Wijsman, 2000; Torgesen,
Wagner, Rashotte, Alexander, & Conway, 1997), the
extent to which the TOWRE discriminates at lower
levels of ability has not been empirically tested.
Second, although a large body of evidence bolsters
the ascertainment of individuals with reading diffi-
culties regardless of overall cognitive abilities, there
is some evidence that environmental influences may
be more salient as a cause of reading difficulties in
children with lower IQ scores (Knopik et al., 2002;
Olson, Datta, Gayan, & DeFries, 1999; Wadsworth,
Olson, Pennington,& DeFries,
research on these issues is warranted.
A third caveat concerns the reliance on a twin
sample. Twinning is associated with an elevated risk
for reading difficulties (e.g., Johnston, Prior, & Hay,
1984; Hay, O’Brien, Johnston, & Prior, 1984; Watts
& Lytton, 1981), which may limit generalisability of
the current findings to singletons (i.e., to the great
majority of people). In order to address this issue
empirically, we are currently collecting similar data
on the younger non-twin siblings of twins in TEDS,
whose genetic background and early familial envir-
onments closely match those of the twins. This
design will make it possible to compare results for
twins and singletons, both in terms of mean and
Finally, it is important to emphasize that the bal-
ance of genetic and environmental influences on
reading abilities and deficits in any one population
will be influenced by the amount of genetic and
environmental variance within the population. Thus,
the evidence for genetic influences on reading diffi-
culty in the present study does not imply genetic
influences on reading differences between groups
within a population for which there are confounding
environmental differences (Olson, 1999). Such group
differences may include English as a second lan-
guage, which was an exclusionary criterion in the
present study. There may also be broader environ-
mental differences between populations. In this
context, it is notable that the majority of children
attending schools in the UK follow a nationwide lit-
eracy programme as part of the UK National Curri-
estimates that are higher than those found in popu-
lations where school literacy environments are more
to produce heritability
Despite these limitations, the results of the pres-
ent study demonstrate in a large and representative
community sample of twins that individual differ-
ences and deficits in reading ability in early child-
hood are substantially heritable and that there are
strong genetic links between ability and disability.
They also suggest some sex-related heterogeneity
in the aetiology of individual differences and defi-
cits in reading. Overall, these findings set a context
for further genetically informative research on typ-
ical and atypical reading development during the
formative school years and beyond.
The authors are indebted to the families in TEDS for
making this study possible. TEDS is supported by a
programme grant from the UK Medical Research
Council to the Institute of Psychiatry, King’s College,
Nicole Harlaar, Social, Genetic, & Developmental
Psychiatry (SGDP) Centre,
Denmark Hill, London, SE5 8AF, UK; Tel: +44 (0)
207848 0600; Fax: +44 (0) 207848 0092; Email:
Box Number PO83,
Adams, M.J., & Bruck, M. (1993). Word recognition:
The interface of educational policies and scientific
research. Reading and Writing: An Interdisciplinary
Journal, 5, 113–139.
Alarco ´n, M., DeFries, J.C., & Fulker, D.W. (1995).
Etiology of individual differences in reading perform-
ance: A test of sex-limitation. Behavior Genetics, 25,
Boada, R., Willcutt, E.G., Tunick, R.A., Chhabildas,
N.A., Olson, R.K., DeFries, J.C., & Pennington, B.F.
(2002). A twin study of the etiology of high reading
ability. Reading and Writing: An Interdisciplinary
Journal, 15, 683–707.
Boomsma, D., Busjahn, A., & Peltonen, L. (2002).
Classical twin studies and beyond. Nature Reviews
Neuroscience, 3, 872–882.
Castles, A., Datta, H., Gaya ´n, J., & Olson, R.K. (1999).
Varieties of developmental reading disorder: Genetic
and environmental influence. Journal of Experimental
Child Psychology, 72, 73–94.
