ArticlePDF Available

Abstract and Figures

On average, students attending selective schools outperform their non-selective counterparts in national exams. These differences are often attributed to value added by the school, as well as factors schools use to select pupils, including ability, achievement and, in cases where schools charge tuition fees or are located in affluent areas, socioeconomic status. However, the possible role of DNA differences between students of different schools types has not yet been considered. We used a UK-representative sample of 4814 genotyped students to investigate exam performance at age 16 and genetic differences between students in three school types: state-funded, non-selective schools (‘non-selective’), state-funded, selective schools (‘grammar’) and private schools, which are selective (‘private’). We created a genome-wide polygenic score (GPS) derived from a genome-wide association study of years of education (EduYears). We found substantial mean genetic differences between students of different school types: students in non- selective schools had lower EduYears GPS compared to those in grammar (d = 0.41) and private schools (d = 0.37). Three times as many students in the top EduYears GPS decile went to a selective school compared to the bottom decile. These results were mirrored in the exam differences between school types. However, once we controlled for factors involved in pupil selection, there were no significant genetic differences between school types, and the variance in exam scores at age 16 explained by school type dropped from 7% to <1%. These results show that genetic and exam differences between school types are primarily due to the heritable characteristics involved in pupil admission.
No caption available
No caption available
Content may be subject to copyright.
Differences in exam performance between pupils attending
selective and non-selective schools mirror the genetic
differences between them
Emily Smith-Woolley
, Jean-Baptiste Pingault
, Saskia Selzam
, Kaili Rimfeld
, Eva Krapohl
, Sophie von Stumm
, Kathryn Asbury
Philip S. Dale
, Toby Young
, Rebecca Allen
, Yulia Kovas
and Robert Plomin
On average, students attending selective schools outperform their non-selective counterparts in national exams. These differences
are often attributed to value added by the school, as well as factors schools use to select pupils, including ability, achievement and,
in cases where schools charge tuition fees or are located in afuent areas, socioeconomic status. However, the possible role of DNA
differences between students of different schools types has not yet been considered. We used a UK-representative sample of 4814
genotyped students to investigate exam performance at age 16 and genetic differences between students in three school types:
state-funded, non-selective schools (non-selective), state-funded, selective schools (grammar) and private schools, which are
selective (private). We created a genome-wide polygenic score (GPS) derived from a genome-wide association study of years of
education (EduYears). We found substantial mean genetic differences between students of different school types: students in non-
selective schools had lower EduYears GPS compared to those in grammar (d=0.41) and private schools (d=0.37). Three times as
many students in the top EduYears GPS decile went to a selective school compared to the bottom decile. These results were
mirrored in the exam differences between school types. However, once we controlled for factors involved in pupil selection, there
were no signicant genetic differences between school types, and the variance in exam scores at age 16 explained by school type
dropped from 7% to <1%. These results show that genetic and exam differences between school types are primarily due to the
heritable characteristics involved in pupil admission.
npj Science of Learning (2018) 3:3 ; doi:10.1038/s41539-018-0019-8
Achievement at the end of full-time compulsory education
represents a major tipping point in life, opening up avenues for
higher education, including university and beyond. Therefore,
understanding the potential predictors of academic achievement
at this juncture is of great importance. One such predictor that has
been hotly debated is school type. In England, when students
transition from primary to secondary school at age 11, they have
the option of attending one of three school types. Ninety-three
percent of children attend state-funded schools, the majority of
which are non-selective
(state non-selective). A small proportion
of state-funded schools (163 schools out of 3113 schools in
England) are academically selective grammarschools. These
schools select their intake based on achievement and ability,
assessed by an entrance exam. The remainder of students
(approximately 7%), are private educated. As well as being fee-
paying, private schools are often also academically selective.
These school types are assumed to set children on different
trajectories, with research linking selective schools (grammar and
private schools) to later success, including higher levels of
academic achievement, acceptance at university, and even higher
earning potential compared to pupils educated in non-selective
However, by design, selective schools are able to choose their
student intake based on certain pupil characteristics. This can
include selection on ability or achievement on an entrance test;
both of which have been shown to correlate positively with life
outcomes, including later academic achievement.
by virtue of being fee-paying, entrance into private schools is
usually dependent on whether the family can afford it (their
socioeconomic status (SES)), which also correlates with future
Even for state schools, family SES may play a role in
what school type a student attends, with grammar schools
typically located in more afuent areas and attracting higher SES
students on average.
It is, therefore, possible that improved
outcomes for pupils in selective schools do not necessarily reect
a higher quality of education, but may simply be the consequence
of selectioneither active, as in the case of ability or achievement,
or passive, as in the case of family SES.
Given the considerable fees charged by private schools, in
addition to the potential stress of selective school entrance exams,
Received: 18 July 2017 Revised: 17 November 2017 Accepted: 9 February 2018
Kings College London, MRC Social, Genetic and Developmental Psychiatry Centre, Institute of Psychiatry, Psychology & Neuroscience, London SE5 8AF, UK;
Clinical, Educational
& Health Psychology, Division of Psychology & Language Sciences, Faculty of Brain Sciences, University College London, 26 Bedford Way, London WC1H 0DS, UK;
London School
of Economics and Political Science, Houghton Street, London WC2A 2AE, UK;
Department of Education, Psychology in Education Research Centre, University of York, York YO10
5DD, UK;
Department of Speech and Hearing Sciences, University of New Mexico, Albuquerque, NM, USA;
New Schools Network, 3 Albert Embankment, London SE1 7SP, UK;
Education Datalab, 1st Floor, 11 Tufton Street, London SW1P 3QB, UK;
Laboratory for Cognitive Investigations and Behavioural Genetics, Tomsk State University, Lenin Avenue,
36, Tomsk Oblast, Tomsk 634050, Russia and
Department of Psychology, Goldsmiths, University of London, 8 Lewisham Way, London SE14 6NW, UK
Correspondence: Emily Smith-Woolley (
Published in partnership with The University of Queensland
why do families choose these schools? Among the many possible
reasons is superior academic achievement. The nding that pupils
at selective schools outperform their non-selective school counter-
parts in exams has been frequently reported.
At age 16,
students in the UK typically take the General Certicate of
Secondary Education (GCSE) exams. The UK Department for
Education shows that 99% of grammar school students obtain top
GCSE grades (A*C grade) in English and mathematics, compared
to 64% for all state-funded mainstream school students.
However, academic achievement at age 16 is positively correlated
with the factors involved in pupil selection, such as prior
achievement, ability and SES.
Therefore, this raises the
questionare selective schools adding anything over and above
these factors in the prediction of academic achievement?
Several studies have attempted to elucidate the effect of school
type on achievement over and above factors on which schools can
select (for example,
for a review, see Coe et al.).
many of these have not been published in peer-reviewed journals
for example
and we are not aware of a recent peer-
reviewed study looking at all three school types: state non-
selective, grammar and private schools in the UK. However, the
non-peer-reviewed reports support the conclusion that there are
only small academic advantages to attending a selective school,
after student factors such as achievement, ability and family SES
have been taken into account.
Traditionally, the relationship between the factors involved in
school admission and later achievement have been thought to
operate environmentally. For example, parents with higher SES
may invest more time in their childrens education
and can
afford more resources (e.g., more books or private tuition), which
in turn may lead to better opportunities and improved achieve-
ment. However, a less frequently investigated factor inuencing
both selection factors, as well as achievement, is genetics. In the
example above, parents with higher SES are not only passing on
educationally relevant environments, but they are also passing on
educationally relevant genes, a concept referred to as gene-
environment correlation (rGE).
