Twin designs offer a powerful framework for disentangling genetic and environmental sources of individual differences, but require zygosity information that is often unavailable. When zygosity is unknown, analyses may still be possible using a normal finite-mixture-distribution model. This model assumes that the bivariate twin phenotype is normally distributed within each zygosity group, an assumption that cannot be directly verified because zygosity is unobserved. We conducted a simulation study to determine how well the normal finite-mixture-distribution model recovered parameters if this assumption is violated and if the twin difference distribution could serve as a diagnostic for violations. In univariate and bivariate simulations, we varied sample size, variance structure and the degree and source of kurtosis. We examined whether marginal score transformations reduce bias. We applied univariate and bivariate models to educational gaming data with unknown zygosity. Simulation estimates were most precise at high heritability and remained reasonably stable with samples as small as 200 pairs. Excess kurtosis in twin difference scores substantially biased estimates, and marginal transformations did not meaningfully reduce bias. Kurtosis-related bias was smaller in the bivariate model than in the univariate model. Empirical estimates based on the univariate model seemed highly biased, producing very high heritability estimates (0.91–0.99), except for the measure of division-skills (A = 0.58, C= 0, E= 0.42). Estimates based on the bivariate model seemed less biased, but unstable. Although mixture modelling may be used for twin-analyses without zygosity information, violations of bivariate normality may introduce considerable, difficult-to-detect bias, warranting careful interpretation of results.

Importance: Although boys, children of parents with low education, and children younger than their classmates have higher rates of ADHD, it is unclear to what extent this reflects underlying risk for ADHD. Clarifying this question is important for equitable identification. Objective: To examine whether sex, parental educational attainment (EA), and birth month were associated with ADHD diagnosis after accounting for symptom severity and polygenic index (PGI). Design: Data from the population-based Norwegian Mother, Father, and Child Cohort Study (MoBa) linked with demographic and health register data. Recruitment from 1999 to 2008. Diagnostic follow-up available from 2008 to 2023. Data were analyzed from May 2025 to April 2026. Setting: Nationwide pregnancy cohort study in Norway. Participants: 110,905 children, of whom 85,424 had valid ADHD symptoms and/or ADHD PGI data before multiple imputation. Exposures: ADHD symptom scores based on 12 maternal-reported items from the Conners Parent Rating Scale-Revised: Short Form at age 5; ADHD-PGIs; Sex; Parental EA; and birth month. Main outcomes and Measures: Age at ADHD diagnosis based on International Classification of Diseases, code F90 in the Norwegian Patient Registry. Cox proportional hazards models were used to estimate the relationships of time to diagnosis with symptoms, PGI, and demographic variables. Results: Boys were more likely to be diagnosed with ADHD in childhood (<13 years; HR, 0.42 [95% CI, 0.39-0.45]), whereas girls were more likely to be diagnosed in adolescence (=>13 years; HR, 1.48 [95% CI, 1.38-1.59]). Children were also more likely to be diagnosed with ADHD if they were born into families with low education (HR, 0.30 [95% CI, 0.27-0.33] comparing a Master’s degree with basic education) or if they were relatively younger than their classmates (HR, 1.50 [95% CI, 1.34-1.68] comparing birth month December with January). These group differences persisted even if children had the same level of symptoms or genetic predisposition for ADHD. Conclusions and Relevance: Sex, parental EA, and birth month were related to ADHD diagnosis beyond symptom severity and genetic predisposition, which may indicate the influence of demographic factors in how children are referred and diagnosed.

Children of parents with psychopathology generally perform less well in school than their peers. Parental symptoms might not, however, account for their lower achievement, as other familial factors might contribute. To examine the role of parental symptoms, we analyse data from 9,000 families in the Norwegian Mother, Father, and Child Study (MoBa). Parents reported their symptoms of anxiety, depression, ADHD, eating disorders, and alcohol use disorder during pregnancy and during their child’s childhood. Children in 5th grade (aged 10) completed nationally-standardised tests of mathematics, reading comprehension, and English (as an additional language). Comparing children who are cousins (children-of-siblings design), we controlled for unmeasured factors shared among family members that confound the relationship between parental mental health and children’s academic achievement. We found small associations between greater parental symptoms and lower scores in reading and mathematics (− 0.060 ≤ β ≤ −0.013). However, these associations were no longer significant when comparing cousins whose parents differ in mental health. Results suggest that parental symptoms do not exert a direct, clinically meaningful effect on children’s academic achievement. Although non-significant, associations with parental symptoms measured close to the child’s academic test are less attenuated than those measured during pregnancy. Notably, comparison between the overall MoBa sample and the children-of-siblings samples suggests ascertainment differences that hinder strong generalisations. Our findings underscore the importance of quasi-experimental family designs in identifying modifiable targets in intergenerational health and education research. Supporting parental mental health is crucial but its direct effect on children’s achievement may be smaller than often thought.

