CoChat
Sign Up
Cited by 362Cognitive Abilities and TestingIntergenerational and Educational Inequality StudiesGenetic Associations and Epidemiology

Large Cross-National Differences in Gene × Socioeconomic Status Interaction on Intelligence

Open access
Tucker-Drob, Elliot M., Bates, Timothy C.·2016·Psychological Science
DOI 10.1177/0956797615612727PMC PMC4749462
Open this paper in CoChat — free account

Abstract

A core hypothesis in developmental theory predicts that genetic influences on intelligence and academic achievement are suppressed under conditions of socioeconomic privation and more fully realized under conditions of socioeconomic advantage: a Gene × Childhood Socioeconomic Status (SES) interaction. Tests of this hypothesis have produced apparently inconsistent results. We performed a meta-analysis of tests of Gene × SES interaction on intelligence and academic-achievement test scores, allowing for stratification by nation (United States vs. non–United States), and we conducted rigorous tests for publication bias and between-studies heterogeneity. In U.S. studies, we found clear support for moderately sized Gene × SES effects. In studies from Western Europe and Australia, where social policies ensure more uniform access to high-quality education and health care, Gene × SES effects were zero or reversed.

AI snapshot

Written from the paper’s full text.

TL;DR

This meta-analysis finds a positive gene × socioeconomic status (G × SES) interaction on intelligence in US studies but no such effect—or a reversed effect—in Western Europe and Australia. The cross-national difference is statistically significant and not explained by publication bias.

Key findings

  • In US studies, the G × SES interaction was significant and positive (a′ = .074, SE = .020, p < .0005); in non-US studies from Western Europe and Australia, it was trivial and non-significant (a′ = −.027, SE = .022, p = .223). The difference between US and non-US effects was significant (b = .101, SE = .032, p = .001).
  • The unconditional meta-analytic mean G × SES effect was non-significant (a′ = .029, SE = .019, p = .136), with significant heterogeneity (random-effect SD = .048, SE = .019, p = .011). Excluding Netherlands studies made non-US effects essentially zero (a′ = −.001, SE = .033, p = .979).
  • In the US model, genetic variance in intelligence increased from .24 at 2 SD below mean SES to .61 at 2 SD above mean SES, corresponding to heritability estimates of 26% and 61%, respectively. Shared and non-shared environmental variance decreased with SES, but these interactions were not statistically significant.
  • Publication-bias checks found no evidence of bias: funnel asymmetry tests were non-significant, and p-curve analysis indicated evidential value for US studies (z = −3.427, p < .001) but not for non-US studies (z = −1.264, p = .103). A power simulation estimated that 3,300 twin pairs are needed for 80% power based on the meta-analytic US estimate, versus 500 based on Turkheimer et al. (2003).

Methods

  • The authors meta-analyzed published and unpublished G × SES studies on objective intelligence/achievement measures, requiring sibling/twin genetic relatedness, known zygosity, childhood family SES as moderator, and no selection on diagnoses or test scores. The final dataset included k = 43 effect sizes from K = 14 independent studies and N = 24,926 twin/sibling pairs (approximately 50,000 individuals); US samples contributed k = 18, K = 8, N = 10,831, and non-US samples k = 25, K = 6, N = 14,095.
  • To ensure comparable estimates, the authors reanalyzed raw data, obtained reanalyses from original authors, or reanalyzed published subgroup covariance matrices. They fitted Purcell’s (2002) biometric model of latent Gene × Measured Shared Environment interaction, with Z-transformed phenotypes and SES, extracting a, c, e, s, a′, c′, and e′ parameters.
  • Effect sizes were meta-analyzed using random-effects meta-regression in Mplus, allowing multiple effect sizes per sample, weighting by the reciprocal number of effect sizes and inverse sampling variance, and using a sandwich estimator to correct standard errors for non-independence. Moderator analyses and publication-bias/robustness checks were also conducted.