A vast literature from quantitative genetics has shown that
genetic factors explain a substantial amount of variance in
selection factors, including ability and achievement.
ability estimates of general cognitive ability (g) from twin studies
range from around 30% in childhood, to 4050% in adolescence
and approximately 60% in adulthood.
Twin studies also show
that much of the relationship between selection factors, such as g,
and later achievement, are substantially inuenced by genetics.
Because twins typically grow up in the same family, the aetiology
of traits such as family SES, which do not vary between twins,
cannot be estimated in this way. However, heritability can be
estimated by genome-wide complex trait analysis (GCTA),
which uses DNA from unrelated individuals to estimate the
proportion of phenotypic variance explained by hundreds of
thousands of single-nucleotide polymorphisms (SNPs) genotyped
on DNA arrays. This method has also shown that genetics
accounts for a signicant amount of individual differences in
family SES,
as well as gand achievement.
School type, like SES, does not tend to vary within twin pairs.
However, because GCTA requires large sample sizes, it has so far
not been possible to look at the genetic differences between
students of different school types. However, powerful genome-
wide association (GWA) studies of behavioural traits, which test
associations between specic SNPs and traits are starting to make
this possible. Although individually these SNPs, identied through
GWA studies, are of small effect, by summing their effects
together it is possible to create a genetic score for each individual
in an independent sample, which explains a substantial propor-
tion of the genetic variation.
These scores, dubbed genome-
wide polygenic scores(GPS) are a game-changer for genetic
research and have already proved insightful within the area of
educational achievement. For example, a recent study
using a
GPS derived from a 2016 GWA study of years of education
has shown educational achievement scores at age
16 differ as a function of GPS. There was approximately one
standard deviation difference between those in the highest GPS
septile and those in the lowest; representing almost a whole
school grade difference. Furthermore, while 65% of students in
the highest GPS septile went on to university, only 37% in the
lowest septile progressed to university-level education.
For the rst time, we assess differences in a polygenic score for
years of education (EduYears) between students from three school
types: non-selective, grammar and private schools. We predict that
selection involving heritable traits such as achievement, ability
and family SES will be reected in the genetic differences between
students of different school types. Furthermore, in line with
previous literature, we expect that selection will also create large
achievement differences between students attending the three
school types, which will reduce substantially once controlling for
the selection factors.
Polygenic score differences between school types
Students attending different school types (state non-selective,
grammar and private schools) differed genetically, as shown by
their mean EduYears GPS (see Fig. 1, analysis of variance (ANOVA)
details in Table S1). Non-selective state school students had
signicantly lower EduYears GPS scores compared to grammar
school students (t=4.87, p< 0.001) and private school students (t
=7.17, p< 0.001). These differences translate to more than a third
of a standard deviation difference (d=0.41 and 0.37, respectively).
There were no signicant mean differences in EduYears GPS scores
between grammar and private school students (t=0.44, p=0.66).
There were also no signicant mean differences between state
non-selective schools in varying selectivity areas (see Table S2 and
Supplementary Fig. S1).
Associations between EduYears GPS and selection factors
EduYears GPS was positively correlated with each of the selection
factors (see Supplementary Table S3), explaining 2.1% of the
variance in ability, 5.2% in achievement and 6.6% in family SES.
EduYears GPS was also positively correlated with GCSE, explaining
7.6% of the variance in GCSE scores, similar to previous analysis of
these data.
Because selective schools actively select for
achievement and ability and passively select for SES, all of which
correlate with EduYears GPS, we tested whether mean differences
in EduYears GPS remained once controlling for these factors.
We found that, after accounting for the variance explained by
heritable selection factors, there were no signicant EduYears GPS
differences between students of the three school types: state non-
selective, grammar and private (see Supplementary Fig. S2 and
Supplementary Table S4). Similar results also emerged when we
looked at differences between state non-selective schools in
varying selectivity areas (see Supplementary Table S5 and
Supplementary Fig. S3), showing small differences in EduYears
between school types.
GCSE differences
Supplementary Table S6 and Fig. 2show unadjusted average
GCSE grades for state non-selective, grammar and private school
students, as well as average GCSE score adjusting separately for
EduYears GPS, family SES, prior ability and prior achievement, and
for all variables together. Unadjusted GCSEs between school types
mirrored unadjusted EduYears GPS results, with large differences
between non-selective and selective schools (see Unadjusted
GCSEin Fig. 2, details in Supplementary Table S6). Indeed, the
School type, all hype?
E Smith-Woolley et al.
npj Science of Learning (2018) 3 Published in partnership with The University of Queensland
mean GCSE score of students attending state non-selective
schools was approximately 1 SD below the mean GCSE score of
those attending grammar schools (d=1.05, 95% CIs =0.831.28)
and private school students (d=0.92, 95% CIs =0.751.09). This
translates to around a whole grade difference between average
GCSE scores for state non-selective school students and selective
school students. There was no difference between grammar and
private school studentsaverage GCSE score (t=1.00, p=0.32).
There were also no signicant differences between non-selective
schools in areas that varied in the selectivity of their schools (see
Supplementary Table S7 and Supplementary Fig. S4).
Controlling for selection factors
Controlling for EduYears GPS had a small effect on average GCSE
grades, with the GCSE variance explained by school type dropping
slightly from R
=0.07 to 0.06, see Fig. 2, details in Supplementary
Table S6). This relatively small effect is to be expected given that
EduYears GPS accounts for only 8% of the variance in GCSE (see
Supplementary Table S3). Controlling for family SES and prior
ability had a slightly larger effect on GCSE, in line with the GCSE
variance they account for (=24% and 27%, respectively). Out of
all of the selection factors, prior achievement had the biggest
impact on GCSE grades between school type, with average GCSE
for grammar schools falling from 10.12 (grade A) to 9.21 (grade B).
After controlling for prior achievement, the variance in GCSE
explained by school type dropped from 7.1 to 1.3%.
Controlling for all of the selection factors and EduYears GPS
together saw a further reduction in average GCSE between school
types, with average GCSE score for grammar (M=9.14; t=2.35, p
< 0.019) and private (M=9.32, t=6.16, p< 0.001) similar to that of
state non-selective school studentsaverage grade (M=8.96).
Although these mean differences between school types remained
signicant, they were greatly reduced. Standardised betas
Fig. 2 Plotted means (and 95% condence intervals) for unadjusted GCSE, GCSE controlling for GPS, GCSE controlling for SES, GCSE
controlling for prior ability, GCSE controlling for prior achievement and GCSE controlling for all variables between three school types: state
non-selective, grammar and private. Note: Details can be found in Supplementary Table S6
Fig. 1 EduYears GPS plotted means (and 95% condence intervals) between state non-selective, grammar and private school students. Note:
There were signicant EduYears GPS mean differences between state non-selective school students and both grammar (t=4.869, p< 0.001; d
=0.413) and private school students (t=7.170, p< 0.001; d=0.372). There was not a signicant difference between grammar and private
school students (t=0.436, p=0.659)
School type, all hype?
E Smith-Woolley et al.
Published in partnership with The University of Queensland npj Science of Learning (2018) 3
indicated that attending a grammar school compared to a non-
selective state school was associated with an increase of just 0.03
of a standard deviation in GCSE, and for private schools, the
increase was 0.07. In addition, no signicant differences emerged
between non-selective schools in varying selectivity areas (see
Supplementary Table S7 and Supplementary Fig. S4).
