Within- and between-family genetic effects on educational achievement vary across countries and ages

Alexandra Starr; Mirko Ruks; Alexandros Giannelis; Emily Willoughby; Oskar Pettersson; Charlotte K L Pahnke; Carlo Maj; Anastasia Andreas; Rafael Ahlskog; Louise Arseneault; Helen L Fisher; Andreas J Forstner; Christian Kandler; Matt McGue; Markus M Nöthen; Sven Oskarsson; Frank M Spinath; Scott Vrieze; Jasmin Wertz; Sophie von Stumm

doi:10.1038/s41380-025-03342-0

Within- and between-family genetic effects on educational achievement vary across countries and ages

Alexandra Starr
, Mirko Ruks
, Alexandros Giannelis
, Emily Willoughby
, Oskar Pettersson
, Charlotte K L Pahnke
, Carlo Maj
, Anastasia Andreas
, Rafael Ahlskog
, Louise Arseneault
, Helen L Fisher
, Andreas J Forstner
, Christian Kandler
, Matt McGue
, Markus M Nöthen
, Sven Oskarsson
, Frank M Spinath
, Scott Vrieze
, Jasmin Wertz
, Sophie von Stumm

Research output: Contribution to Journal › Article › Academic › peer-review

Abstract

Polygenic score (PGS) predictions of educational achievement are sizeable at the population level. Yet, population-level PGS predictions are environmentally confounded, due to gene-environment correlations, assortative mating, and population stratification. This confounding complicates the interpretation and application of PGS predictions of educational achievement. Here, we charted the variability of PGS predictions in N = 8115 dizygotic twins from UK, US, Swedish, and German samples aged 7 to 19 years. Population-level PGS predictions of educational achievement ranged from β = 0.16 to β = 0.37 across ages and countries. Discerning within- and between-family level estimates, we found that 10 to 65% of the population-level PGS predictions were due to environmental confounding, of which 29 to 100% were accounted for by family socioeconomic status. Variability in within-family and population-level PGS predictions was largely unsystematic across countries' school systems (multi-tiered vs. comprehensive) and children's ages. Therefore, interpretations regarding the sources of environmental confounding effects on educational achievement remain, at present, speculative.

Original language	English
Pages (from-to)	2029-2037
Number of pages	9
Journal	Molecular Psychiatry
Volume	31
Issue number	4
Early online date	20 Nov 2025
DOIs	https://doi.org/10.1038/s41380-025-03342-0
Publication status	Published - Apr 2026

Bibliographical note

Funding

TEDS is supported by the UK Medical Research Council (MR/V012878/1 and previously MR/M021475/1), with additional support from the US National Institutes of Health (AG046938). The Environmental Risk (E-Risk) Study is funded by grants from the UK Medical Research Council (MRC) [G1002190; MR/X010791/1]. Additional support for E-Risk was provided by the US National Institute of Child Health and Human Development [HD077482] and the Jacobs Foundation. The Minnesota Twin Family Study was supported in part through grants from the National Institutes of Health R01 DA 042755, R01HG011035, R01 DA 054087, R01 DA 044283. The Swedish Twin Registry (STR) is managed by the Karolinska Institute and receives additional funding through the Swedish Research Council (2017-00641). Additional funding for STR comes from the Ragnar Söderberg Foundation (E9/11), and the Swedish Research Council (421-2013-1061). TwinLife is funded by grants from the German Research Foundation (DFG; project number: 220286500 and 428902522). SvS held fellowships from the Jacobs Foundation (2022-2027) and the Paris Institute of Advanced Study (2023-2024) during the writing of this manuscript. HLF was supported by the UK ESRC Centre for Society and Mental Health at King’s College London [ES/S012567/1]. The position of CP is funded by the DFG (funding number 428902552). OP, RA, and SO were supported by the Swedish Research Council (2019-00244; 2023-01343). TEDS is supported by the UK Medical Research Council (MR/V012878/1 and previously MR/M021475/1), with additional support from the US National Institutes of Health (AG046938). The Environmental Risk (E-Risk) Study is funded by grants from the UK Medical Research Council (MRC) [G1002190; MR/X010791/1]. Additional support for E-Risk was provided by the US National Institute of Child Health and Human Development [HD077482] and the Jacobs Foundation. The Minnesota Twin Family Study was supported in part through grants from the National Institutes of Health R01 DA 042755, R01HG011035, R01 DA 054087, R01 DA 044283. The Swedish Twin Registry (STR) is managed by the Karolinska Institute and receives additional funding through the Swedish Research Council (2017-00641). Additional funding for STR comes from the Ragnar Söderberg Foundation (E9/11), and the Swedish Research Council (421-2013-1061). TwinLife is funded by grants from the German Research Foundation (DFG; project number: 220286500 and 428902522). SvS held fellowships from the Jacobs Foundation (2022-2027) and the Paris Institute of Advanced Study (2023-2024) during the writing of this manuscript. HLF was supported by the UK ESRC Centre for Society and Mental Health at King’s College London [ES/S012567/1]. The position of CP is funded by the DFG (funding number 428902552). OP, RA, and SO were supported by the Swedish Research Council (2019-00244; 2023-01343).