Cunningham, A.E., & Stanovich, K.E. (1997). Early
reading acquisition and its relation to reading experi-
ence and ability ten years later. Developmental
Psychology, 33, 934–945.
Dale, P.S., Harlaar, N., & Plomin, R. (submitted). A
comparison of two nonconventional procedures for
the research assessment of reading skills in 7-year-
Genetic influences on early reading
Davis, C.J., Gaya ´n, J., Knopik, V.S., Smith, S.D.,
Cardon, L.R., Pennington, B.F., Olson, R.K., &
DeFries, J.C. (2001). Etiology of reading difficulties
and rapid naming: The Colorado Twin Study of
Reading Disability. Behavior Genetics, 31, 625–635.
DeFries, J.C. & Alarco ´n, M. (1996). Genetics of specific
reading disability. Mental Retardation Developmental
Disabilities Research Review, 2, 39–47.
DeFries, J.C., Filipek, P.A., Fulker, D.W., Olson, R.K.,
Pennington, B.F., Smith, S.D., & Wise, B.W. (1997).
Colorado Learning Disabilities
Learning Disabilities: A Multidisciplinary Journal, 8,
DeFries, J.C., & Fulker, D.W. (1985). Multiple regres-
sion analysis of twin data. Behavior Genetics, 15,
DeFries, J.C., & Fulker, D.W. (1988). Multiple regres-
sion analysis of twin data: Etiology of deviant scores
versus individual differences. Acta Geneticae Medicae
et Gemmellologiae: Twin Research, 37, 205–216.
DeFries, J.C., Gillis, J.J., & Wadsworth, S.J. (1993).
Genes and genders: A twin study of reading disability.
In A.M. Galaburda (Ed.), Dyslexia and development:
(pp. 187–204). Cambridge, MA: Harvard University
Eaves, L.J., Silberg, J.L., Meyer, J.M., Maes, H.H.,
Simonoff, E., Pickles, A., Rutter, M, Reynolds, C.A.,
Heath, A.C., Truett, K.R., Neale, M.C., Erikson, M.T.,
Loeber, R., & Hewitt, J.K. (1997). Genetics and
developmental pathology: 2. The main effects of genes
and environment on behavioral problems in the
Virginia Twin Study of Adolescent Behavioral Devel-
opment. Journal of Child Psychology and Psychiatry,
Fisher, S.E., & DeFries, J.C. (2002). Developmental
dyslexia: Genetic dissection of a complex cognitive
trait. Nature Reviews: Neuroscience, 3, 767–782.
Flannery, K.A., Liederman, J., Daly, L., & Schultz, J.
(2000). Male prevalence for reading disability is found
in a large sample of black and white children free from
ascertainment bias. Journal of the International
Neuropsychological Society, 6, 433–442.
Fletcher, J.M., Francis, D.J., Shaywitz, S.E., Lyon, G.R.,
Foorman, B.R., Stuebing, K.K., & Shaywitz, B.A.
(1998). Intelligent testing and the discrepancy model
for children with learning disabilities. Learning Dis-
abilities Research and Practice, 13, 186–203.
Liberman, I.Y., Stuebing, K.K., Francis, D.J., Fowler,
A.E., & Shaywitz, B.A. (1994). Cognitive profiles of
reading disability: Comparisons of discrepancy and
low achievement definitions. Journal of Educational
Psychology, 86, 6–23.
Gaya ´n, J., & Olson, R.K. (2001). Genetic and environ-
mental influences on orthographic and phonological
skills in children with reading disabilities. Develop-
mental Neuropsychology, 20, 483–507.
Gaya ´n, J., & Olson, R.K. (2003). Genetic and environ-
mental influences on individual differences in printed
word recognition. Journal of Experimental Child
Psychology, 84, 97–123.
Gilger, J.W., Borecki, I.B., Smith, S.D., DeFries, J.C., &
Pennington, B.F. (1996). The etiology of extreme
scores for complex phenotypes: An illustration using
reading performance. In C. Chase, G. Rosen, & G.
