Cited by 364Cognitive Abilities and TestingPsychometric Methodologies and TestingIntergenerational and Educational Inequality Studies

The DOI matches its Crossref record; 268 of the 268 references it cites were found in a registry or an index; none of the 34 sources its claims cite is checked for support yet.

The Flynn effect: A meta-analysis.

Open access

Abstract

The Flynn effect refers to the observed rise in IQ scores over time, which results in norms obsolescence. Although the Flynn effect is widely accepted, most efforts to estimate it have relied upon "scorecard" approaches that make estimates of its magnitude and error of measurement controversial and prevent determination of factors that moderate the Flynn effect across different IQ tests. We conducted a meta-analysis to determine the magnitude of the Flynn effect with a higher degree of precision, to determine the error of measurement, and to assess the impact of several moderator variables on the mean effect size. Across 285 studies (N = 14,031) since 1951 with administrations of 2 intelligence tests with different normative bases, the meta-analytic mean was 2.31, 95% CI [1.99, 2.64], standard score points per decade. The mean effect size for 53 comparisons (N = 3,951, excluding 3 atypical studies that inflate the estimates) involving modern (since 1972) Stanford-Binet and Wechsler IQ tests (2.93, 95% CI [2.3, 3.5], IQ points per decade) was comparable to previous estimates of about 3 points per decade but was not consistent with the hypothesis that the Flynn effect is diminishing. For modern tests, study sample (larger increases for validation research samples vs. test standardization samples) and order of administration explained unique variance in the Flynn effect, but age and ability level were not significant moderators. These results supported previous estimates of the Flynn effect and its robustness across different age groups, measures, samples, and levels of performance.

Overview

Integrity signals

Registry

Crossref · 10.1037/a0037173

Cites

268 of 268 references verified0 citation errors · 0 not checkable

Statistics

21 tests recomputed · 0 with the reported and recomputed p on opposite sides of .05

Verification

Claims the paper makes

0 of 34 cited sources checked · 0 supported · 0 flagged

Show all 81 claims

  • It may state two separate claims as one. 42
  • Probably not a substantial claim. 35
67 of 67
  • The paper relies on this without support.

    Research on the Flynn effect has focused almost exclusively on the effect produced from administrations of the Stanford-Binet and Wechsler tests.

    “Research on the Flynn effect has focused almost exclusively on the effect produced from administrations of the Stanford-Binet and Wechsler tests.”
  • The paper relies on this without support.

    Validation studies conducted by test publishers likely employed the most rigorous procedures with regard to sampling, selection of administrators, and adherence to administration and scoring protocols.

    “Validation studies conducted by test publishers likely employed the most rigorous procedures with regard to sampling, selection of administrators, and adherence to administration and scoring protocols.”
  • The paper relies on this without support.

    The 27 multifactorial tests chosen represent the major IQ tests in use over the past few decades.

    “There may be other multifactorial tests to consider, but the 27 we chose represent the major IQ tests in use over the past few decades.”
  • The authors state this as an assumption.

    The results of test validation studies provide important information irrespective of the findings and would therefore constitute publishable data.

    “A formal search for unpublished studies was not undertaken; it was presumed that the results of test validation studies would provide important information irrespective of the findings and would therefore constitute publishable data.”
  • The paper relies on this without support.

    The greater impact of the Flynn effect on fluid versus crystallized intelligence is well-established.

    “because the greater impact of the Flynn effect on fluid versus crystallized intelligence is well-established.”
  • The paper relies on this without support.

    The Flynn effect is a cohort phenomenon.

    “For the study of a cohort phenomenon like the Flynn effect, this approach is appropriate.”
  • The paper relies on this without support.

    The convention in society is to use a cut point of 70 for intellectual disability.

    “Because the convention in our society is to use a cut point of 70”
  • The paper relies on this without support.

    No comparable data exist for American young men.

    “Unfortunately, no comparable data exist for American young men.”
  • The paper relies on this without support.

    The Raven’s test administered to Scandinavian young men has not changed in format or content since its development.

    “Whereas the Raven’s test administered to Scandinavian young men has not changed in format or content since its development”
  • The paper relies on this without support.

    The Armed Services Vocational Aptitude Battery administered to potential conscripts in the United States has changed in format or content.

    “this is not the case for the Armed Services Vocational Aptitude Battery”
10 of 10

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Statistics that don’t add up

Each reported p-value recomputed from its test statistic and degrees of freedom.

