Table 5.
Detailed characteristics of studies investigating offspring educational attainment and cognition (N = 21).
Offspring educational attainment and cognition | ||||||||
---|---|---|---|---|---|---|---|---|
Study | Design | Sample | Parental attribute (predictor) | Child attribute (outcome) | Control variables | Genetic overlap | Environmental transmission | G–E interplay |
Kendler et al. 124 | Adoption (siblings-reared-apart) |
Snr 436 sibships, one member reared by biological, other by adoptive parents Age: 18–20 years |
EA: highest education achieved by both parents, five-point-scale | IQ: Military Conscription Register, standardised test | Clustering of siblings within biological families | Not studied | Yes, adoptive parent EA predicted offspring IQ | |
Conley et al.82 | Within-family PGS: genetic sensitivity analysis, and genetic nurture (statistical control method) |
FHS, HRS 6186 individuals from 4867 households Mean age: 39.49 years (FHS), 68.17 years (HRS) |
Parental education Genetic transmission: effect of parental EA PGS Genetic nurture: effect of parental EA PGS, after adjusting for child EA PGS |
EA: self-report, highest grade completed | Child sex, age | Yes, parental EA PGS predicted offspring EA (effect size not clear) | Genetic sensitivity analysis: After controlling for offspring EA PGS, parental EA was still associated with offspring EA. Genetic nurture: no evidence of genetic nurture as parental EA PGS was not associated with offspring EA after controlling for offspring EA PGS (effect size not clear) | No G × E interaction found between maternal EA and offspring PGS |
Ayorech et al.79 | Extended twin, within-family PGS |
TEDS Twin analyses: 6105 twin pairs PGS analyses: 5825 individuals Age: 18 years |
EA (extended twin): self-reported highest qualification Genetic transmission (within-family PGS): effect of parental EA PGS |
EA: self or parent report, A levels qualification Intergenerational EA (extended twin): similarity between parental and offspring EA, two levels Intergenerational EA (within-family PGS): similarity between parental and offspring EA, four levels |
PGS analyses: previous school performance (GCSE grades) |
Twin analyses: yes, additive genetic effects underlying intergenerational EA were found (R2 = ~50%) PGS analyses: yes, parental EA PGS was associated with intergenerational EA |
Twin analyses: yes, shared environmental effects underlying intergenerational EA were found (R2 = ~40%) PGS analyses: Not studied |
|
Scheeren et al.90 | Adoption |
NLnr 1792 adopted children, 424,928 biological children Age: 15 years |
EA: register-based, highest education level Parental income: yearly household income |
EA: level of enrolment in secondary school, four levels | Father and mother year of birth, family structure, number of children in the household, observation year, adoption age, country of adoption, gender | Not studied | Adoptive parents’ income (but not EA) predicted offspring EA | Passive rGE: family income was more strongly associated with offspring EA in biological families than adoptive families |
Bates et al.14 | Within-family PGS: genetic nurture (transmitted/non-transmitted method) |
BATS 2,335 children and their genotyped parents Age: 17 years |
Genetic nurture: effect of EA PGS based on non-transmitted alleles SES: ASI-2006 |
EA: Queensland Core Skills Test | Sex, age at test, offspring EA PGS | Not studied | PGS for EA based on non-transmitted alleles were associated with offspring EA, but this relationship disappeared after adjusting for parental SES | No G × E interaction found between PGS and SES |
Belsky et al.86 | Within-family PGS: genetic nurture (statistical control method) |
E-RISK, NLAAH 1574 & 5526 individuals Age: 18 years, late 20 s to early 30 s |
Genetic nurture: effect of parental EA PGS, after adjusting for child EA PGS | EA: GCSE attainment; four levels | Genetic principal components | Not studied | Yes, parental EA PGS was associated with offspring EA after adjusting for offspring EA PGS | Passive rGE: individuals with higher PGS grew up in better-educated households |
Kong et al.83 | Within-family PGS: genetic nurture (transmitted/non-transmitted method) |
deCODE 21,637 probands with at least one genotyped parent Age: not reported |
Genetic transmission: effect of EA alleles PGS based on transmitted alleles Genetic nurture: effect of EA PGS based on non-transmitted alleles |
EA | Sex, year of birth, the interaction between sex and year of birth, 100 principal components | Yes, EA PGS based on transmitted parental alleles was associated with offspring EA (direct effect explained 70% of the overall observed effect of EA PGS) | Yes, EA PGS based on non-transmitted parental alleles was associated with offspring PGS (genetic nurture explained explaining 22.4% of the overall effect of EA PGS) | |
Liu et al.84 | Within-family PGS: genetic nurture (statistical control method) |
FHS, HRS 8639 individuals from three generations and 9342 individuals over age 50 Age: not reported |
Genetic transmission (FHS sample): effect of parental EA PGS Genetic nurture (FHS sample): effect of parental EA PGS, after adjusting for child PGS EA (HRS sample): self-report, years of education |