One of our main ndings was that after accounting for the
variance explained by the selection factors and EduYears GPS, the
variance in GCSE explained by school type dropped from 7.1% to
only 0.5% (see Supplementary Table S6 for regression results).
We report genetic mean differences between students attending
three different types of school: state non-selective, grammar and
private schools. We nd that, on average, students in state non-
selective schools have lower polygenic scores for years of
education (EduYears) compared to their peers in selective schools.
Furthermore, following the same pattern of results as EduYears,
there are also substantial mean differences in GCSE performance
between pupils in selective and non-selective school types.
However, almost all of these differences are explained by
heritable, individual-level factors, which schools actively or
passively use in the pupil selection process.
Although nding genetic differences between state non-
selective, grammar and private school students may initially seem
surprising, when we consider the heritable traits that selection is
based on, this difference is less unexpected. Put another way,
students with higher polygenic score for years of education have,
on average, higher cognitive ability, better grades and come from
families with higher SES, and these students are subsequently
more likely to be accepted into selective schools. This results in a
system in which children are intentionally phenotypically selected,
but unintentionally genetically selected.
However, despite nding mean genetic differences between
students of different school types, it should be noted that the
majority of the variation in EduYears GPS occurs within the school
type, not between the school types. For example, a Cohensdof
0.41, (the difference between mean EduYears scores for state non-
selective school students and grammar school students), which is
classed as a small-medium effect size, translates to an overlap of
approximately 83% between the two distributions.
Nevertheless, nding an association between genotype and
school type suggests that genetic factors are contributing to
variation in educational environments, a concept known as gene-
environment correlation (rGE). This occurs when individuals select,
modify and inherittheir environment, in part based on their
Putting our research within the context of rGE, we
suggest that in addition to students being selected into schools
based on their genetically inuenced traits (evocative rGE),
children themselves also actively select educational environments
that correlate with their genotype (active rGE). In the case of high
achieving students, these environments might be challenging or
competitive academic institutions, which grammar and private
schools are often reputed to be. Finally, because we know that the
factors used in school selection are substantially heritable, it is
likely that academically gifted children will come from academi-
cally gifted parents. These parents not only provide the genes but
also the environments to help them progress academically.
As well as having a higher average EduYears polygenic score,
students attending selective schools also achieve better GCSE
results on average.
There has been some debate in the
literature as to the size of this achievement gap, with studies
accounting for different background characteristics in their
analysis. We nd that almost all of the selective school advantage
in GCSE can be explained by family SES, achievement, ability and
EduYears GPS. After controlling for these factors, going to a
grammar vs. a state non-selective school is associated with a mean
GCSE grade increase of just 0.026 of a standard deviation and for
private schools, 0.070 of a standard deviation. Furthermore, the
variance in GCSE that school type explains falls from 7% to <1%.
Controlling for EduYears alone had a fairly small effect on
average GCSE grades between school types. However, this is to be
expected considering that EduYears GPS currently predicts
approximately 8% of the variance in GCSE15% of the heritability
estimated by the twin design
and approximately one-third of
the heritable variance from SNP-based studies of GCSE at age 16.
The predictive nature of EduYears is likely to increase with more
powerful GWA studies. For example, there was a threefold
increase in prediction of educational achievement at age 16 from
the 2016 EduYears GPS (based on a GWA study with N=293,723)
as compared to the 2013 EduYears GPS (N=126,559).
Although there were only small mean differences between
school types once selection factors and EduYears were controlled
for, this does not mean that other factors are not important for
achievement at age 16. Altogether, these factors do not predict all
of the variance in GCSE (R²=0.69). As shown previously,
achievement is the result of many genetically inuenced traits,
including behaviour, personality, home environment and health.
Furthermore, by nding a small effect of school type, we are not
saying schools are unimportant, or that teaching does not work.
Without schools, it is hard to imagine a successful education
system that allows children to reach their academic potential.
However, while schools themselves are important for academic
achievement, the type of school appears less so. Educational
achievement is not necessarily the only reason parents opt to
send their children to selective schools. A recent report on private
schools found that these students earned about £200,000 more in
their early career (between ages 26 and 42) as compared to state
school students.
However, this report did not distinguish
between non-selective and selective state schools. More research
is needed to see whether differences in university attendance,
career choice and earnings are still predicted by school type once
individual student factors have been accounted for. In addition to
differences in university and career outcomes, it would also be of
interest to identify potential differences between school types in
terms of non-cognitive traits as outcomes, with one survey nding
66% of parents believing that private schools instil a sense of
condence in pupils.
There are several limitations to our study. First, we recognise
that there is considerable variation in schools within our three
school typeswithin each of the school types, there will be
examples of exceptional and under-performing schools. In
particular, there is more variance in the state non-selective
schools category as it includes most of the schools. It also includes
a wide variety of other categories, such as schools that are allowed
to select for religion and schools that are allowed to select up to
10% of their pupils for talent in specialist subjects, such as sport,
performing or visual arts, and languages. These schools are not
allowed to select directly on academic grounds. However, there is
some evidence that they do in fact select more able students.
Nonetheless, accounting for prior achievement and ability at age
11, before most children enter secondary school, adjusts for this.
Another limitation of the present study is access to school type.
Grammar and private schools are not evenly distributed around the
country. Therefore, in local authority areas where there are no
selective schools, the average GCSE grade of pupils in non-selective
schools may be higher and in areas where there are a greater
number of selective schools, the average GCSE grade of non-
selective schools may be lower. Because there are far fewer
selective schools, this geographical effect may potentially inate the
average non-selective school GCSE grade. To see whether this had
an impact on GCSE differences, we split the non-selective school
group into three further groups: non-selective schools in selective
areas, partially selective areas and non-selective areas. Once we
controlled for all of the selection factors, we found that there were
School type, all hype?
E Smith-Woolley et al.
npj Science of Learning (2018) 3 Published in partnership with The University of Queensland
no differences between non-selective schools in areas of varying
selectivity (see Supplementary Table S7 and Supplementary Fig. S4).
Anal limitation to note is that the GCSE variable we used in
the analysis is a composite of only the three core subjects taken at
age 16English, science and mathematics. For other subjects,
such as languages, art and social sciences, school type may have a
greater inuence. However, because different school types
prioritise different subjects,
it is difcult to untangle the effect
of school type on optional rather than core subjects, although this
would be a useful direction for future research.
In the current study, we nd genetic differences between
students attending three school types: state non-selective schools,
grammar schools and private schools. We nd that selective
school students have higher polygenic scores for years of
education on average compared to students attending non-
selective schools. Furthermore, we nd substantial mean differ-
ences in GCSE between school types. However, once student and
family factors have been accounted for, as well as EduYears GPS,
the type of school that a child attends explains less than one
percent of the individual differences in educational achievement
(GCSE mean grade) at age 16.
This study included unrelated individuals from the Twins Early Develop-
ment Study (TEDS). TEDS is a large, representative sample of 16,000 twin
pairs born in England and Wales between 19941996 and followed from
birth to the present day.
Ethical approval for this study was received from
Kings College London Ethics Committee. Although there has been some
attrition throughout the years, approximately 10,000 twin pairs are still
actively involved in the study and provide rich behavioural and cognitive
data. Importantly, TEDS was and still is a representative sample of England
and Wales, as described in detail elsewhere.