  • Inconsistent p-value: z = 3.60, p < .0001

    Minor

    The paper reports p < .0001 here; recomputing from the test statistic and its degrees of freedom gives p < .001 (two-tailed). The check cannot tell whether the statistic or the p-value is the misprint.

    Reported p < .0001 · recomputed p < .001

  • p-value fits an unstated one-tailed test: z = −3.37, p < .0007

    Minor

    The paper reports p < .0007 here; recomputing from the test statistic and its degrees of freedom gives p < .001 (two-tailed). The reported value is what a one-tailed test gives, but the paper does not say near this result that a one-tailed test was used.

    Reported p < .0007 · recomputed p < .001

2 of 2

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.

  • Doesn’t add upz = 3.6reported p < .0001 · recomputed p < .001

    “z = 3.60, p < .0001”

  • Doesn’t add upz = -3.37reported p < .0007 · recomputed p < .001

    “z = −3.37, p < .0007”

  • Not confirmedQ(1) = 3.14reported p < .076 · recomputed p = .076

    “Q(1) = 3.14, p < .076”

  • Consistentz = 14.1reported p < .0001 · recomputed p < .001

    “z = 14.10, p < .0001”

  • ConsistentQ(284) = 4710reported p < .0001 · recomputed p < .001

    “Q(284) = 4710, p < .0001”

  • ConsistentQ(5) = 231reported p < .0001 · recomputed p < .001

    “Q(5) = 231, p < .0001”

  • Consistentz = 2.58reported p < .01 · recomputed p = .010

    “z = 2.58, p < .01”

  • Consistentz = -1.08reported p < .28 · recomputed p = .280

    “z = −1.08, p < .28”

  • Consistentz = 11.12reported p < .0001 · recomputed p < .001

    “z =11.12, p < .0001”

  • ConsistentQ(151) = 3237reported p < .0001 · recomputed p < .001

    “Q(151) = 3237, p < .0001”

  • Consistentz = 0.21reported p < .84 · recomputed p = .834

    “z = 0.21, p < .84”

  • ConsistentQ(16) = 232reported p < .0001 · recomputed p < .001

    “Q(16) = 232, p < .0001”

  • Consistentz = 0.19reported p = .85 · recomputed p = .849

    “z = 0.19, p = .85”

  • ConsistentQ(33) = 295reported p < .0001 · recomputed p < .001

    “Q(33) = 295, p < .0001”

  • Consistentz = 6.13reported p < .0001 · recomputed p < .001

    “z = 6.13, p < .0001”

  • ConsistentQ(11) = 44reported p < .0001 · recomputed p < .001

    “Q(11) = 44, p < .0001”

  • ConsistentQ(13) = 74reported p < .0001 · recomputed p < .001

    “Q(13) = 74, p < .0001”

  • Consistentz = 10.06reported p < .00001 · recomputed p < .001

    “z = 10.06, p < .00001”

  • ConsistentQ(55) = 597.34reported p < .0001 · recomputed p < .001

    “Q(55) = 597.34, p < .0001”

  • Consistentz = 16.83reported p < .0001 · recomputed p < .001

    “z = 16.83, p < .0001”

  • ConsistentQ(377) = 5581reported p < .0001 · recomputed p < .001

    “Q(377) = 5581, p < .0001”

21 of 21

Self-check

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

Each contradiction between two passages listed here was confirmed by an independent check.

  • Forty-year WISC-R example applies only nine points of adjustment

    Minor

    The example spans 1973 to 2013, or four decades. At the paper's approximately three-point-per-decade rate, the adjustment would be about 12 points, yielding a mean near 112 and a cut point near 82, not 109 and 79.

  • Comparisons are counted as studies in the abstract

    Minor

    The methods and results identify 285 qualifying comparisons extracted from 154 empirical studies and 27 manuals, rather than 285 distinct studies.

  • WISC-R norming year changes from 1972 to 1973

    Minor

    The paper uses 1972 as the WISC-R norming year in its worked example, but calls the same test standardized in 1973 in the conclusion.