Limitations

  • The authors state they could not fully test whether the US G × SES effects were driven by greater racial and ethnic diversity in US samples, though they cite individual studies suggesting this is not the case. They also note that candidate mechanisms for national differences—such as educational quality, medical/educational access, social mobility, and income support—were not directly tested.
  • The non-US evidence is limited to Western Europe and Australia, so the “non-US” estimate may not generalize to other nations. The authors also excluded several notable studies for methodological reasons (e.g., Scarr-Salapatek, 1971; NLSY kinship data; Nagoshi & Johnson, 2005; Friend et al.; Prescott et al.).
  • Evident from the text: the meta-analysis cannot establish causal environmental mechanisms, and residual non-US heterogeneity was driven by Netherlands studies, meaning the non-US summary may mask country-level variation. Publication-bias tests reduce but cannot entirely rule out bias.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1177/0956797615612727

Cites

50 of 50 references verified0 citation errors · 0 not checkable

Statistics

Claims the paper makes

10 of 26 cited sources checked · 6 supported · 0 flagged · accuracy not yet measured

  • Premisecitation

    Some subsequent US research has replicated the G × SES interaction.

    “Some subsequent US research has replicated the interaction (Bates, Lewis, & Weiss, 2013; Harden, Turkheimer, & Loehlin, 2007; Tucker-Drob, Rhemtulla, Harden, Turkheimer, & Fask, 2011; Turkheimer, Haley, Waldron, D’Onofrio, & Gottesman, 2003).”
    Quote in the paperNot a substantial claim1 of 3 sources checked
    • cites Bates, Lewis, & Weiss, 2013 source

      The indexed record’s title (metadata; no section or page) identifies a study of childhood socioeconomic status amplifying genetic effects on adult intelligence, which supports the interaction component. The available passages do not establish that this was US research or that it replicated an earlier finding; searches for the study setting and replication returned only metadata. (LLM, conf=0.72)

      Experimental — accuracy not yet measured
    • cites Harden, Turkheimer, & Loehlin, 2007

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites Tucker-Drob, Rhemtulla, Harden, Turkheimer, & Fask, 2011

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Findingown data

    The well-established phenomenon of increasing genetic influences and decreasing shared environmental influences on intelligence with childhood age was replicated in the current meta-analytic dataset.

    “We also replicated the well-established phenomenon of increasing genetic influences, and decreasing shared environmental influences, on intelligence with childhood age (e.g. Briley & Tucker-Drob, 2013; Haworth et al., 2009) in the current meta-analytic dataset.”
    Quote in the paperNot a substantial claim2 of 2 sources checked
    • cites Briley & Tucker-Drob, 2013 source
    • cites Haworth et al., 2009 source
  • Findingown data

    The non-US p-curve is not flatter than expected if studies were powered at 33% (z = .122, p = .548).

    “It also indicated that the p-curve is not flatter than one would expect if studies were powered at 33% (z = .122, p = .548)”
    Quote in the paperNot a substantial claim
  • Findingown data

    Excluding primary investigations authored by Eric Turkheimer, non-US samples had a null G × SES effect (a′ = −.027, SE = .022, p = .220).

    “A model that excluded these effect sizes (reported in Harden et al., 2007; Tucker-Drob et al., 2011; Turkheimer et al., 2003), yielded a null G × SES effect in non-US samples (a′ = −.027, SE = .022, p = .220)”
    Quote in the paperSubstantial claim
  • Findingown data

    The US meta-analytic G × SES finding was based on a total sample size of over 10,000 pairs of US twins and siblings.

    “Our meta-analytic finding of a G × SES interaction in the US, was based on a total sample size of over 10,000 pairs of US twins and siblings.”
    Quote in the paperNot a substantial claim
  • Classified as probably not a substantial claim. 6
  • May combine two claims. 20
  • The quote only partly states this claim. 2

  • Classified as probably not a substantial claim.

    Study design and conductreasoning

    Phenotypes and continuous SES measures were Z-transformed prior to fitting the Purcell model to ensure effect sizes were on a comparable metric across studies.

    “In order to ensure that effect sizes were placed on a metric that was comparable across studies, we ensured that phenotypes and continuous SES measures were Z-transformed (based on the M and SD from data pooled across all twins and siblings of all relatedness types) prior to fitting the Purcell (2002) model.”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Purcell (2002) source

  • May combine two claims.

    Premisecitation

    Heritability estimates from genome-wide data on unrelated individuals generally support large and increasing heritability of intelligence.