In the present study, we
included 4814 unrelated individuals (one twin randomly in a pair) who had
data present for three key variables: genotype data, educational
achievement at age 16 and school type data. This sample included 2597
females (54%) and 2217 males (46%). Of this sample, 2533 individuals also
had data present for the selection factors: ability, achievement and SES,
which included 1427 females (56.3%) and 1106 males (43.7%). For a
breakdown of sample sizes by school type, see Supplementary Table S8.
Written informed consent was given for all participants involved for each
wave of data collection.
For information on how the sample were genotyped and the quality
control process, please see Supplementary Methods S1.
School type. When TEDS twins were 18, they received a questionnaire that
included a series of questions asking what type of school they attended
when they took exams at age 16the GCSEs. Respondents were asked to
indicate either Yesor Nofor different school types. We classied all
respondents who reported attending either a state non-selective school as
State non-selective, all those who indicated that they went to a grammar
school as Grammarand all those indicating that they went to a private
school as Private. In addition to TEDS data, we also accessed school type
information through the National Pupil Database (NPD;
government/collections/national-pupil-database). By supplementing TEDS
data with that from NPD, our nal school type numbers were: state non-
selective: n=4263, grammar: n=143, private: n=408. We also further split
state non-selective schools into three categories for follow-up analysis: non-
selective schools in fully selective areas (n=331), non-selective schools in
partially selective areas (n=905) and non-selective schools in non-selective
areas (n=3027). For more information on how and why we created these
groupings, including accuracy between data sources and selective area
groupings, please see Supplementary Methods S2.
Educational achievement at age 16. The GCSE is a standardised UK-based
examination administered at the end of compulsory education at age 16
(M=16.31, SD =0.29). Almost all students take the three core subjects:
English, mathematics and science. In addition, students are allowed to
choose a range of other subjects such as geography, history and art. These
subjects were graded from 4 (G, the minimum pass grade) to 11 (A*, the
best possible grade). In the current sample, GCSE results were obtained
from questionnaires sent via mail, in addition to telephone interviews with
twins and their parents. We further supplemented this with data from NPD.
Our analyses focused on the three core subjects: English, mathematics and
science taken by all students. Students taking science GCSE are either
awarded separate GCSEs for physics, chemistry and biology (triple
science) or as one course, which is double weighted (double science),
therefore, we took a mean grade of the science GCSEs. Because English,
mathematics and science grades correlated highly (r=0.700.82), we
created a GCSE composite. There were 3920 individuals for whom we had
both self-reported GCSE and NPD data, this composite correlated at r=
0.99 between both data sources, which supported the high accuracy of
TEDS data.
Selection factors
Socioeconomic status. Family SES was measured by taking the arithmetic
mean of ve measures: maternal and paternal education (measured on a
scale from 18, where 1 =no education and 8 =postgraduate qualica-
tions), occupation (indexed by the Standard Occupational Classication
(2000) on a scale from 19, where 1 =elementary administration and
service occupations and 9 =managers, directors and senior ofcials) and
maternal age at birth of rst child. All measures were standardised to have
a mean of 0 and a SD of 1 and at least three measures were required to
calculate the arithmetic mean.
Achievement tests at age 11. We did not have access to selective school
entrance exams, however, before children transition to secondary school,
they are usually required to take exams, which include English and
mathematics tests. In our sample these tests comprise two English tests
(reading and writing) and three maths tests (calculator and non-calculator
test as well as a mental arithmetic test). Due to the high correlation
between maths and English scores (r=0.67), we created a composite of
these test scores requiring both to be present.
Ability (general cognitive ability, g). To measure general cognitive ability,
participants were asked to complete an online battery of cognitive tests
administered as part of TEDS testing at age 11. These tests included verbal
and non-verbal abilities (M=11.2, SD =0.69). A mean score was derived
from four tests, two verbal tests (the Wechsler Intelligence Scale for
Children (WISC) Vocabulary Multiple-Choice and the WISC General
Knowledge test)
and two non-verbal tests (Ravens Progressive
and the WISC Picture Completion task).
Data availability
For information on data availability, please see the Twins Early Develop-
ment Study data access policy. This can be found at:
Genome-wide polygenic scores. We calculated polygenic scores that were
based on the summary statistics of the largest GWA study for years of
education (N=293,723 individuals).
A GPS is calculated by using
information from GWA study summary statistics about the strength of
association between a genetic variant and a trait, to score individuals
genotypes in independent samples. For each genotype in the independent
sample, all trait-associated alleles are counted and multiplied by their
effect size (i.e., their strength of association with a trait as reported in GWA
summary statistics). The sum of these weighted and counted alleles forms
a polygenic score for each individual. We used the software PRSice to
create individual GPS. Those SNPs that passed quality control were
clumped for linkage disequilibrium by applying an R
=0.1 cutoff within a
250-kb window. It is possible to calculate various GPS based on different
GWA study signicance thresholds for genetic variants, with less stringent
p-value thresholds resulting in GPS that include more SNPs. Here, we
calculated GPS for seven p-value thresholds (0.001, 0.05, 0.1, 0.2, 0.3, 0.4,
0.5). We report analyses for the p-value threshold of 0.05 in the main text;
however, the analyses for the other p-value thesholds are reported in
Supplementary Fig. S5. We regressed all GPS on the rst ten principal
components and used these standardised residuals in our analyses to
account for population stratication.
School type, all hype?
E Smith-Woolley et al.
Published in partnership with The University of Queensland npj Science of Learning (2018) 3
Mean differences. To estimate differences between the three school
types: state non-selective, grammar and private schools, we used a one-
way ANOVA with planned contrasts. In addition to the three-level school
type analysis, we also conducted follow-up analysis looking at differences
between state non-selective schools in areas with and without grammar
schools: non-selective schools in fully selective areas, non-selective schools
in partially selective areas and non-selective schools in non-selective areas.
As the sample sizes varied between groups, we used adjusted Cohensdto
estimate effect size. This test adjusts the calculation of the pooled standard
deviation with weights for the sample sizes.
To test the effect of school type after controlling for selection factors
(SES, prior achievement and prior ability) and EduYears GPS, we conducted
hierarchical linear regression with dummy coding. See Supplementary
Methods S3 for further information on analysis.
All methods were performed in accordance with relevant regulations
and guidelines.
We gratefully acknowledge the ongoing contribution of the participants in the Twins
Early Development Study (TEDS) and their families. TEDS is supported by a
programme grant to R.P. from the UK Medical Research Council (MR/M021475/1 and
previously G0901245), with additional support from the US National Institutes of
Health (AG046938). The research leading to these results has also received funding
from the European Research Council under the European Union's Seventh Frame-
work Programme (FP7/2007-2013)/ grant agreement n° 602768 and ERC grant
agreement n° 295366. RP is supported by a Medical Research Council Professorship
award (G19/2). S.S. is supported by the MRC/IoPPN Excellence Award and by the EU
Framework Programme 7 (602768). E.K. and K.R. were supported by a Medical
Research Council studentship. S.v.S. is supported by a Jacobs Foundation Research
Fellowship award (20172019). Y.K. is supported by the Tomsk State University
competitiveness improvement programme grant J.B.P. is a fellow of MQ:
Transforming Mental Health (MQ16IP16).
R.P. directs and received funding for the Twins Early Development Study (TEDS). R.P.
and E.S.W. conceived the present study. E.S.W. analysed and interpreted the data with
advice from all co-authors. R.P. supervised the project and interpreted the data. R.P.
and E.S.W. wrote the manuscript with help from all authors (J.B.P., S.S., K.R., E.K., S.v.S.,
K.A., P.S.D., T.Y., R.A., and Y.K.).