Sherman (Eds.), Developmental dyslexia: Neural,
Baltimore, MD: York Press.
Grigorenko, E.L. (2001). Developmental dyslexia: An
update on genes, brains, and environments. Journal
of Child Psychology and Psychiatry, 42, 91–125.
Hay, D.A., O’Brien, P.J., Johnston, C.J., & Prior, M.
(1984). The high incidence of reading disability in
twin boys and its implications for genetic analyses.
Acta Geneticae Medicae et Gemmellologiae: Twin
Research, 33, 223–236.
Hedges, L.V., & Nowell, A. (1995). Sex differences in
mental test scores, variability and numbers of high
scoring individuals. Science, 269, 41–45.
Hohnen, B., & Stevenson, J. (1999). The structure of
genetic influences on general cognitive, language,
phonological, and reading abilities. Developmental
Psychology, 35, 590–603.
Jacobson, C. (1999). How persistent is reading disabil-
ity? Individual growth curves in reading. Dyslexia, 5,
Johnston, C., Prior, M., & Hay, D. (1984). Prediction of
reading disability in twin boys. Developmental Medi-
cine and Child Neurology, 26, 588–595.
study of 54 children from first through fourth grades.
Journal of Educational Psychology, 80, 437–447.
Knopik, V.S., Alarco ´n, M., & DeFries, J.C. (1998).
Common and specific gender influences on individual
differences in reading performance: A twin study.
Personality and Individual Differences, 25, 269–277.
Knopik, V.S., Smith, S.D., Cardon, L., Pennington, B.,
Gaya ´n, J., Olson, R.K., & DeFries, J.C. (2002).
Differential genetic etiology of reading component
processes as a function of IQ. Behavior Genetics, 32,
Martin, N., Boomsma, D.I., & Machin, G. (1997). A twin-
pronged attack on complex traits. Nature Genetics,
Maughan, B. (1995). Annotation: Long-term outcomes
of developmental reading problems. Journal of Child
Psychology and Psychiatry, 36, 357–371.
Meltzer, H., Gatward, R., Goodman, R., & Ford, T.
(2000). The mental health of children and adolescents
in Great Britain. London: The Stationery Office.
McGue, M.K. (1992). When assessing twin concor-
dance, use the probandwise and not the pairwise
rate. Schizophrenia Bulletin, 18, 171–176.
McGue, M., & Bouchard, T.J. (1984). Adjustment of
twin data for the effects of age and sex. Behavior
Genetics, 14, 325–343.
Neale, M.C., Boker, S.M., Xie, G. & Maes, H.M. (2002).
Mx: Statistical modeling (6th edn). VCU Box 900126,
Richmond, VA 23298: Department of Psychiatry.
Neale, M.C. & Maes, H.H.M. (1999). Methodology for
genetic studies of twins and families. Dordrecht:
Kluwer Academic Publishers.
Office for National Statistics (ONS). (2002). Living in
Britain: Results from the General Household Survey
2000. London: The Stationery Office.
Olson, R.K. (1999). Genes, environment, and reading
disabilities. In R.J. Sternberg & L. Spear-Swerling
(Eds.), Perspectives on learning disabilities (pp. 3–21).
Boulder, CO: Westview Press.
Nicole Harlaar et al.
Olson, R.K., Datta, H., Gayan, J., & DeFries, J.C.
(1999). A behavioral-genetic analysis of reading
disabilities and component processes. In R.M. Klein
& P.A. McMullen (Eds.), Converging methods for
understanding reading and dyslexia (pp. 133–151).
Cambridge, MA: The MIT Press.
Olson, R.K., & Gaya ´n, J. (2001). Brains, genes, and
environment in reading development. In S.B. Neuman
& D.K. Dickinson (Eds.), Handbook of early literacy
research (pp. 81–94). New York: Guilford Press.