Why do children of more highly educated parents obtain better school grades? We addressed whether this reflects environmental effects of parents or familial factors shared across generations, using Norwegian population-register data (1.6 million individuals from 200k extended families) and genetically informative family designs. At age 16, teacher evaluations and national written exams in Norwegian (L1), English (L2), and mathematics were highly similar, with strong phenotypic correlations (r = .68-.84) and near-unit genetic correlations (r = .87-1.00). Associations between parental educational attainment and offspring grades were comparable across teacher evaluations, national exams, and grade point averages (r = .28-.36). For grade point averages, children-of-twins-and-siblings modelling indicated that most parent-offspring resemblance reflected genetic transmission (66%), alongside significant contributions from extended-family (25%) and parental transmission (10%). These findings suggest that intergenerational stability of education largely reflects genetic and extended-family transmission rather than teacher bias or direct effects of parental education alone.

Background: Studies on the home numeracy environment (HNE) show mixed evidence of its link with children’s mathematical skills. A rarely considered aspect is parents’ dual role as providers of the home environment and transmitters of genetic predispositions relevant for learning. Parents’ maths abilities may serve as a proxy for the genetic transmission that confounds the home-child association, as proposed in the Familial Control Method. We examined whether numeracy activities at home predict children’s maths development, while accounting for parents’ maths ability and children’s initial skill level. Methods: We analysed data from 244 Norwegian school children, 201 mothers, and 146 fathers. Children’s maths fluency was assessed in Grades 1, 3, 4, 5 and 6. Parents completed an HNE questionnaire and a maths assessment when children were in Grade 2. Relations among HNE, parental maths, and children’s maths development were tested using structural equation modelling. Results: Both formal and informal HNE predicted Grade 3 maths. After controlling for parental maths, only mothers’ informal HNE remained associated with Grade 3 performance. When Grade 1 maths was included, only fathers’ formal activities were related to Grade 3 skills, while both parents’ maths still predicted Grade 3 performance. Only Grade 1 maths predicted Grade 3 as well as subsequent growth. Conclusions: The finding that parental maths predicted children’s maths beyond family activities underscores the importance of addressing familial confounding in research on the home environment. To better understand the role of the HNE in maths development, research should consider parents’ skills, children’s early skills and HNE as a dynamic process.

Background: Children with attention-deficit/hyperactivity disorder (ADHD) are at increased risk of learning difficulties, which may reflect shared risk factors, ADHD symptoms hindering learning, or learning difficulties increasing ADHD symptoms. We aimed to map rates of co-occurrence among ADHD, dyslexia, dyscalculia, and reading comprehension difficulties and to investigate directional influences between ADHD symptoms and reading fluency, arithmetic, and reading comprehension. Methods: Reading fluency, arithmetic, and reading comprehension were assessed longitudinally in approximately 2,000 Finnish children across Grades 1 to 9 (ages 7–16). Attention/hyperactivity symptoms were measured with teacher ratings (Grades 1–4) and self-reports (Grades 6–9). ADHD, dyslexia, dyscalculia, and reading comprehension difficulties were based on a 10th percentile cut-off. Directionality between ADHD symptoms and academic skills was examined using random intercept cross-lagged panel models (RI-CLPMs) to separate stable traits from time-specific states. Results: Children with ADHD were 1.14 to 1.75 times more likely to have a learning difficulty, whereas those with one learning difficulty were 4.32 to 4.62 times more likely to have another. The RI-CLPMs showed consistent autoregressive effects across all domains, indicating stable within-person variation over time. At the between-person level (i.e., trait), academic skills were negatively related to ADHD symptoms, especially those reported by teachers. At the within-person level (i.e., state), ADHD symptoms were unrelated to reading fluency and arithmetic. However, at multiple time points, cross-lagged effects emerged from ADHD symptoms to subsequent reading comprehension, indicating that elevated ADHD symptoms predicted subsequent declines in reading comprehension. Conclusions: Learning difficulties co-occurred more frequently with each other (homotypic) than with ADHD (heterotypic co-occurrence). Both ADHD symptoms and learning difficulties showed trait-like stability. After accounting for these, ADHD symptoms were no longer linked to reading fluency or arithmetic. However, more ADHD symptoms predicted later declines in reading comprehension, likely because comprehension requires sustained attention—often impaired in children with hyperactivity and inattention.

