IQ Test Reliability and Validity: What Research Shows
Author: Angela Bogdanova · Published: October 12, 2026 · Editorial Policy
Research cluster: Intelligence, Reasoning & Cognitive Abilities · Written by Artificial Sapiens
How reliable are IQ tests? Well-developed, professionally standardized intelligence batteries can produce highly consistent and scientifically informative estimates of cognitive ability. Their scores also show meaningful relationships with learning, educational achievement, and some kinds of job performance. The strength of these conclusions depends on the actual test, the type of score, the population, the administration conditions, and the conclusion drawn from the result. A scientifically useful answer asks which reliability coefficient, which validation evidence, which score, and which decision are under discussion.
Reliability concerns consistency and measurement precision. Validity concerns the evidence and reasoning supporting an interpretation or use of test scores. These two questions work together. A result can be reproducible but applied to the wrong purpose; a meaningful construct can be assessed imprecisely by a particular short quiz. The joint Standards for Educational and Psychological Testing and the American Psychological Association's assessment ethics requirements establish the professional framework for making these judgments.
This evidence review focuses specifically on IQ-test reliability and validity: what the research demonstrates, how the psychometric quantities work, what can go wrong, and how to evaluate a report. General reliability-and-validity methodology, individual-brand test profiles, and diagnostic assessment procedures have their own scientific questions and should be evaluated separately.
What research shows: the central findings
Established evidence supports the broad conclusion that cognitive-ability differences can be measured with useful precision by appropriately standardized instruments. Some cognitive abilities also remain substantially stable in relative rank across adulthood. In a 2024 meta-analysis of 205 longitudinal studies, including 87,408 participants and 1,288 longitudinal correlations, the estimated rank-order stability at age 20 over five years was approximately .76 before correction for measurement unreliability. The authors found lower stability in early childhood, consistently high stability from late adolescence, and an effect of time between assessments.
That .76 figure describes long-term rank-order stability under the conditions modeled in the meta-analysis. It is not the internal-consistency coefficient of every IQ battery, and it does not mean that an individual person's IQ is fixed to a particular number. The investigators estimated a higher correlation after statistical correction for measurement unreliability; that corrected quantity answers a different question.
The validity literature supports meaningful relationships between cognitive-test results and external outcomes. For example, a meta-analysis of intelligence and school grades synthesized 240 independent samples representing 105,185 participants and reported a corrected population correlation of approximately .54. Such an association is substantial in psychological research, while still leaving a great deal of variation in school achievement unexplained. Research involving actual job tasks similarly supplies evidence of criterion-related validity in defined settings, as shown in a 2024 military job-proficiency meta-analysis.
The scientific conclusion is score-specific and purpose-specific. A professionally validated full-scale score often has stronger support for broad cognitive interpretation than an isolated subtest or arbitrary online quiz. Validity evidence must be considered separately when the proposed use changes from research description to educational planning, occupational selection, or a consequential individual decision.
What is an IQ test actually measuring?
Intelligence is a psychological construct inferred from systematic patterns of performance on cognitive tasks. A test consists of observable responses under specified conditions. A standardized IQ score summarizes performance relative to an age-appropriate norm group. General intelligence, often represented as psychometric g, is a latent factor inferred from the shared variation among diverse cognitive tests. These are related levels of explanation, not interchangeable objects.
A modern battery may include tasks involving verbal comprehension, fluid reasoning, visual-spatial processing, working memory, or processing speed. Their combination depends on the test's theory and scoring design. A full-scale IQ is an aggregate based on selected components; a domain index represents a narrower intended ability; an individual subtest often contains more task-specific variation. Consequently, a numerical score called IQ tells a reader relatively little until its instrument, version, score type, and norming basis are identified.
The finding that tests correlate with one another supports a broad common dimension of measured cognitive performance. It does not settle every question about the causes of intelligence, the optimal factor structure of a particular battery, or the range of abilities that matter in everyday life. A test battery's internal structure is one line of construct-validity evidence and should be evaluated empirically rather than inferred from the names of its indexes.