Pennington, B.F. (1999). Toward an integrated under-
standing of dyslexia: Genetic, neurological, and
cognitive mechanisms. Development and Psycho-
pathology, 11, 629–654.
Petrill, S.A., Pike, A., Price, T., & Plomin, R. (submitted).
Chaos in the home and socioeconomic status are
associated with cognitive development in early child-
hood: Environmental risks identified in a genetic
Petrill, S.A., & Thompson, L.A. (1994). The effect of
gender upon heritability and common environmental
estimates in measures of scholastic achievement.
Personality and Individual Differences, 16, 631–640.
Phillips, L.M., Norris, S.P., Osmond, W.C., & Maynard,
A.M. (2002). Relative reading achievement: A long-
itudinal study of 187 children from first through sixth
grades. Journal of Educational Psychology, 94, 3–13.
Plomin, R., DeFries, J.C., Craig, I.W., & McGuffin, P.
(2003). Behavioral genetics. In R. Plomin, J.C.
DeFries, I.W. Craig, & P. McGuffin (Eds.), Behavioral
genetics in the postgenomic era (pp. 3–15). Washing-
ton, DC: American Psychological Association.
(2001). Behavioral Genetics (4th edn). New York:
Plomin, R., Owen, M.J., & McGuffin, P. (1994). The
genetic basis of complex human behaviors. Science,
Price, T.S., Freeman, B., Craig, I., Petrill, S.A., Ebersole,
L., & Plomin, R. (2000). Infant zygosity can be
assigned by parental report questionnaire data. Twin
Research, 3, 129–133.
Purcell, S., & Sham, P.C. (2003). A model-fitting
implementation of the DeFries–Fulker model for
selected twin data. Behavior Genetics, 33, 271–278.
Raskind, W.H. (2001). Current understanding of the
genetic basis of reading and spelling disability.
Learning Disability Quarterly, 24, 141–157.
Raskind, W.H., Hsu, L., Berninger, V.W., Thomson,
J.B., & Wijsman, E.M. (2000). Familial aggregation of
dyslexia phenotypes. Behavior Genetics, 5, 385–396.
Reynolds, C.A., Hewitt, J.K., Erickson, M.T., Silberg,
J.L., Rutter, M., Simonoff, E., Meyer, J., & Eaves, L.J.
(1996). The genetics of children’s oral reading per-
formance. Journal of Child Psychology and Psychi-
atry, 37, 425–434.
Reynolds, C.R. (1985). Measuring the aptitude–achieve-
ment discrepancy in learning disability diagnosis.
Remedial and Special Education, 6, 37–55.
Schulte-Ko ¨rne, G. (2001). Annotation: Genetics of read-
ing and spelling disorder. Journal of Child Psychology
and Psychiatry, 42, 985–997.
Shankweiler, D., Lundquist, E., Katz, L., Stuebing, K.K.,
Fletcher, J.M., Brady, S., Fowler, A., Dreyer, L.G.,
Marchione, K.E., Shaywitz, S.E., & Shaywitz, B.A.
(1999). Comprehension and decoding: Patterns of
association in children with reading difficulties.
Scientific Studies of Reading, 3, 69–94.
Share, D.L., & Stanovich, K.E. (1995). Cognitive pro-
cesses in early reading development: Accommodating
individual differences into a model of acquisition.
Issues in Education: Contributions from Educational
Psychology, 1, 1–57.
Shaywitz, S.E., Fletcher,
Shneider, A.E., Marchione, K.E., Stuebing, K.K.,
Francis, D.J., Pugh, K.R., & Shaywitz, B.A. (1999).
Persistence of dyslexia: The Connecticut Longitudinal
Study at adolescence. Pediatrics, 194, 1351–1359.
Shrout, P.E., & Fleiss, J.L. (1979). Intraclass correla-
tion: Use in assessing rater reliability. Psychological
Bulletin, 86, 420.