3 of 3

Citations

Cites

Flagged first, then by citations

Show all 262 references

Research works printed but not linked: 17

  1. Doll B Boren R Performance of severely language-impaired students on the WISC-III, language scales, and academic achievement measures Journal of Psychoeducational Assessment, Advances in Psychological Assessment Monograph Series 1993 77 86
  2. Edwards BR Klein M Comparison of the WAIS and the WAIS-R with Ss of high intelligence Journal of Clinical Psychology 1984 40 1 300 302
  3. Elley WB Changes in mental ability in New Zealand New Zealand Journal of Educational Studies 1969 4 140 155
  4. Flynn JR Flores-Mendoza C Colom R Efeito Flynn: Repensando a inteligência e seus efeitos [The Flynn effect: Rethinking intelligence and what affects it] Introdução à psicologia das diferenças individuais [Introduction to the psychology of individual differences] 2006b Porto Alegre, Brasil ArtMed 387 411 (English trans. available from jim.flynn@stonebow.otago.ac.nz).
  5. Flynn JR What is intelligence? 2007 Cambridge Cambridge University Press
  6. Klanderman J Devine J Mollner C The K-ABC: A construct validity study with the WISC-R and Stanford-Binet Journal of Clinical Psychology 1985 41 2 273 281
  7. McGinley P A comparison of WISC and WISC-R test results The Irish Journal of Psychology 1981 1 23 24 doi:
  8. Phelps L Leguori S Nisewaner K Parker M Practical interpretations of the WISC-III with language-disordered children Journal of Psychoeducational Assessment WISC-III Monograph 1993 71 76
  9. Quereshi MY McIntire DH The comparability of the WISC, WISC-R, and WPPSI Journal of Clinical Psychology 1984 40 4 1036 1043
  10. Robinson EL Nagle RJ The comparability of the Test of Cognitive Skills with the Wechsler Intelligence Scale for Children-Revised and the Stanford-Binet: Fourth Edition with gifted children Psychology in the Schools 1992 29 2 107 112
  11. Rothlisberg BA Comparing the Stanford-Binet, Fourth Edition to the WISC-R: A concurrent validation study Journal of School Psychology 1987 25 2 193 196
  12. Thorndike RL Hagen EP Sattler JM Stanford-Binet Intelligence Scale, Fourth Edition, Technical manual 1986 Chicago, IL The Riverside Publishing Company
  13. Wechsler D Wechsler Intelligence Scale for Children, Third Edition, Manual 1991 San Antonio, TX The Psychological Corporation
  14. Wechsler D Wechsler Adult Intelligence Scale, Third Edition, Technical manual 1997 San Antonio, TX The Psychological Corporation
  15. Wechsler D Wechsler Preschool and Primary Scale of Intelligence, Third Edition, Technical manual 2002 San Antonio, TX The Psychological Corporation
  16. Wechsler D Wechsler Adult Intelligence Scale, Fourth Edition, Technical manual 2008 San Antonio, TX The Psychological Corporation
  17. Zimmerman IL Woo-Sam J The utility of the Wechsler Preschool and Primary Scale of Intelligence in the public school Journal of Clinical Psychology 1970 26 4 472

Cited by

Sorted by citations

Metadata

Authors
Trahan, L. H., Stuebing, K. K., Fletcher, J. M., & Hiscock, M.
Affiliations
University of Houston (US)
Venue
Psychological Bulletin
Published
2014
Volume
140
Issue
5
Pages
1332-1360
ISSN
  • 0033-2909 (ISSN-L)
  • 1939-1455
OpenAlex
W2164636756
Cited by
364
References
262
Citations by year
2026: 24 · 2025: 38 · 2024: 33 · 2023: 30 · 2022: 36 · 2021: 56 · 2020: 34 · 2019: 20 · 2018: 32 · 2017: 16 · 2016: 20 · 2015: 22 · 2014: 2
FWCI
27.60
Type
journal article
Language
en
Open access
Locations
  • Psychological Bulletin (publishedVersion)
  • PubMed (publishedVersion)
  • PubMed Central (acceptedVersion)
Funders
  • Novartis Foundation
  • University of Cambridge
  • Eunice Kennedy Shriver National Institute of Child Health and Human Development (P50HD052117)
Topics
  • Social Sciences › Psychology › Experimental and Cognitive Psychology › Cognitive Abilities and Testing
  • Social Sciences › Decision Sciences › Management Science and Operations Research › Psychometric Methodologies and Testing
  • Social Sciences › Sociology and Political Science › Intergenerational and Educational Inequality Studies
Keywords
Flynn effect, IQ trends, Stanford-Binet, measurement error, effect size, validation samples
MeSH terms
Capital Punishment, Education, Special, Humans, Intelligence, Intelligence Tests, Intellectual Disability

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.

A language model drew the claims from the paper on 2026-11-03, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-11-03 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

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