We examined how the home math environment (HME) relates to children’s arithmetic fluency development. Unlike most prior research, we used the Familial Control Method to account for familial confounding by leveraging parental skills. Children’s arithmetic fluency was tested in Grades 1, 2 and 3 (n = 498). Mothers’ (n = 406) and fathers’ (n = 176) mathematical skills were assessed using math tests and self-reports. Parents reported formal and informal HME activities. We used latent growth curve models. More informal math-related activities (e.g. playing games) at home predicted higher arithmetic fluency in Grade 3 and faster development from Grades 1 to 3 (β’s .26 to .37). In contrast, more formal math-related activities (e.g. explicit parental teaching) predicted lower arithmetic fluency in Grade 3 and slower development from Grades 1 to 3 (β’s -.56 to -.43). These associations attenuated, but most remained significant after controlling for parental skills (|β|’s .15-.44). Informal play-based interactions were positively linked to arithmetic development, whereas formal instruction and homework help were negatively linked. However, this negative association may reflect parents increasing their formal support in response to children’s difficulties—suggesting a possible evocative effect. Parents with lower math skills tended to have children with lower arithmetic fluency (r = .31-.34), and to engage more in formal and less in informal home math activities. This study underscores the intergenerational transmission of children’s arithmetic skills. Future research should continue to account for familial confounding to better understand how the home environment shapes children’s learning.

Abstract Purpose: Individual differences in literacy and numeracy – that is, how well one can read, write, and work with numbers – emerge early in childhood and are crucial indicators of later educational success. Twin and family studies revealed substantial heritability of literacy and numeracy, and genetic overlap between these domains. Yet, it is unclear whether measured genomic variation shows a similarly overlapping or domain-specific pattern, and whether such molecular signals also predict adults’ educational attainment. Method: In N=15,669 participants (52.7% female; Netherlands Twin Register), we predicted children’s literacy, numeracy, and academic achievement, and adults’ educational attainment from genetic predispositions. Children’s learning outcomes were assessed via teacher and parent reports and a nationwide test (age M=12.25 years). Adults reported their highest educational qualification (age M= 42.47 years). We derived polygenic scores (PGSs) for years of education, dyslexia, and reading skills to predict these outcomes. Results: All three PGSs significantly predicted educational outcomes, but the broad years-of-education PGS exceeded the domain-specific dyslexia and reading-skills PGSs. The years-of-education PGS explained up to 10.7% in children’s academic achievement and 13.2% in adults’ attainment, suggesting that broader, large-scale PGS of general education capture much of the genetic signal shared across learning domains. Conclusion: Findings indicate both shared and specific genetic influences on learning. Broad, high-power PGSs maximise prediction, whereas literacy-focused PGSs highlight domain-specific signal. While current PGSs are not suitable for individual prediction, they are valuable for accounting for genetic propensities and improving causal inference in developmental and education research.

Children's academic achievement is linked to their parents’ education, a link often attributed to resources and support in the home. Yet children also inherit genes and grow up in complex social networks, requiring genetically-informed designs to uncover causal environmental effects. We used Norwegian register data from 569,035 children (aged 10–14), linked to the Norwegian Twin Registry, and applied extended family behaviour-genetic models to disentangle sources of intergenerational transmission. Parental education correlated .31 with children’s school-achievement scores; this correlation was due to substantial genetic (68%) and smaller parental-environmental (12%) and extended-family environmental (20%) contributions. Parents and children are alike in educational outcomes because of a complex mix of genetic similarity, environmental effects of parental education, and environmental effects shared with the extended family.