Siegel, L.S. (1989). IQ is irrelevant to the definition of
learning disabilities. Journal of Learning Disabilities,
Silventoinen, K. (2003). Determinants of variation in
adult body height. Journal of Biosocial Science, 35,
Smith, S.D., Kelley, P.M., & Brower, A.M. (1998).
Molecular approaches to the genetic analysis of speci-
fic reading disability. Human Biology, 70, 239–256.
St. Sauver, J.L., Katusic, S.K., Bararesi, W.J., Colligan,
R.C., & Jacobsen, S.J. (2001). Boy/girl differences in
risk for reading disability: Potential clues? American
Journal of Epidemiology, 154, 787–794.
Stanovich, K.E. (1986). Matthew effects in reading:
Some consequences of individual differences in the
acquisition of literacy. Reading Research Quarterly,
Stevenson, J. (1991). Which aspects of processing text
mediate genetic effects? Reading and Writing: An
Interdisciplinary Journal, 3, 249–269.
Stevenson, J. (1992). Identifying sex differences in
reading disability: Lessons from a twin study. Read-
ing and Writing: An Interdisciplinary Journal, 4, 307–
(1987). A twin study of genetic influences on reading
and spelling ability and disability. Journal of Child
Psychology and Psychiatry, 28, 229–247.
Stevenson, J., Pennington, B.F., Gilger, J.W., DeFries,
J.C., & Gillis, J.J. (1993). Hyperactivity and spelling
disability: Testing for shared genetic aetiology. Jour-
nal of Child Psychology and Psychiatry, 34, 1137–
Thompson, J.S., & Thompson, M.W. (1986). Genetics in
medicine. Philadelphia: W. B. Saunders.
Torgesen, J.K., & Burgess, S.R. (1998). Consistency of
reading-related processes throughout early child-
hood: Evidence from longitudinal-correlational and
instructional studies. In J. Metsala & L. Ehri (Eds.),
Word recognition in beginning reading (pp. 161–188).
Hillsdale, NJ: Erlbaum.
Torgesen, J.K., Wagner, R.K., & Rashotte, C.A. (1999).
Test of Word Reading Efficiency (TOWRE). Austin, TX:
Torgesen, J.K., Wagner, R.K., Rashotte, C.A., Alexan-
der, A.W., & Conway, T. (1997). Preventive and
remedial interventions for children with severe read-
ing disabilities. Learning Disabilities: An Interdisci-
plinary Journal, 8, 51–62.
Genetic influences on early reading
Trouton, A., Spinath, F.M., & Plomin, R. (2002). The Download full-text
Twins Early Development Study (TEDS): A multi-
variate longitudinal genetic investigation of language,
cognition, and problem behaviour in childhood. Twin
Research, 5, 544–548.
Viding, E., Spinath, F.M., Price, T.S., Bishop, D.V.M.,
environmental influence on language impairment
on 4-year-old same sex and opposite sex twins.
Journal of Child Psychology and Psychiatry, 45,
Wadsworth, S.J., Knopik, V.S., & DeFries, J.C. (2000).
Reading disability in boys and girls: No evidence for a
differential genetic etiology. Reading and Writing: An
Interdisciplinary Journal, 13, 133–145.
Wadsworth, S.J., Olson, R.K., Pennington, B.F., &
DeFries, J.C. (2000). Differential genetic etiology of
reading disability as a function of IQ. Journal of
Learning Disabilities, 33, 192–199.
Wajnrajch, M.P. (2002). Genetic disorders of human
growth. Journal of Pediatric Endocrinology and Meta-
bolism, 15(Suppl. 2), 701–714.
Watts, D., & Lytton, H. (1981). Twinship as handicap:
fact or fiction? In L. Gedda, P. Parisi, & P.W.E. Nance
(Eds.), Twin Research 3: Part B. Intelligence, person-
ality and development (pp. 283–266). New York: Alan
Manuscript accepted 19 January 2004
Nicole Harlaar et al.