Funders	Funder number
Economic and Social Research Council
Karolinska Institutet
King's College London	2019-00244, 2023-01343, ES/S012567/1, 428902552
National Institutes of Health	AG046938
Deutsche Forschungsgemeinschaft	220286500, 2022-2027, 428902522
Medical Research Council	MR/V012878/1, MR/X010791/1, G1002190, MR/M021475/1
Institut d’études avancées de Paris	2023-2024
Ragnar Söderbergs stiftelse	421-2013-1061, E9/11
Vetenskapsrådet	2017-00641
Jacobs Foundation	R01HG011035, R01 DA 042755, R01 DA 044283, R01 DA 054087
National Institute of Child Health and Human Development	HD077482

Access to Document

10.1038/s41380-025-03342-0

Persistent URL (handle)

Persistent link

Embargoed Document

Within- and between-family genetic effects on educational achievement vary across countries and ages
Final published version, 1.38 MB
Licence: Article 25fa Dutch Copyright Act
Embargo ends: 22/05/26

Cite this

Starr, A., Ruks, M., Giannelis, A., Willoughby, E., Pettersson, O., Pahnke, C. K. L., Maj, C., Andreas, A., Ahlskog, R., Arseneault, L., Fisher, H. L., Forstner, A. J., Kandler, C., McGue, M., Nöthen, M. M., Oskarsson, S., Spinath, F. M., Vrieze, S., Wertz, J., & von Stumm, S. (2026). Within- and between-family genetic effects on educational achievement vary across countries and ages. Molecular Psychiatry, 31(4), 2029-2037. https://doi.org/10.1038/s41380-025-03342-0

@article{64627d03061d473d9c2aae21f5174554,

title = "Within- and between-family genetic effects on educational achievement vary across countries and ages",

abstract = "Polygenic score (PGS) predictions of educational achievement are sizeable at the population level. Yet, population-level PGS predictions are environmentally confounded, due to gene-environment correlations, assortative mating, and population stratification. This confounding complicates the interpretation and application of PGS predictions of educational achievement. Here, we charted the variability of PGS predictions in N = 8115 dizygotic twins from UK, US, Swedish, and German samples aged 7 to 19 years. Population-level PGS predictions of educational achievement ranged from β = 0.16 to β = 0.37 across ages and countries. Discerning within- and between-family level estimates, we found that 10 to 65\% of the population-level PGS predictions were due to environmental confounding, of which 29 to 100\% were accounted for by family socioeconomic status. Variability in within-family and population-level PGS predictions was largely unsystematic across countries' school systems (multi-tiered vs. comprehensive) and children's ages. Therefore, interpretations regarding the sources of environmental confounding effects on educational achievement remain, at present, speculative.",

author = "Alexandra Starr and Mirko Ruks and Alexandros Giannelis and Emily Willoughby and Oskar Pettersson and Pahnke, \{Charlotte K L\} and Carlo Maj and Anastasia Andreas and Rafael Ahlskog and Louise Arseneault and Fisher, \{Helen L\} and Forstner, \{Andreas J\} and Christian Kandler and Matt McGue and N{\"o}then, \{Markus M\} and Sven Oskarsson and Spinath, \{Frank M\} and Scott Vrieze and Jasmin Wertz and \{von Stumm\}, Sophie",