EA FHS: self-report, years of education completed HRS: parent report |
7 principal components HRS sample: child’s EA PGS |
Yes, parental EA PGS was associated with offspring EA (FHS sample; β = 0.345), and offspring EA PGS attenuated the association between parental and offspring EA (HRS sample; from β = 0.314 to β = 0.292) | Yes, parental EA PGS was associated with offspring EA, after adjusting for offspring EA PGS (β = 0.076) | |
Young et al.17 | Relatedness disequilibrium regression |
deCODE 12,035 individuals who had parents and grandparents genotyped Age: not reported |
Genetic nurture: estimated variance in offspring trait explained by parental genes acting indirectly via the environment | Educational attainment: self-report, number of years of schooling | Sex, year of birth | Not studied | Yes, after accounting for shared genetic effects, parental genes explained variance in offspring EA | |
Pingault et al.13 | Within-family PGS: genetic sensitivity analysis |
TEDS 3663–4693 individuals Age: 8–16 years |
Maternal EA: self-report, eight levels | EA: mean of three standardised tests | Sex, age and ten principal components of ancestry, PGS for EA | Yes, association between maternal EA and offspring EA decreased after adjusting for EA PGS (from β = 0.40 to 0.33) | Under a twin-heritability scenario, the association between maternal and offspring EA was expected to be null if EA PGS captured all heritability | |
Bates et al.15 | Within-family PGS: genetic nurture (transmitted/non-transmitted method) |
BATS 2335 children and their genotyped parents Age: 17 years |
Genetic nurture: effect of parental EA PGS based on non-transmitted alleles SES: ASI-2006 |
EA: Queensland Core Skills Test | Sex, age at test, offspring EA PGS | Not studied | PGS for EA based on non-transmitted alleles were associated with offspring EA, but this relationship disappeared after adjusting for parental SES | No G × E interaction found between PGS and SES |
Willoughby et al.88 | Within-family PGS: genetic nurture (statistical control method) |
MCTFR 1223 families, 2446 offspring Age: varied |
Genetic nurture: effect of parental EA PGS, on top of child EA PGS SES: composite score, family income, parent education level, parent occupation level Parental IQ: WIS |
Years of education: self-report, mean age 29 High-school grade-point-average: self-report, age 17 IQ: WIS, mean age 14.4 |
Height and BMI used as negative controls | Not studied | Yes, parental EA PGS was associated with offspring EA traits after adjusting for offspring EA PGS, and this association was mediated by parental SES and IQ | |
Armstrong-Carter et al.89 | Within-family PGS: genetic nurture (statistical control method) |
BiBs 2077 mother–child dyads Age: 7 years |
Genetic nurture: effect of maternal EA PGS, after adjusting for child EA PGS Maternal health: composite score, self-reported mental health, smoking, indirect smoke exposure, alcohol and drug use, vitamin use, sleep problems, and BMI SES: composite score, self-reported education, cohabitation status, employment, maternity leave, governmental benefits, perceived financial difficulty, and governmental index of neighbourhood-level deprivation |
Academic performance: standardised national exam | Child EA PGS, maternal age, first ten principal components | Not studied | Yes, maternal EA PGS was associated with offspring academic performance, after adjusting for offspring EA PGS, and this association was mediated by maternal health and SES during pregnancy | |
Borriello et al.80 | Adoption |
EGDS 195 families Age: 7 years |
Mathematical achievement: standardised scores on the mathematics fluency subtest of WJ-III | Mathematical achievement: standardised scores on the mathematics fluency subtest of the WJ-III | Obstetric complications, adoption opennness, parent education level, non-mathematical cognitive skills | Yes, birth parent and offspring mathematic achievement were correlated (β = 0.17) | Yes, paternal (but not maternal) mathematic achievement was correlated with adopted-offspring mathematical achievement (β = 0.15) | No G × E interaction found |
Domingue et al.85 | Adoption (PGS study) |
WLS 855 adopted and 20,939 biological offspring Age: not reported |
Genetic transmission: association between parental EA PGS and EA of biological offspring Genetic nurture: association between parental EA PGS and EA of adoptive offspring |
Educational attainment: parent-reported, highest grade of school attended | Child sex, age, ten principal components | Yes, parental EA PGS was associated with EA of biological offspring (effect size not clear) | Yes, parental EA PGS was associated with EA of adoptive offspring (effect size not clear) | Passive rGE implied: higher association in biological families than adoptive families |
de Zeeuw et al.75 | Within-family PGS: genetic nurture (transmitted/non-transmitted method) |
NTR 5900 offspring from 2649 families Age: 10–12, 25–64 years |
Genetic transmission: effect of EA and ADHD ADHD PGS based on transmitted alleles Genetic nurture: effect of EA and ADHD PGS based on non-transmitted alleles |