    “findings generally supported by heritability estimates from genome-wide data on unrelated individuals (Davies et al., 2011; Plomin et al., 2013).”
    Quote in the paperSubstantial claim2 of 2 sources checked
    • cites Davies et al., 2011 source

      The Abstract and Discussion report genome-wide estimates from unrelated adults that support substantial heritability of intelligence. The paper does not show that heritability increases; its estimates concern fluid and crystallized intelligence in middle to older adulthood, not a change over time. (LLM, conf=0.84)

      Experimental — accuracy not yet measured
    • cites Plomin et al., 2013 source

      The Discussion reports a substantial GCTA estimate for general cognitive ability, and the Genomewide Complex-Trait Analysis section reports estimates from unrelated adults and children. The paper describes increasing heritability from childhood to adulthood as a finding from developmental research, but does not show that genome-wide estimates establish that increase. (LLM, conf=0.86)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Findingown data

    There was significant support for age differences in the a and c parameters, indicating that genetic variance increases and shared environmental variance decreases with age.

    “there was significant support for age differences in the a and c parameters, indicating that genetic variance increases and shared environmental variance decreases with age”
    Quote in the paperNot a substantial claim1 of 2 sources checked
    • cites Briley & Tucker-Drob, 2013

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites Haworth et al., 2009 source

      The Abstract reports significant age-related increases in heritability and a decline in shared environmental influence for the broader 11,000-pair sample, citing an earlier analysis. But this paper’s Results (Heterogeneity model-fitting analyses; page unavailable) found no significant age-group differences in A, C, or E for its high-cognitive-ability sample. The claim does not distinguish the earlier finding from this paper’s own result. (LLM, conf=0.84)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Premisecitation

    Other studies have failed to replicate the G × SES finding, including the largest sample to date.

    “Other studies, however, have failed to replicate the finding (Bartels, van Beijsterveldt, & Boomsma, 2009; Kremen et al., 2005; Soden-Hensler, 2012; van der Sluis, Willemsen, de Geus, Boomsma, & Posthuma, 2008), including the largest sample to date (Hanscombe et al., 2012).”
    Quote in the paperSubstantial claim2 of 3 sources checked
    • cites Bartels, van Beijsterveldt, & Boomsma, 2009 source
    • cites Kremen et al., 2005 source
    • cites Hanscombe et al., 2012

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • May combine two claims.

    Interpretationcitation

    The increasing genetic influence with age and the US G × SES effect may both reflect a process of increased and accumulated effects of gene-environment transactions with increased opportunity that comes with both social class and age.

    “Indeed, both phenomena may reflect a process of increased and accumulated effects of gene-environment transactions with the increased opportunity that comes with both social class and age (Tucker-Drob et al., 2013; Turkheimer & Horn, 2014).”
    Quote in the paperNot a substantial claim1 of 2 sources checked
    • cites Tucker-Drob et al., 2013 source
    • cites Turkheimer & Horn, 2014

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Findingown data

    There is some indication that shared and non-shared environmental variance components decrease with SES, although these interactions were not statistically significant.

    “There is also some indication that the shared and non-shared environmental variance components decrease with SES, although these interactions were not statistically significant.”
    Quote in the paperNot a substantial claim

  • The quote only partly states this claim.

    Findingown data

    Excluding studies from the Netherlands, there was no G × SES effect for non-US studies, a′ = −.001, SE = .033, p = .979.

    “but no G × SES effect for non-US studies (a′ = −.001, SE = .033, p = .979).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    None of the additional moderators examined in the childhood subsample-childhood age, single vs. composite measure of SES, achievement/knowledge test score vs. intelligence test score, and single cognitive ability vs. composite cognitive measure-achieved statistical significance.

    “None of these moderators achieved statistical significance.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    All robustness checks indicated a significantly positive G × SES interaction in the US but not in Western Europe/Australia, with the difference between US and non-US studies being statistically significant.

    “In conclusion, all robustness checks indicated the presence of a significantly positive G × SES interaction in the US but not in Western Europe/Australia, with the difference between US and non-US studies being statistically significant.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The meta-analytic dataset consisted of k = 43 effect sizes from N = 24,926 pairs of twins and siblings, participating in K = 14 independent studies.

    “our meta-analytic dataset consisted of k = 43 effect sizes from a total of N = 24,926 pairs of twins and siblings (approximately 50,000 individuals) participating in K = 14 independent studies.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Data were split across US samples (k = 18, K = 8, N = 10,831) and non-US Western Europe and Australia samples (k = 25, K = 6, N = 14,095).