note = "{\textcopyright} 2025. The Author(s), under exclusive licence to Springer Nature Limited.",

year = "2026",

month = apr,

doi = "10.1038/s41380-025-03342-0",

language = "English",

volume = "31",

pages = "2029--2037",

journal = "Molecular Psychiatry",

issn = "1359-4184",

publisher = "Springer Nature",

number = "4",

}

Starr, A, Ruks, M, Giannelis, A, Willoughby, E, Pettersson, O, Pahnke, CKL, Maj, C, Andreas, A, Ahlskog, R, Arseneault, L, Fisher, HL, Forstner, AJ, Kandler, C, McGue, M, Nöthen, MM, Oskarsson, S, Spinath, FM, Vrieze, S, Wertz, J & von Stumm, S 2026, 'Within- and between-family genetic effects on educational achievement vary across countries and ages', Molecular Psychiatry, vol. 31, no. 4, pp. 2029-2037. https://doi.org/10.1038/s41380-025-03342-0

TY - JOUR

T1 - Within- and between-family genetic effects on educational achievement vary across countries and ages

AU - Starr, Alexandra

AU - Ruks, Mirko

AU - Giannelis, Alexandros

AU - Willoughby, Emily

AU - Pettersson, Oskar

AU - Pahnke, Charlotte K L

AU - Maj, Carlo

AU - Andreas, Anastasia

AU - Ahlskog, Rafael

AU - Arseneault, Louise

AU - Fisher, Helen L

AU - Forstner, Andreas J

AU - Kandler, Christian

AU - McGue, Matt

AU - Nöthen, Markus M

AU - Oskarsson, Sven

AU - Spinath, Frank M

AU - Vrieze, Scott

AU - Wertz, Jasmin

AU - von Stumm, Sophie

PY - 2026/4

Y1 - 2026/4

N2 - Polygenic score (PGS) predictions of educational achievement are sizeable at the population level. Yet, population-level PGS predictions are environmentally confounded, due to gene-environment correlations, assortative mating, and population stratification. This confounding complicates the interpretation and application of PGS predictions of educational achievement. Here, we charted the variability of PGS predictions in N = 8115 dizygotic twins from UK, US, Swedish, and German samples aged 7 to 19 years. Population-level PGS predictions of educational achievement ranged from β = 0.16 to β = 0.37 across ages and countries. Discerning within- and between-family level estimates, we found that 10 to 65% of the population-level PGS predictions were due to environmental confounding, of which 29 to 100% were accounted for by family socioeconomic status. Variability in within-family and population-level PGS predictions was largely unsystematic across countries' school systems (multi-tiered vs. comprehensive) and children's ages. Therefore, interpretations regarding the sources of environmental confounding effects on educational achievement remain, at present, speculative.

AB - Polygenic score (PGS) predictions of educational achievement are sizeable at the population level. Yet, population-level PGS predictions are environmentally confounded, due to gene-environment correlations, assortative mating, and population stratification. This confounding complicates the interpretation and application of PGS predictions of educational achievement. Here, we charted the variability of PGS predictions in N = 8115 dizygotic twins from UK, US, Swedish, and German samples aged 7 to 19 years. Population-level PGS predictions of educational achievement ranged from β = 0.16 to β = 0.37 across ages and countries. Discerning within- and between-family level estimates, we found that 10 to 65% of the population-level PGS predictions were due to environmental confounding, of which 29 to 100% were accounted for by family socioeconomic status. Variability in within-family and population-level PGS predictions was largely unsystematic across countries' school systems (multi-tiered vs. comprehensive) and children's ages. Therefore, interpretations regarding the sources of environmental confounding effects on educational achievement remain, at present, speculative.

U2 - 10.1038/s41380-025-03342-0

DO - 10.1038/s41380-025-03342-0

M3 - Article

C2 - 41266877

SN - 1359-4184

VL - 31

SP - 2029

EP - 2037

JO - Molecular Psychiatry

JF - Molecular Psychiatry

IS - 4

ER -