Childhood academic achievement: nationwide standardised test at age 12 Adult EA: self-report, highest degree; four levels |
Sex, birth year (EA), interaction between sex and birth year (EA), ten principal components, genotyping platform | EA PGS based on transmitted parental alleles were associated with offspring academic achievement in childhood and EA in adulthood (R2 = 5.7–7.6%) but there was no association with ADHD PGS | EA PGS based on non-transmitted parental alleles were associated with offspring EA in adulthood (R2 = 1.7%), but not academic achievement in childhood (which was also not associated with non-transmitted PGS for ADHD) | |
Halpern-Manners et al.81 | Adoption |
EGDS 340 families Age: first-graders (6–7 years) |
Adoptive and birth parent education attainment: self-report, highest level of education completed by adoptive or birth parents | Early educational achievement: WJ-III | Obstetric complications, adoption opennness, child sex, child and adoptive parents’ ethnicity, adoptive parents’ age, type of adoption agency | Yes, birth parent EA was associated with offspring EA (effect size not clear) | Yes, adoptive parent EA was associated with offspring EA (effect size not clear) | No G × E interaction |
Torvik et al.45 | Children-of-twins and siblings |
MoBa 34,958 children Age: 8 years |
Educational attainment: self-report, highest level completed | Academic problems: maternal report, three-point scale | Yes, there were shared genetic effects between parental EA and offspring academic problems (effect size not clear) | Yes, after accounting for genetic relatedness, parental EA was associated with offspring academic problems (effect size not clear) | ||
Ellingson et al.71 | Sibling comparison |
CNLSY 10,251 children of 4827 mothers Age: 4–14 years |
Smoking during pregnancy: self-report, mean number of packs smoked per day | Cognitive functioning: PPVT-R (math, reading and reading Recognition subtests) and digit span test | Maternal age at birth, EA, intelligence, delinquency, offspring sex, birth order, ethnicity, household income, geographic location | Not studied | Exposed children had poorer reading recognition than their unexposed siblings, but there were no other group differences | |
Kuja-Halkola et al.67 | Sibling comparison, children-of-twins |
Snr 2,754,626 children Age: up to 20 years |
Maternal smoking during pregnancy: self-report |
Academic achievement: class 9 records General cognitive ability: Military Conscription Register, nine levels |
Maternal age at childbirth, child sex, birth year | Yes, there were shared genetic effects between maternal smoking during pregnancy and offspring EA traits (effect size not clear) | No, exposed children did not differ from their unexposed siblings, and after accounting for genetic relatedness, maternal smoking was not associated with offspring EA traits | |
Wertz et al.78 | Within-family PGS: genetic nurture (statistical control method) |
E-RISK 860 mothers and their children Age: 18 years |
Genetic nurture: effect of maternal EA PGS, after adjusting for child PGS Parenting behaviour: mother, child and interviewer report, cognitive stimulation, warmth and sensitivity, household chaos, and safety and tidiness of the family home |
EA: self-report, highest educational attainment, 18 years | Sex, first ten principal components, offspring EA PGS | Yes, controlling for offspring EA PGS attenuated the association between parenting behaviours and offspring EA (from β range = 0.33–0.52 to β range = 0.30–0.48) | Genetic nurture: yes, maternal EA PGS was associated with offspring EA after adjusting for offspring EA PGS (β = 0.11), and this effect was mediated by parenting behaviours including cognitive stimulation, household chaos and a safe, tidy home (but not parental warmth) | Evocative rGE: mother and offspring PGS for EA predicted cognitive stimulation and warm, sensitive parenting |
G–E gene–environment, G × E gene–environment interaction, rGE gene–environment correlation.
Design = PGS Polygenic scores.
Samples = BATS Brisbane Adolescent Twin Study, BiBs Born in Bradford study, CNLSY Children of the National Longitudinal Survey of Youth, EGDS Early Growth and Development Study, deCODE Icelandic Genealogy Database, FHS Framingham Heart Study, HRS Health Retirement Study, MoBa Norwegian Mother Father and Child Study, MCTFR Minnesota Centre for Twin and Family Research, NLNR Dutch national registers, NLAAH National Longitudinal study of Adolescent to Adult Health, NTR Netherlands Twin Register, SNR Swedish national registers, TEDS Twins Early Development Study, WLS Wisconsin Longitudinal Study.
Measures = ASI Australian Socioeconomic Index occupational status scale, PPVT-R Peabody Picture Vocabulary Test-Revised, QCST Queensland Core Skills Test, WIS Weschler Intelligence Scale, WJ-III Woodcock–Johnson Test of Achievement III.
Statistics = β standardised parameter estimate, R2 percentage of variance explained. Effect sizes are not reported for studies that did not investigate both genetic and environmental transmission.