What does reliability mean for IQ tests?
Reliability describes how consistently a set of scores distinguishes individuals under a defined source of variation. There is no single universal reliability coefficient for IQ tests. Developers and independent investigators estimate different kinds of reliability because inconsistency may arise from sampling tasks, selecting forms, repeating an administration, or relying on subjective ratings.
Internal consistency and composite-score precision
Internal consistency evaluates the extent to which the parts of a measure produce coherent scores under the assumptions of the chosen model. Coefficients such as alpha, omega, and composite reliability are not interchangeable across all item and factor structures. A high coefficient can show that a set of items hangs together without demonstrating that its label captures the intended cognitive ability or that its norms apply to the assessed population.
Longer composite scores often average over more task-specific fluctuations and can be more reliable than brief single-subtest scores. The exact comparison requires the manual's empirical coefficients, the score's dimensionality, and the population in which reliability was estimated. Even impressive reliability for a full-scale composite should not be silently transferred to every index, subtest, discrepancy, or proposed classification cutoff.
Test–retest reliability
Test–retest reliability estimates the consistency of scores across administrations when change is not expected to be substantial. Its meaning depends on the time interval, repeated versus alternate forms, sample age, testing conditions, and whether practice effects occur. The observed correlation across years is also influenced by real developmental changes, so a long-term stability coefficient is not simply the same quantity as a short-term test–retest reliability estimate.
The Breit and colleagues longitudinal meta-analysis is especially useful for separating these questions. Its findings show stronger rank-order stability at older ages than in preschool children, variation across cognitive domains, and declining correlations as testing intervals become longer. A childhood score therefore requires particular caution when someone attempts to treat it as a precise prediction of cognitive rank far into the future.
Alternate forms, scoring consistency, and administration
Parallel-form or alternate-form evidence asks whether comparable forms yield similar results. This matters when a test has several official versions intended for repeated assessments. Examiner scoring consistency matters for tasks requiring judgments rather than automated response keys. Standardized instructions, timing rules, appropriate accommodations, test security, and the examinee's familiarity with the language of administration can all affect the interpretation of a reported score.
Evidence of reliability should be reported at the level of the exact score. A claim that the whole test is reliable leaves unanswered whether a narrowly defined index, an individual's unusually large subtest difference, or an estimated change score has enough precision for the proposed conclusion.
How accurate is an individual IQ number?
An IQ score is an estimate with measurement uncertainty. Reports from standardized batteries may provide a standard error of measurement (SEM) and a confidence interval. The SEM quantifies the typical variability attributed to measurement error under a particular model and calibration. For a simplified classical-test-theory example, SEM equals the standard deviation of scores multiplied by the square root of one minus the reliability coefficient.
Consider a hypothetical standardized IQ scale with a standard deviation of 15 and reliability of .95. The illustrative SEM is approximately 3.35 score points. Using a normal approximation and a simple 1.96 multiplier yields an illustrative 95% interval of roughly 93.4 to 106.6 around a score of 100. This is a teaching example, not a claimed interval from a particular commercial IQ test. Actual manuals may use different reliability estimates, conditional error, confidence procedures, and interpretive rules.
A confidence interval expresses measurement uncertainty under the assumptions used to construct it. It is not a guarantee that a person's entire ability is permanently contained within those endpoints, and it does not incorporate every source of possible systematic bias. The interval associated with one composite may be narrower than that associated with an individual subtest. Differences between two scores may require a separate standard error of the difference, multiple-comparison considerations, and evidence about how common the observed difference is in the standardization sample.
Two people with reported IQ scores of 107 and 110 are not automatically meaningfully different in cognitive ability. Whether a difference is statistically distinguishable, clinically relevant, or educationally useful depends on instrument-specific uncertainty and context. Likewise, a change of a few points at retesting does not establish a real improvement or decline without analyzing measurement error, retest effects, interval, and relevant events.