Supplementary information accompanies the paper on the npj Science of Learning
website (
Competing interests: The authors declare no competing interests.
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims
in published maps and institutional afliations.
1. Department for Education UK Government. National Statistics: Schools, Pupils and
Their Characteristics. National Tables, available at:
SFR20_2016_Main_Text.pdf (2016).
2. Broughton, N., Ezeyi, O., Hupkau, C., Keohane, N. & Shorthouse, R. Open Access: An
Independent Evaluation, available at:
3. Coe, R. et al. Evidence on the Effects of Selective Educational Systems, available at:
4. Dearden, L., Ferri, J. & Meghir, C. The effect of school quality on educational
attainment and wages. Rev. Econ. Stat. 84,120 (2002).
5. Schmidt, F. L. & Hunter, J. General mental ability in the world of work:
occupational attainment and job performance. J. Pers. Soc. Psychol. 86, 162
6. Deary, I. J., Strand, S., Smith, P. & Fernandes, C. Intelligence and educational
achievement. Intelligence 35,1321 (2007).
7. White, K. R. The relation between socioeconomic status and academic achieve-
ment. Psychol. Bull. 91, 461 (1982).
8. Strenze, T. Intelligence and socioeconomic success: A meta-analytic review of
longitudinal research. Intelligence 35, 401426 (2007).
9. Sirin, S. R. Socioeconomic status and academic achievement: A meta-analytic
review of research. Rev. Educ. Res. 75, 417453 (2005).
10. Bradley, R. H. & Corwyn, R. F. Socioeconomic status and child development. Annu.
Rev. Psychol. 53, 371399 (2002).
11. Andrews, J., Hutchinson, J. & Johnes, R. Grammar Schools and Social Mobility,
(Education Policy Institute, London, 2016). Available at:
12. Bolton, P. Grammar School Statistics, available at: http://researchbriengs.les. (2017).
13. Sullivan, A. & Heath, A. State and Private Schools in England and Wales. (University
of Oxford, 2002).
14. Department for Education UK Government. Revised GCSE and Equivalent Results in
England, available at:
and-equivalent-results-in-england-2015-to-2016 (2016).
15. Goldstein, H. & Sammons, P. The inuence of secondary and junior schools on
sixteen year examination performance: A crossclassied multilevel analysis. Sch.
Eff. Sch. Improv. 8, 219230 (1997).
16. Clark, D. Selective schools and academic achievement. BE J Econ Anal Policy 10,
2024 (2010).
17. Anderson, K., Gong, X., Hong, K. & Zhang, X. Do selective high schools improve
student achievement? Effects of exam schools in China. China Econ. Rev. 40,
121134 (2016).
18. Atkinson, A., Gregg, P. & McConnell, B. The Result of 11 plus Selection: an Inves-
tigation Into Opportunities and Outcomes for Pupils in Selective Leas. (Centre for
Market and Public Organisation Working Paper, UK, Bristol, 2006).
19. Waldfogel, J. & Washbrook, E. On Your Marks: Measuring the School Readiness of
Children in Low-to-middle Income Families (Resolution Foundation, London, 2011).
20. Knopik, V. S., Neiderheiser, J., DeFries, J. C. & Plomin, R. Behavioral Genetics 7th
edn (Worth Publishers, New York, 2017).
21. Plomin, R. & Deary, I. J. Genetics and intelligence differences: ve special ndings.
Mol. Psychiatry 20,98108 (2015).
22. Krapohl, E. et al. The high heritability of educational achievement reects many
genetically inuenced traits, not just intelligence. Proc. Natl Acad. Sci. 111,
1527315278 (2014).
23. Wainwright, M. A., Wright, M. J., Luciano, M., Geffen, G. M. & Martin, N. G. Mul-
tivariate genetic analysis of academic skills of the Queensland core skills test and
IQ highlight the importance of genetic g. Twin Res. Human. Genet. 8, 602608
24. Bartels, M., Rietveld, M. J., Van Baal, G. C. M. & Boomsma, D. I. Heritability of
educational achievement in 12-year-olds and the overlap with cognitive ability.
Twin Res. 5, 544553 (2002).
25. Thompson, L. A., Detterman, D. K. & Plomin, R. Associations between cognitive
abilities and scholastic achievement: Genetic overlap but environmental differ-
ences. Psychol. Sci. 2, 158165 (1991).
26. Calvin, C. M. et al. Multivariate genetic analyses of cognition and academic
achievement from two population samples of 174,000 and 166,000 school chil-
dren. Behav. Genet. 42, 699710 (2012).
27. Yang, J., Lee, S. H., Goddard, M. E. & Visscher, P. M. Genome-wide Association
Studies and Genomic Prediction(GCTA): Methods, Data Analyses, and Interpreta-
tions. In: Genome-Wide Association Studies and Genomic Prediction. Methods in
Molecular Biology (Methods and Protocols), vol 1019. (eds Gondro C., van der
Werf J., Hayes B.) 215236 (Humana Press, Totowa, NJ, 2013).
28. Yang, J., Lee, S. H., Goddard, M. E. & Visscher, P. M. GCTA: a tool for genome-wide
complex trait analysis. Am. J. Human. Genet. 88,7682 (2011).
29. Trzaskowski, M. et al. Genetic inuence on family socioeconomic status and
childrens intelligence. Intelligence 42,8388 (2014).
30. Krapohl, E. & Plomin, R. Genetic link between family socioeconomic status and
childrens educational achievement estimated from genome-wide SNPs. Mol.
Psychiatry 21, 437 (2016).
31. Trzaskowski, M., Shakeshaft, N. G. & Plomin, R. Intelligence indexes generalist
genes for cognitive abilities. Intelligence 41, 560565 (2013).
32. Trzaskowski, M. et al. DNA evidence for strong genome-wide pleiotropy of
cognitive and learning abilities. Behav. Genet. 43, 267273 (2013).
33. Plomin, R. et al. Common DNA markers can account for more than half of the
genetic inuence on cognitive abilities. Psychol. Sci. 24, 562568 (2013).
34. Dudbridge, F. Power and predictive accuracy of polygenic risk scores. PLoS Genet.
9, e1003348 (2013).
35. Harlaar, N. et al. A behavioural genomic analysis of DNA markers associated with
general cognitive ability in 7yearolds. J. Child Psychol. Psychiatry 46, 10971107
36. Wray, N. R. et al. Research review: polygenic methods and their application to
psychiatric traits. J. Child Psychol. Psychiatry 55, 10681087 (2014).
School type, all hype?
E Smith-Woolley et al.
npj Science of Learning (2018) 3 Published in partnership with The University of Queensland
37. Selzam, S. et al. Predicting educational achievement from DNA. Mol. Psychiatry
22, 267272 (2017).
38. Rietveld, C. A. et al. GWAS of 126,559 individuals identies genetic variants
associated with educational attainment. Science 340, 14671471 (2013).
39. Okbay, A. et al. Genome-wide association study identies 74 loci associated with
educational attainment. Nature 533, 539542 (2016).
40. Magnusson, K. Interpreting Cohens d Effect Size: An interactive Visualization,
available at: (2014).
41. Rutter, M. et al. Integrating nature and nurture: Implications of
personenvironment correlations and interactions for developmental psycho-
pathology. Dev. Psychopathol. 9, 335364 (1997).