ADHD, dyslexia, and dyscalculia often co-occur, and the underlying continuous traits are correlated (ADHD symptoms, reading, spelling, and math skills). This may be explained by trait-to-trait causal effects, shared genetic and environmental factors, or both. We studied a sample of ≤ 19,125 twin children and 2,150 siblings from the Netherlands Twin Register, assessed at ages 7 and 10. Children with a condition, compared to those without that condition, were 2.1 to 3.1 times more likely to have a second condition. Still, most children (77.3%) with ADHD, dyslexia, or dyscalculia had just one condition. Cross-lagged modeling suggested that reading causally influences spelling (β = 0.44). For all other trait combinations, cross-lagged modeling suggested that the trait correlations are attributable to genetic influences common to all traits, rather than causal influences. Thus, ADHD, dyslexia, and dyscalculia seem to co-occur because of correlated genetic risks, rather than causality.

BACKGROUND:Numerous studies have investigated the associations between the home literacy environment (HLE) and children's word reading skills. However, these associations may partly reflect shared genetic factors since parents provide both the reading environment and their child's genetic predisposition to reading. Hence, the relationship between the HLE and children's reading is genetically confounded. To address this, parents' reading abilities have been suggested as a covariate, serving as a proxy for genetic transmission. The few studies that have incorporated this covariate control have made no distinction between the HLE reported by each parent or controlled for different skills in parents and children. We predicted children's reading development over time by the reading abilities of both parents as covariates and both parents' self-reported HLE as predictors. METHODS:We analyzed data from 242 unrelated children, 193 mothers, and 144 fathers. Children's word reading was assessed in Grades 1 and 3, and parents' word reading was assessed on a single occasion. Predictors of children's reading development included literacy resources and shared reading activities. RESULTS:Children's reading in Grade 3 was predicted by mothers' engagement in reading activities and by literacy resources at home, even after controlling for the genetic proxy of parental reading abilities. The longitudinal rate of change from Grades 1 to 3 was not associated with the HLE or parental reading. CONCLUSIONS:Our finding that parental reading skills predicted children's word reading beyond children's initial word reading underscores the importance of considering genetic confounding in research on the home environment. Beyond parental reading abilities, children's skills were predicted by literacy resources in the home and by how often mothers engage in reading activities with their children. This suggests true environmental effects.

Numerous studies have investigated the associations between the home literacy environment (HLE) and children's word reading skills. However, these associations may partly reflect shared genetic factors since parents provide both the reading environment and their child's genetic predisposition to reading. Hence, the relationship between the HLE and children's reading is genetically confounded. To address this, parents' reading abilities have been suggested as a covariate, serving as a proxy for genetic transmission. The few studies that have incorporated this covariate control have made no distinction between the HLE reported by each parent or controlled for different skills in parents and children. We predicted children's reading development over time by the reading abilities of both parents as covariates and both parents' self-reported HLE as predictors. We analyzed data from 242 unrelated children, 193 mothers, and 144 fathers. Children's word reading was assessed in Grades 1 and 3, and parents' word reading was assessed on a single occasion. Predictors of children's reading development included literacy resources and shared reading activities. Children's reading in Grade 3 was predicted by mothers' engagement in reading activities and by literacy resources at home, even after controlling for the genetic proxy of parental reading abilities. The longitudinal rate of change from Grades 1 to 3 was not associated with the HLE or parental reading. Our finding that parental reading skills predicted children's word reading beyond children's initial word reading underscores the importance of considering genetic confounding in research on the home environment. Beyond parental reading abilities, children's skills were predicted by literacy resources in the home and by how often mothers engage in reading activities with their children. This suggests true environmental effects.

In the classical twin design, the assumption that the additive genetic (A) and shared environment (C) variance components are uncorrelated may not hold. If there is positive AC covariance, the C component is overestimated. While many processes can lead to AC covariance, in this study, we focus on two widely-studied mechanisms: Cultural transmission (e.g., genetic nurture), when the parents' genotypes contribute to the effective environment of the child, and sibling interaction, when the genotype of one sibling contributes to the effective environment of another. Several designs use polygenic scores of parents or siblings to detect AC covariance, but these models cannot unambiguously identify the source. A combined model has been proposed, but its power to identify both processes has not been well-studied yet. This study uses exact data simulation to investigate the power to disentangle these processes. Results demonstrated that we can detect AC covariance using either genotyped-sibling or genotyped-parent data, but we cannot resolve its source and thus risk making wrong inferences. These sources of AC covariance can be resolved using genotyped data of both siblings and parents. However, the power analyses show that large samples are required to do so. The sample sizes of published studies may be too small to unambiguously resolve the contributions of the two processes to AC covariance, especially if the effect of one is relatively small compared to the effect of the other. We implement these findings in an R package for genomic simulations, gnomesims, and emphasize the need for whole-family genotyping and modeling.