    “Data were relatively evenly split across US (k = 18, K = 8, N = 10,831) and Non-US (Western Europe and Australia) samples (k = 25, K = 6, N = 14,095).”
    Quote in the paperSubstantial claim

  • The quote only partly states this claim.

    Findingown data

    Excluding both TEDS and MTFS, the difference between US and non-US studies in G × SES effect was significant (difference = .135, SE = .042, p = .001).

    “and a significant difference between US and non-US studies (difference = .135, SE = .042, p = .001).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The p-curve for US studies is right skewed, indicating evidential value (z = −3.427, p < .001).

    “When only US studies were entered into p-curve analysis, p-checker indicated that the p-curve is, in fact, right skewed, indicating that these studies contain evidential value (z = −3.427, p < .001).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    No evidence of publication or reporting bias was found in the meta-analytic dataset; visual and formal tests of funnel plot asymmetry were not significant overall and separately for US and non-US samples.

    “We did not find evidence of publication or reporting bias in our meta-analytic dataset. Both visual and formal tests of funnel plot asymmetry were not significant, both when applied to the overall meta-analytic dataset, and when separately applied to effects sizes from US and non-US samples.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingreasoning

    Genes account for considerably more variation in intelligence at both higher ages and in higher US socioeconomic contexts.

    “Genes account for considerably more variation in intelligence at both higher ages and in higher US socioeconomic contexts.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    p-curve analysis indicated evidential value for a non-zero G × SES effect in US studies but not in non-US studies.

    “Consistent with the cross national difference identified in our primary meta-analysis, p-curve analysis further indicated evidential value for a non-zero G × SES effect in US studies, but did not indicate evidential value for a nonzero G × SES effect in the non-US studies.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    In studies conducted outside the US (Western Europe and Australia), the best estimate for G × SES magnitude was very slightly negative and not significantly different from zero.

    “Second, in studies conducted outside the US (in Western Europe and Australia), the best estimate for G × SES magnitude was very slightly negative, and not significantly different from zero.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Study design and conductown data

    The meta-analysis weighted effect sizes by the reciprocal number of effect sizes per sample and by the inverse sampling variance, and implemented a sandwich estimator correcting standard errors for non-independence of data points from the same sample.

    “We included multiple effect sizes per sample (e.g. for different intelligence measures, different waves, or different measures of SES), weighted meta-regression models by the reciprocal number of effect sizes included for the corresponding sample and by the inverse of each parameter’s sampling variance, and implemented a sandwich estimator that corrects standard errors for non-independence of data”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    Harden et al. (2007) reported evidence for G × SES interaction in a positively selected sample of US adolescents that contained very few racial minorities.

    “Additionally, Harden et al. (2007) reported evidence for G × SES interaction in a positively selected sample of US adolescents that contained very few racial minorities.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Harden et al. (2007)

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • May combine two claims.

    Study design and conductreasoning

    The meta-analysis used random effects meta-regression models in nested data, allowing for population variation in the true effect size above sampling error.

    “We used sophisticated methods for estimating random effects meta-regressions in nested data (i.e., in a meta-analytic dataset that contained multiple effect sizes per study).”
    • Also stated in: § Meta-Analyzing Effect Sizes · plain
      “Random effects models do not require the traditional assumption that all effect sizes were derived from the same population with a single true effect size. Rather, they allow for population variation in the true effect size above and beyond variation attributable to sampling error.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Study design and conductreasoning

    To ensure all effect sizes and standard errors were calculated using a consistent modeling approach, the authors obtained raw data and re-analyzed it, obtained parameter estimates from re-analyses by original study authors, or re-analyzed published sub-group covariance matrices.

    “Because this methodology required that all effect sizes and standard errors be calculated using a consistent modeling approach, we typically (a) obtained the raw data and re-analyzed it ourselves, (b) were provided with parameter estimates from re-analyses by the original study authors (all authors contacted complied with our request), or (c) re-analyzed the published sub-group covariance matrices”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Heritability estimates from twin and pedigree studies of intelligence increase with age.

    “and increase with age (Briley & Tucker-Drob, 2013)”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Briley & Tucker-Drob, 2013 source

      Not checked in the free run.

  • Classified as probably not a substantial claim. May combine two claims.

    Study design and conductown data

    The meta-analysis conducted rigorous tests for publication bias and between-study heterogeneity.