42. Gibbons, S. & Silva, O. Faith primary schools: better schools or better pupils? J.
Labor. Econ. 29, 589635 (2011).
43. Iannelli, C. The role of the school curriculum in social mobility. Br. J. Sociol. Educ.
34, 907928 (2013).
44. Haworth, C. M., Davis, O. S. & Plomin, R. Twins Early Development Study
(TEDS): a genetically sensitive investigation of cognitive and behavioral devel-
opment from childhood to young adulthood. Twin Res. Human. Genet. 16,
117125 (2013).
45. Kovas, Y. et al. The genetic and environmental origins of learning abilities and
disabilities in the early school years. Monogr. Society Res. Child Dev. 72, vii144
46. Kaplan, E., Fein, D., Kramer, J., Delis, D. & Morris, R. The WISC-III as A Process
Instrument (The Psychological Corporation, New York, 1999).
47. Raven, J. Raven Progressive Matrices. In: Handbook of Nonverbal Assessment (ed.
McCallum R.S.) (Springer, Boston, MA, 2003).
48. Wechsler, D. Wechsler Intelligence Scale for Children Third Edition UK (WISC-
IIIUK) Manual (Psychological Corporation, London, 1949).
Open Access This article is licensed under a Creative Commons
Attribution 4.0 International License, which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as long as you give
appropriate credit to the original author(s) and the source, provide a link to the Creative
Commons license, and indicate if changes were made. The images or other third party
material in this article are included in the articles Creative Commons license, unless
indicated otherwise in a credit line to the material. If material is not included in the
articles Creative Commons license and your intended use is not permitted by statutory
regulation or exceeds the permitted use, you will need to obtain permission directly
from the copyright holder. To view a copy of this license, visit http://creativecommons.
© The Author(s) 2018
School type, all hype?
E Smith-Woolley et al.
Published in partnership with The University of Queensland npj Science of Learning (2018) 3
... A family's socioeconomic status (SES) describes their level of access to and control over economic and social resources relative to that of other families. Family SES has been recognized as an important determinant of children's educational opportunities and outcomes [1][2][3][4] , but less is known about the stability of the influence of family SES on children's school performance over longer historical periods [5][6][7][8] . ...
... npj Science of Learning (2022)4 Published in partnership with The University of Queensland ...
Full-text available
In Britain and elsewhere, the influence of family socioeconomic status (SES) on education is already evident in primary school, and it persists and increases throughout the school years, with children from impoverished families earning lower grades and obtaining fewer educational qualifications than children from more privileged backgrounds. Reducing the effect of family background on children’s education is a pivotal aim of educators, policymakers, and researchers, but the success of their efforts is poorly evidenced to date. Here, we show for the first time that over 95 years in Britain the association between family SES and children’s primary school performance has remained stable. Across 16 British population cohorts born between 1921 and 2011 (N = 91,935), we confirmed previous findings of a correlation between family SES and children’s school performance of 0.28 [95% Confidence Interval 0.22–0.34], after adjusting for cohort-specific confounders. Contrary to the popular assumption that family background inequality has increased over time, we observed only minimal differences in the association between family SES and school performance across British cohorts. We argue that education policies must prioritize equity in learning outcomes over equality in learning opportunities, if they seek to disrupt the perpetuation of social and economic inequality across generations. We speculate that the influence of family SES on children’s education will only noticeably weaken if primary education settings become better equipped to meet and remediate the children’s differential learning needs.
... However, withinfamily EA-PGS are not clustered in schools. The absence of selective elementary and middle schools in Norway is also reassuring, in contrast to the United Kingdom, where exam differences between selective and non-selective schools primarily reflect heritable characteristics involved in admission 33 . ...
A child’s environment is thought to be composed of different levels that interact with individual genetic propensities, with less advantaged environments suppressing genetic effects on achievement. However, studies have not tested this theory comprehensively across multiple environmental levels. Here, we quantify the contributions of child, parent, school, neighbourhood, district, and municipality factors to achievement, and investigate interactions between polygenic scores for educational attainment (EA-PGS) and environmental levels. We link population-wide administrative data on children’s standardised test results, schools and residential identifiers to the Norwegian Mother, Father and Child Cohort Study (MoBa), which includes >23,000 genotyped parent-child trios. We test for gene-environment interactions using multilevel models with interactions between EA-PGS and random effects for school and residential environments (thus remaining agnostic to specific features of environments) and use parent EA-PGS to control for gene-environment correlation. Our within-family results suggest that students’ EA-PGS interact with schools but not residential environments (neighbourhoods, districts, and municipalities), which explain negligible variance. Students’ EA-PGS explain four times the variance in achievement in the 2.5% of schools where EA-PGS associations are strongest as in the 2.5% where effects are weakest. Contrary to theory, PGS effects are stronger in less advantaged environments (lower-performing schools). None of the school sociodemographic measures we tested could explain the interaction. Schools make a greater difference to the achievements of students with lower EA-PGS, explaining 4% of the variance for students 2SD below the mean PGS, but 2% for students 2SD above the mean. Policy to reduce social inequality in achievement in Norway should focus on tackling unequal support across schools for children with difficulties, and not on differences between residential areas.
... This research contends that grammar schools are no more or less effective than non-selective schools, once their clear difference in intake has been taken into account (Gorard & Siddiqui, 2018). In addition, further research in the United Kingdom shows that exam differences between school types, including state-funded selective and non-selective schools, are primarily due to the heritable characteristics involved in pupil admission (Smith-Woolley et al., 2018), and that grammar school attendance has little positive effect on other essential aspects of school life such as school engagement, academic wellbeing, peer relationships, self-esteem, aspirations for the future, and mental health (Jerrim & Sims, 2020). ...
Almost all countries throughout the world have in place various means of academically sorting students that in most cases occurs at the higher educational level. In the main, those students who achieve the highest scores in externally devised examinations have a greater choice of what higher education course they wish to pursue. However, in Northern Ireland, there also exists a unique situation whereby primary school students also sit a highly competitive examination referred to as the 11 plus examination in order to gain entry into what is referred to as a selective grammar school. The purpose of this chapter is to provide an overview of the cause and effect of maintaining such a sorting and testing regime that has almost vanished from other regions education systems. The chapter begins by providing an overview of the establishment of Academic selection in Northern Ireland from 1947 to present which is subsequently followed by a review of the literature relating to the benefits and limitations of Academic Selection. The penultimate section provides an overview of the impact of and unintended consequences of Academic Selection and concludes with a discussion and analysis of the place of Academic Selection in Northern Irelands Education system.
... Results such as these highlight that school sector is not a strong predictor of basic skills achievement, and suggest that it is the social background and academic ability of children who attend private schools which support the appearance of better quality schooling (e.g. Smith-Woolley, 2018). ...
A higher proportion of students are privately educated in Australia, compared with many other nations. In this paper, we tested the assumption that private schools offer better quality education than public schools. We examined differences in student achievement on the National Assessment Programme: Literacy and Numeracy (NAPLAN) between public, independent, and catholic schools. Cross-sectional regressions using large samples of students (n = 1583–1810 ) at Years 3, 5, 7 and 9 showed few sector differences in NAPLAN scores in any domain. No differences were evident after controlling for socioeconomic status and prior NAPLAN achievement. Using longitudinal modelling, we also found no sector differences in the rate of growth for reading and numeracy between Year 3 and Year 9. Results indicate that already higher achieving students are more likely to attend private schools, but private school attendance does not alter academic trajectories, thus undermining conceptions of private schools adding value to student outcomes.