Children born to parents with fewer years of education are more likely to have depression, anxiety, and attention-deficit hyperactivity disorder (ADHD), but it is unclear to what extent these associations are causal. We estimated the effect of parents' educational attainment on children's depressive, anxiety, and ADHD traits at age 8 years, in a sample of 40,879 Norwegian children born in 1998-2009 and their parents. We used within-family Mendelian randomization, which employs genetic variants as instrumental variables, and controlled for direct genetic effects by adjusting for children's polygenic indexes. We found little evidence that mothers' or fathers' educational attainment independently affected children's depressive, anxiety, or ADHD traits. However, children's own polygenic scores for educational attainment were independently and negatively associated with these traits. Results suggest that differences in these traits according to parents' education may reflect direct genetic effects more than genetic nurture. Consequences of social disadvantage for children's mental health may however be more visible in samples with more socioeconomic variation, or contexts with larger socioeconomic disparities than present-day Norway. Further research is required in populations with more educational and economic inequality and in other age groups.

This study examines the role of genes and environments in predicting educational outcomes. We test the Scarr-Rowe hypothesis, suggesting that enriched environments enable genetic potential to unfold, and the compensatory advantage hypothesis, proposing that low genetic endowments have less impact on education for children from high socioeconomic status (SES) families. We use a pre-registered design with Netherlands Twin Register data (426 <= N-individuals <= 3875). We build polygenic indexes (PGIs) for cognitive and noncognitive skills to predict seven educational outcomes from childhood to adulthood across three designs (between-family, within-family, and trio) accounting for different confounding sources, totalling 42 analyses. Cognitive PGIs, noncognitive PGIs, and parental education positively predict educational outcomes. Providing partial support for the compensatory hypothesis, 39/42 PGI x SES interactions are negative, with 7 reaching statistical significance under Romano-Wolf and 3 under the more conservative Bonferroni multiple testing corrections (p-value < 0.007). In contrast, the Scarr-Rowe hypothesis lacks empirical support, with just 2 non-significant and 1 significant (not surviving Romano-Wolf) positive interactions. Overall, we emphasise the need for future replication studies in larger samples. Our findings demonstrate the value of merging social-stratification and behavioural-genetic theories to better understand the intricate interplay between genetic factors and social contexts.

We investigate the causal relationship between educational attainment (EA) and mental health conditions using two research designs. Here we first compare the relationship between EA and 18 psychiatric diagnoses within-sibship in Dutch national registry data (N = 1.7 million), thereby controlling for unmeasured familial factors. Second, we apply two-sample Mendelian randomization, which uses genetic variants related to EA or psychiatric diagnosis as instrumental variables, to test whether there is a causal relation in either direction. Our results suggest that lower levels of EA causally increase the risk of major depressive disorder, attention-deficit/hyperactivity disorder, alcohol dependence, generalized anxiety disorder and post-traumatic stress disorder diagnoses. We also find evidence of a causal effect of attention-deficit/hyperactivity disorder on EA. For schizophrenia, anorexia nervosa, obsessive-compulsive disorder and bipolar disorder, the results were inconsistent across the different approaches, highlighting the importance of using multiple research designs to understand complex relationships, such as between EA and mental health conditions.

Non-cognitive skills, such as motivation and self-regulation, are partly heritable and predict academic achievement beyond cognitive skills. However, how the relationship between non-cognitive skills and academic achievement changes over development is unclear. The current study examined how cognitive and non-cognitive skills are associated with academic achievement from ages 7 to 16 years in a sample of over 10,000 children from England and Wales. The results showed that the association between non-cognitive skills and academic achievement increased across development. Twin and polygenic scores analyses found that the links between non-cognitive genetics and academic achievement became stronger over the school years. The results from within-family analyses indicated that non-cognitive genetic effects on academic achievement could not simply be attributed to confounding by environmental differences between nuclear families, consistent with a possible role for evocative/active gene-environment correlations. By studying genetic associations through a developmental lens, we provide further insights into the role of non-cognitive skills in academic development. Malanchini et al. find that non-cognitive skills increasingly predict academic achievement over development, driven by shared genetic factors whose influence grows over school years. These effects persist across socio-economic contexts and suggest the importance of fostering non-cognitive skills in education.