CoChat
Sign Up
Cited by 0Cognitive Abilities and TestingPsychometric Methodologies and TestingSocial and Intergroup Psychology

The DOI matches its Crossref record; 264 of the 264 references it cites were found in a registry or an index; 23 of the 186 sources its claims cite were checked for support.

Factors Related to Mean Score Group Differences on Cognitive Ability Tests: A Systematic Review

Open access
Cuppello, Stephen, Zibarras, Lara D., Corr, Philip J.·2025·Trends in Psychology
DOI 10.1007/s43076-025-00504-5
Open this paper in CoChat — free accountOpen PDF

Abstract

Commonly used in employee selection, cognitive ability tests (CATs) show mean score group differences by gender, ethnicity and socioeconomic status (SES). Prior research supports both substantive accounts, such as biological and environmental factors, and methodological and contextual accounts, such as stereotype threat and differential item functioning, related to group differences. This systematic review explores the evidence for the theoretical explanations that purport to explain these differences. Two hundred and twenty-five papers met inclusion criteria which explored potential explanations of group differences. These were grouped into 10 factors: biological, environmental, latent trait / measurement invariance, criterion validity, item bias, behaviour in test, anxiety, attitudes, experience, and stereotype threat. Results showed that whilst biological and environmental factors are related to group differences, so are several factors relevant to test design and administration, most notably stereotype threat, differences in domain experience and exposure and self-confidence. Theoretical and practical implications for the occupational use of CATs as well as limitations and recommendations are discussed.

Overview

AI snapshot

Written from the paper’s full text.

TL;DR

This systematic review of 225 papers identifies biological and environmental factors, plus test-design and administration factors—especially stereotype threat, domain experience/exposure, and self-confidence—as related to mean score group differences on cognitive ability tests (CATs). It concludes that methodological and contextual factors offer promising avenues for fairer occupational CAT use, but the evidence is limited by few high-stakes occupational studies and heavy reliance on student and US samples.

Key findings

  • Across 225 papers grouped into 10 factors, biological (63 papers; 67% impact “yes”) and environmental (15; 93% “yes”) factors were supported, but so were test-design/administration factors: stereotype threat (75; 81% “yes”), experience (18; 72% “yes”), and attitudes/self-confidence (23; 52% “yes”).
  • Stereotype threat had the largest evidence base (75 papers; 51 gender, 27 ethnicity, 7 SES), with supported mechanisms including working-memory/executive-resource depletion and interpretation of anxious arousal; interventions included stating no group differences, role models, self-affirmation, and mindfulness. However, few studies replicated effects in high-stakes occupational settings.
  • Experience/exposure findings showed training, practice tests/items, and video games improved spatial performance more for women and sometimes eliminated the gender gap; practice/training improved Black performance more than White performance. Self-confidence mediated gender–spatial performance, and manipulating confidence increased performance.
  • Item bias/DIF exists but generally had negligible impact on overall group mean differences; criterion validity findings were inconsistent; latent trait/measurement invariance findings were discordant. Only 14% of studies met ecological validity criteria, 71% used student samples, 56% used US samples, and average MMAT score was 43%.

Methods

  • PRISMA-P systematic review, not pre-registered. Searched PsycINFO, PsychArticles, Web of Science, and Business Source Ultimate with broad terms; backward and limited forward citation searching (first 20 papers and first 10 from the last five years per paper).
  • Inclusion: adult humans (mean age 18+), original research, English peer-reviewed; excluded descriptive-only papers, bias-measurement-only papers, clinical/cognitive-decline samples, and tests of knowledge/crystallised intelligence or standardised academic tests such as SATs. Initial 1,124 records; 316 duplicates; 541 abstract-excluded; 140 excluded; 127 fully reviewed; plus 98 from citations = 225 papers.
  • Data extraction and synthesis by the first author, reviewed by coauthors; papers tagged into 10 factors; MMAT used for risk of bias; narrative synthesis; extracted “Impact” (yes/no/maybe) and “Ecological Validity” (yes/no/partially).

Limitations

  • Authors state: excluded child samples and high-stakes standardised education testing; broad scope and heterogeneity sacrificed specificity; gender/sex and ethnicity/race terms plus global inclusion may overlook nuances; screening, extraction, and synthesis were completed by one person; forward citation searching was not expansive.
  • Text-evident: only 14% of studies met ecological validity criteria; 71% used student samples; 56% used US samples; only 8% included any participants outside North America/Europe and 4% exclusively outside those regions; average MMAT was 43%, with common risks (confounders accounted for only twice in nonrandomized studies; 18% representative; RCT blinding 10%, baseline comparability 11%).
  • Age differences could not be reviewed: only two papers measured age and neither found reliable causal evidence; only 44 of 225 papers addressed multiple factors, and these rarely provided consistent interaction evidence.