... Later, it turned out that this PRS explained approximately 15% of the variability in the duration of education, correlated (r = 0.4) with students' final grades [187] and predicted educational success almost as well as the best predictor, family's socioeconomic status [188,189]. One study even showed that genetic differences are entirely responsible for the difference in learning outcomes in UK schools both with and without pre-selection of students [190]. We should note, however, that the utility of these results is mainly in informing effective education policies as opposed to using education PRS on an individual level (see [191]). ...
Full-text available
Scientifically interesting as well as practically important phenotypes often belong to the realm of complex traits. To the extent that these traits are hereditary, they are usually ‘highly polygenic’. The study of such traits presents a challenge for researchers, as the complex genetic architecture of such traits makes it nearly impossible to utilise many of the usual methods of reverse genetics, which often focus on specific genes. In recent years, thousands of genome-wide association studies (GWAS) were undertaken to explore the relationships between complex traits and a large number of genetic factors, most of which are characterised by tiny effects. In this review, we aim to familiarise ‘wet biologists’ with approaches for the interpretation of GWAS results, to clarify some issues that may seem counterintuitive and to assess the possibility of using GWAS results in experiments on various complex traits.
... Or the study of the variation of exam scores in relation to school types and polygenic prediction of education (e.g. Smith-Woolley et al. 20 ). ...
Full-text available
The application of polygenic scores has transformed our ability to investigate whether and how genetic and environmental factors jointly contribute to the variation of complex traits. Modelling the complex interplay between genes and environment, however, raises serious methodological challenges. Here we illustrate the largely unrecognised impact of gene-environment dependencies on the identification of the effects of genes and their variation across environments. We show that controlling for heritable covariates in regression models that include polygenic scores as independent variables introduces endogenous selection bias when one or more of these covariates depends on unmeasured factors that also affect the outcome. This results in the problem of conditioning on a collider, which in turn leads to spurious associations and effect sizes. Using graphical and simulation methods we demonstrate that the degree of bias depends on the strength of the gene-covariate correlation and of hidden heterogeneity linking covariates with outcomes, regardless of whether the main analytic focus is mediation, confounding, or gene × covariate (commonly gene × environment) interactions. We offer potential solutions, highlighting the importance of causal inference. We also urge further caution when fitting and interpreting models with polygenic scores and non-exogenous environments or phenotypes and demonstrate how spurious associations are likely to arise, advancing our understanding of such results.
Understanding the factors that impact learners with respect to their academic achievement is critical for enhancing educational provision, and the nature of these factors can vary widely. They could be, for example, cognitive, conative, physiological, or physical. With increased understanding of such factors teachers can better meet learner needs. Investigations into individual differences are not uncommon within technology education, for example much work has been conducted in the area of attitudes towards technology. However, research into individual cognitive differences is an emerging space. In light of the overwhelming evidence illustrating that spatial ability, commonly described as the ability to generate and manipulate abstract visual images, is positively associated with STEM educational performance and retention, understanding the role of spatial ability in technology education is important. Acknowledging the potential implications of such insight but recognizing the lack of contextual evidence, this chapter describes the results of a series of studies conducted with the aim of supporting the development of theory and suggesting recommendations for practice with respect to spatial ability within technology education. A literature review of the extant literature on spatial ability was conducted, and four quantitative studies examined the theorised positionality of spatial ability within technology education, and its relationship with authentic problem solving and other cognitive factors.
In what sense are associations between particular markers and complex behaviors made by genome-wide association studies (GWAS) and related techniques discoveries of, or entries into the study of, the causes of those behaviors? In this paper, we argue that when applied to individuals, the kinds of probabilistic ‘causes’ of complex traits that GWAS-style studies can point towards do not provide the kind of causal information that is useful for generating explanations; they do not, in other words, point towards useful explanations of why particular individuals have the traits that they do. We develop an analogy centered around Galton's “Quincunx” machine; while each pin might be associated with outcomes of a certain sort, in any particular trial, that pin might be entirely bypassed even if the ball eventually comes to rest in the box most strongly associated with that pin. Indeed, in any particular trial, the actual outcome of a ball hitting a pin might be the opposite of what is usually expected. While we might find particular pins associated with outcomes in the aggregate, these associations will not provide causally relevant information for understanding individual outcomes. In a similar way, the complexities of development likely render impossible any moves from population-level statistical associations between genetic markers and complex behaviors to an understanding of the causal processes by which individuals come to have the traits that they in fact have.
Currently, there is more and more research focused on the psychological aspects of teaching mathematics disciplines. At the same time, knowledge in other sciences allows one to learn to draw conclusions about the benefits of genetic factors on academic performance and understanding. The article summarizes the results of recent studies by foreign scientists on this topic. To solve the problem of studying the factors of influence on the understanding of mathematics and factors that influence this, a questionnaire was drawn up. After interviewing students using ANOVA tests, hypotheses about the influence of gender, education level and course of study on students' opinion were tested, and conclusions were drawn. After that, the influence of other indicators on the opinion of students was studied, in particular, self-assessment of their own knowledge in mathematics, the level of their own efforts in the study of mathematics, the knowledge of friends and classmates, assessments of teachers at school and teachers at a university, assessments of parents' understanding of mathematics, assessment of the similarity of understanding in parents and children. As a result, the most determining factors were identified, the main of which was self-assessment of the level of knowledge in mathematics. Based on the identified key factors, using multinomial logistic regression, the probabilities of choosing a particular answer were estimated depending on the three most significant factors. The resulting models were visualized, based on which the final conclusions were drawn. In conclusion, the authors put forward assumptions about the most promising areas of research
The DNA revolution made it possible to use DNA to predict intelligence. We argue that this advance will transform intelligence research and society. Our paper has three objectives. First, we review how the DNA revolution has transformed the ability to predict individual differences in intelligence. Thousands of DNA variants have been identified that – aggregated into genome-wide polygenic scores (GPS) – account for more than 10% of the variance in phenotypic intelligence. The intelligence GPS is now one of the most powerful predictors in the behavioral sciences. Second, we consider the impact of GPS on intelligence research. The intelligence GPS can be added as a genetic predictor of intelligence to any study without the need to assess phenotypic intelligence. This feature will help export intelligence to many new areas of science. Also , the intelligence GPS will help to address complex questions in intelligence research, in particular how the gene-environment interplay affects the development of individual differences in intelligence. Third, we consider the societal impact of the intelligence GPS, focusing on DNA testing at birth, DNA testing before birth (e.g., embryo selection), and DNA testing before conception (e.g., DNA dating). The intelligence GPS represents a major scientific advance, and, like all scientific advances, it can be used for bad as well as good. We stress the need to maximize the considerable benefits and minimize the risks of our new ability to use DNA to predict intelligence.
Full-text available
Molecular Psychiatry publishes work aimed at elucidating biological mechanisms underlying psychiatric disorders and their treatment
Full-text available
A genome-wide polygenic score (GPS), derived from a 2013 genome-wide association study (N=127,000), explained 2% of the variance in total years of education (EduYears). In a follow-up study (N=329,000), a new EduYears GPS explains up to 4%. Here, we tested the association between this latest EduYears GPS and educational achievement scores at ages 7, 12 and 16 in an independent sample of 5825 UK individuals. We found that EduYears GPS explained greater amounts of variance in educational achievement over time, up to 9% at age 16, accounting for 15% of the heritable variance. This is the strongest GPS prediction to date for quantitative behavioral traits. Individuals in the highest and lowest GPS septiles differed by a whole school grade at age 16. Furthermore, EduYears GPS was associated with general cognitive ability (~3.5%) and family socioeconomic status (~7%). There was no evidence of an interaction between EduYears GPS and family socioeconomic status on educational achievement or on general cognitive ability. These results are a harbinger of future widespread use of GPS to predict genetic risk and resilience in the social and behavioral sciences.Molecular Psychiatry advance online publication, 19 July 2016; doi:10.1038/mp.2016.107.