    “we meta-analyzed tests of G × SES interaction on intelligence and academic achievement test scores, allowing for stratification by nation (US vs non-US), and conducted rigorous tests for publication bias and between-study heterogeneity.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    G × SES effects are not uniform, but can take positive, zero, and even negative values depending on factors that differ at the national level.

    “The results indicate that G × SES effects are not uniform, but can rather take positive, zero, and even negative values depending on factors that differ at the national-level.”
    Quote in the paperSubstantial claim
24 of 24

  • The authors state this as an assumption.

    A continuous normal distribution underlies ordered categorical SES distributions.

    “The assumption that a continuous normal distribution underlies ordered categorical SES distributions was considered reasonable, because the studies that did employ continuous measures of SES typically transformed their SES measure to normality if it was not already approximately normally distributed.”
1 of 1

Statistics recomputed

Every reported test result, recomputed from its test statistic and degrees of freedom. A result shows as not adding up only once an independent check has confirmed it.

  • Not confirmedz = 1.253reported p = .895 · recomputed p = .210

    “(z = 1.253, p = .895”

  • Not confirmedz = 3.427reported p = 1 · recomputed p < .001

    “(z = 3.427, p = 1.000”

  • Not confirmedz = -1.264reported p = .103 · recomputed p = .206

    “(z = −1.264, p = .103”

  • Not confirmedz = 0.122reported p = .548 · recomputed p = .903

    “(z = .122, p = .548”

  • Not confirmedz = 1.264reported p = .897 · recomputed p = .206

    “(z = 1.264, p = .897”

  • Consistentz = -3.427reported p < .001 · recomputed p < .001

    “(z = −3.427, p < .001”

6 of 6

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Self-check results appear here once an independent check has confirmed them.

Cites

Flagged first, then by citations

Research works printed but not linked: 4

  1. Jacobson K. C., 2012, Behavior Genetics, 42, 941
  2. Muthén L. K., 2012, Mplus user’s guide
  3. Prescott C. A., 2014, Poster presented at the annual meeting of the Behavior Genetics Association
  4. Soden-Hensler B. (2012). An examination of Gene × Socioeconomic Status interactions for reading achievement (Doctoral dissertation, Florida State University). Retrieved from http://diginole.lib.fsu.edu/cgi/viewcontent.cgi?article=6923&context=etd

Cited by

Sorted by citations

Related

Authors
Tucker-Drob, E. M., & Bates, T. C.
Affiliations
  • Department of Psychology, University of Texas at Austin
  • University of Edinburgh (GB)
ORCID
Bates, Timothy C.
Venue
Psychological Science
Publisher
SAGE Publications
Published
2015-12-15
Volume
27
Issue
2
Pages
138-149
ISSN
  • 0956-7976 (ISSN-L)
  • 1467-9280
DOI
10.1177/0956797615612727
PMC
PMC4749462
OpenAlex
W2197899923
Cited by
359
References
48
Citations by year
2026: 26 · 2025: 28 · 2024: 46 · 2023: 28 · 2022: 45 · 2021: 40 · 2020: 29 · 2019: 39 · 2018: 31 · 2017: 30 · 2016: 14 · 2015: 1
FWCI
38.23
Type
journal article
Language
en
Open access
Licence
journals.sagepub.com/page/policies/text-and-data-mining-license (tdm)
Locations
  • Psychological Science (publishedVersion)
  • Europe PMC (PubMed Central) (publishedVersion)
  • PubMed Central (publishedVersion)
  • PubMed (publishedVersion)
Funders
  • Russell Sage Foundation
  • National Institutes of Health (5p2chd042849-22)
  • Eunice Kennedy Shriver National Institute of Child Health and Human Development (# P2CHD042849, R21 HD081437, R24 HD042849, P01HD031921)
Topics
  • Social Sciences › Psychology › Experimental and Cognitive Psychology › Cognitive Abilities and Testing
  • Social Sciences › Sociology and Political Science › Intergenerational and Educational Inequality Studies
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Genetics › Genetic Associations and Epidemiology
Keywords
intelligence, academic performance, genetic influences on intelligence, cross-national differences, publication bias, study heterogeneity, educational inequality, Western Europe, Australia
MeSH terms
Educational Measurement, Humans, Intelligence, Models, Genetic, Models, Psychological, Social Class, United States, Gene-Environment Interaction

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-29.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-29.

Cited sources are checked by gpt-6-sol.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review