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

Integrity signals

Registry

Crossref · 10.1007/s43076-025-00504-5

Cites

264 of 264 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Verification

Claims the paper makes

23 of 186 cited sources checked · 17 supported · 1 flagged · accuracy not yet measured

Verify all 1 references — free account

Show all 162 claims

  • Probably not a substantial claim. 98
  • It may state two separate claims as one. 73
  • The quote only partly states this claim. 1
148 of 148

  • The authors state this as an assumption.

    For this review, participants' identified gender and the biological sex they were assigned at birth are assumed to be the same.

    “No papers in this review discuss differences between participants' identified genders and the biological sex they were assigned at birth, and so these will be assumed to be the same.”

  • The authors state this as an assumption.

    A lack of studies investigating a factor for a demographic group indicates a gap in the literature, not that the factor is irrelevant to that group.

    “If for example several compelling studies were found supporting a particular factor's contribution to gender score differences on CATs, the lack of studies investigating ethnicity would not indicate that this factor is not relevant, only that there is a gap in the literature.”

  • The paper relies on this without support.

    The workplace is not experienced in the same way by different groups of people.

    “These studies are clearly very important, but even when looking at objective criteria, the workplace is not experienced in the same way by different groups of people.”

  • The paper relies on this without support.

    University students are typically young, educated and more likely to be from higher SES backgrounds.

    “University students are typically young, educated and more likely to be from higher SES backgrounds.”

  • The paper relies on this without support.

    CATs are highly predictive of work success.

    “Whist CATs are highly predictive of work success, their use will result in differing selection rates for groups that exhibit mean score differences.”

  • The paper relies on this without support.

    Recruitment tests are high stakes, applicants are aware they are being evaluated, and they are aware of the competitive context.

    “Recruitment tests are high stakes, applicants are aware they are being evaluated, and they are aware of the competitive context.”
6 of 6

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.

  • Anxiety-stereotype-threat overlap is both denied and tabulated

    Minor

    The discussion says no paper investigating test anxiety examined stereotype threat, but Table 3 lists two papers addressing anxiety, attitudes, and stereotype threat together.

  • Environmental impact percentages exceed 100%

    Minor

    The environmental row assigns 93%, 7%, and 7% to the three impact categories, totaling 107%. These are presented as the alternative values of a single impact classification.

2 of 2

Citations

Cites

Items 201–264 of 264

Flagged first, then by citations

Research works printed but not linked: 4

  1. Jensen, A. R. (1980). Bias in Mental Testing. Free Press.
  2. * McCormick, C. M., & Teillon, S. M. (2001). Menstrual cycle variation in spatial ability: Relation to salivary cortisol levels. Hormones and Behavior, 39(1), 29–38. https://doi.org/10.1207/s15327043hup1603_3 · DOI
  3. *Penner, A. M., & Willer, R. (2011). Stigma and glucose levels: Testing ego depletion and arousal explanations of stereotype threat effects. Current Research in Social Psychology, 16(3), n3.
  4. Thurstone, L. L. (1938). Primary mental abilities. University of Chicago Press.

Related

See all

Metadata

Authors
Cuppello, S., Zibarras, L. D., & Corr, P. J.
Affiliations
City St George's, University of London (GB)
Corresponding author
Cuppello, Stephen
ORCID
Venue
Trends in Psychology
Publisher
Springer Science and Business Media LLC
Published
2025-11-11
ISSN
2358-1883 (ISSN-L)
DOI
10.1007/s43076-025-00504-5
OpenAlex
W4416115091
Semantic Scholar
7ea697cd0a92d42df34ce2d970d7a8ed8e5e0e42
Cited by
0
References
258
FWCI
0.00
Type
journal article
Language
en
Open access
Licence
Locations
  • Trends in Psychology (publishedVersion, cc-by)
  • City Research Online (City University London) (acceptedVersion)
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 › Social and Intergroup Psychology
Keywords
cognitive ability tests, group differences, stereotype threat, differential item functioning, measurement invariance, self-confidence, test anxiety, test administration, personnel selection, gender differences, ethnicity, socioeconomic status, biological and environmental factors

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 2026-10-29, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-29 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (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