Full-text available
This paper focuses on the role of curricular content on social mobility, an issue largely neglected by social mobility studies. Using data from the National Child Development Study we investigate the extent to which secondary school curricula account for social class differences in the chances of entering into the service class and avoiding a low-skilled occupation. The results show that curriculum matters in the acquisition of different social classes of destination but it matters more for children from advantaged social backgrounds than for children from lower classes of origin. This is because of their higher propensity to choose subjects such as languages, English, mathematics and science, which were found to be highly valued in the labour market. Moreover, net of the effect of origin class and individual ability, all or most of the advantage associated with attendance at selective schools is accounted for by the curriculum studied there.
Full-text available
Because educational achievement at the end of compulsory schooling represents a major tipping point in life, understanding its causes and correlates is important for individual children, their families, and society. Here we identify the general ingredients of educational achievement using a multivariate design that goes beyond intelligence to consider a wide range of predictors, such as self-efficacy, personality, and behavior problems, to assess their independent and joint contributions to educational achievement. We use a genetically sensitive design to address the question of why educational achievement is so highly heritable. We focus on the results of a United Kingdom-wide examination, the General Certificate of Secondary Education (GCSE), which is administered at the end of compulsory education at age 16. GCSE scores were obtained for 13,306 twins at age 16, whom we also assessed contemporaneously on 83 scales that were condensed to nine broad psychological domains, including intelligence, self-efficacy, personality, well-being, and behavior problems. The mean of GCSE core subjects (English, mathematics, science) is more heritable (62%) than the nine predictor domains (35–58%). Each of the domains correlates significantly with GCSE results, and these correlations are largely mediated genetically. The main finding is that, although intelligence accounts for more of the heritability of GCSE than any other single domain, the other domains collectively account for about as much GCSE heritability as intelligence. Together with intelligence, these domains account for 75% of the heritability of GCSE. We conclude that the high heritability of educational achievement reflects many genetically influenced traits, not just intelligence.
Full-text available
Intelligence is a core construct in differential psychology and behavioural genetics, and should be so in cognitive neuroscience. It is one of the best predictors of important life outcomes such as education, occupation, mental and physical health and illness, and mortality. Intelligence is one of the most heritable behavioural traits. Here, we highlight five genetic findings that are special to intelligence differences and that have important implications for its genetic architecture and for gene-hunting expeditions. (i) The heritability of intelligence increases from about 20% in infancy to perhaps 80% in later adulthood. (ii) Intelligence captures genetic effects on diverse cognitive and learning abilities, which correlate phenotypically about 0.30 on average but correlate genetically about 0.60 or higher. (iii) Assortative mating is greater for intelligence (spouse correlations ~0.40) than for other behavioural traits such as personality and psychopathology (~0.10) or physical traits such as height and weight (~0.20). Assortative mating pumps additive genetic variance into the population every generation, contributing to the high narrow heritability (additive genetic variance) of intelligence. (iv) Unlike psychiatric disorders, intelligence is normally distributed with a positive end of exceptional performance that is a model for 'positive genetics'. (v) Intelligence is associated with education and social class and broadens the causal perspectives on how these three inter-correlated variables contribute to social mobility, and health, illness and mortality differences. These five findings arose primarily from twin studies. They are being confirmed by the first new quantitative genetic technique in a century-Genome-wide Complex Trait Analysis (GCTA)-which estimates genetic influence using genome-wide genotypes in large samples of unrelated individuals. Comparing GCTA results to the results of twin studies reveals important insights into the genetic architecture of intelligence that are relevant to attempts to narrow the 'missing heritability' gap.Molecular Psychiatry advance online publication, 16 September 2014; doi:10.1038/mp.2014.105.
We use regression discontinuity design to examine the effect of a system of public exam high schools, which admit students solely by pre-existing achievement, on student college entrance exam scores in Beijing, China. More selective exam schools may have higher peer quality and sometimes are equipped with more experienced teachers and better facilities. We find, however, that elite exam high schools, which are the most selective, have no effects on student test scores. We find that on average the system of exam schools improves student performance on the exam, which indicates that students benefit from attending more selective non-elite schools. The results on qualifying for college admission are consistent with our findings about test scores. Differences among schools in peer achievement, student/teacher ratio and the percentage of certificated and experienced teachers partially explain our findings; self-choices of track and exam participation do not explain test scores or college admission.
Educational attainment is strongly influenced by social and other environmental factors, but genetic factors are estimated to account for at least 20% of the variation across individuals1. Here we report the results of a genome-wide association study (GWAS) for educational attainment that extends our earlier discovery sample1, 2 of 101,069 individuals to 293,723 individuals, and a replication study in an independent sample of 111,349 individuals from the UK Biobank. We identify 74 genome-wide significant loci associated with the number of years of schooling completed. Single-nucleotide polymorphisms associated with educational attainment are disproportionately found in genomic regions regulating gene expression in the fetal brain. Candidate genes are preferentially expressed in neural tissue, especially during the prenatal period, and enriched for biological pathways involved in neural development. Our findings demonstrate that, even for a behavioural phenotype that is mostly environmentally determined, a well-powered GWAS identifies replicable associated genetic variants that suggest biologically relevant pathways. Because educational attainment is measured in large numbers of individuals, it will continue to be useful as a proxy phenotype in efforts to characterize the genetic influences of related phenotypes, including cognition and neuropsychiatric diseases.
The Raven Progressive Matrices (RPM) tests measure “general cognitive ability” or, better, eductive, or “meaning making,” ability (Raven, Raven, & Court, 1998a,2000). The term “eductive” comes from the Latin root educere, which means, “to draw out.” The basic version of the test, known as the Standard Progressive Matrices (or SPM), consists of five sets of items of the kind shown in Figures 11.1 and 11.2. Within each set, the items become progressively more difficult. At the beginning of each set, the items, although easy again, follow a different logic. The sets in turn become progressively more difficult. The five sets offer those taking the test five opportunities to become familiar with the method of thought required to solve the problems. In addition to the Standard series, there is the Coloured Progressive Matrices (CPM), which is designed to spread the scores of children and less able adults and the Advanced Progressive Matrices (APM), developed to spread the scores of the top 20% of the population.
One of the best predictors of children's educational achievement is their family's socioeconomic status (SES), but the degree to which this association is genetically mediated remains unclear. For 3000 UK-representative unrelated children we found that genome-wide single-nucleotide polymorphisms could explain a third of the variance of scores on an age-16 UK national examination of educational achievement and half of the correlation between their scores and family SES. Moreover, genome-wide polygenic scores based on a previously published genome-wide association meta-analysis of total number of years in education accounted for ~3.0% variance in educational achievement and ~2.5% in family SES. This study provides the first molecular evidence for substantial genetic influence on differences in children's educational achievement and its association with family SES.Molecular Psychiatry advance online publication, 10 March 2015; doi:10.1038/mp.2015.2.