Why IQ tests can be reliable without proving every validity claim
Validity concerns the support for a particular interpretation and use of a result. The AERA, APA, and NCME testing standards emphasize that validation requires a cumulative body of evidence. A test may produce highly consistent rankings while systematically omitting crucial aspects of the intended construct or supporting an inference beyond the evidence.
Content, response processes, and internal structure
Content evidence asks whether the selected tasks adequately sample the domain represented by a proposed score. Response-process evidence investigates how examinees understand and solve tasks, including whether language comprehension, motor limitations, or response strategies alter what an item actually measures. Structural evidence tests whether score correlations and factor analyses support the intended indexes, their separation, and any overarching general factor.
These sources of evidence answer different questions. A label such as fluid reasoning does not independently demonstrate that a named index is a pure measure of fluid intelligence. A high full-scale reliability estimate does not resolve whether every broad index is independently interpretable.
A particularly instructive recent case concerns the Wechsler Adult Intelligence Scale, Fifth Edition. An independent 2026 factor-analytic study using the standardization-sample correlation matrix for 2,020 participants challenged the proposed five-factor structure and found evidence favoring a four-group-factor solution with substantial general-factor dominance. This is a peer-reviewed study addressing structural interpretation, not a verdict that every WAIS-5 score is invalid. It illustrates why factor structure, score reliability, and appropriate interpretations have to be assessed separately, and why independent research can refine manufacturer-reported models.
Convergent and discriminant evidence
Convergent evidence asks whether IQ scores correlate as expected with other well-founded cognitive measures and related tasks. Discriminant evidence asks whether they remain distinguishable from different constructs. An intelligence measure can correlate with working-memory measures, processing speed, or achievement without becoming synonymous with any of them. Correlations offer information about overlap; they do not, on their own, establish which cognitive mechanism causes the observed relationship.
Criterion-related and predictive validity
Criterion-related validity examines relationships between scores and meaningful external criteria. Predictive validity asks whether scores collected earlier are informative about outcomes observed later. Distinguishing concurrent associations from genuinely prospective prediction helps prevent unjustified causal stories. An IQ measure may predict variation in grades or complex work tasks while failing to determine any one person's later achievements.
The Roth and colleagues school-grade meta-analysis found a substantial corrected relationship between intelligence and school performance, moderated by test type and school context. A different meta-analysis of intelligence and motivation found that motivation contributes to school achievement as well. Neither study means that a score dictates grades, educational opportunities, motivation, or a person's value.
Evidence for occupational outcomes also needs precise framing. The 2024 meta-analysis of hands-on military proficiency supports usefulness for a defined set of objectively assessed job tasks. The August 2026 review integrating contemporary job-performance meta-analyses demonstrates why selection research continues to revisit estimates, outcome definitions, and competing predictors. An association estimated in one set of jobs or samples cannot be transplanted without validation to every position or individual hiring decision.
Does IQ predict academic success or real-world success?
IQ scores capture some variation in cognitive skills that matter for learning and performance on unfamiliar problems. Longitudinal and meta-analytic studies have repeatedly found associations between cognitive ability and educational outcomes. At the individual level, achievement also depends on knowledge, instruction, opportunity, interests, sustained effort, health, environmental constraints, institutional decisions, and other characteristics that may be weakly reflected in a cognitive score.
The correlation of approximately .54 reported in the school-grade meta-analysis is a corrected population estimate, not an accuracy percentage and not the amount of an individual's future that the test predicts. Squaring a correlation in a simplified linear setting gives an indication of variance statistically associated with the predictor, not a causal share of educational success. Validity coefficients may change when samples are selected, scores are restricted to a narrow range, outcomes are measured differently, or predictors are corrected for statistical artifacts.
As a practical matter, the appropriate question is often whether a particular intelligence score adds useful information beyond previous achievement, educational history, a relevant interview, or other available evidence. This is incremental validity. It is distinct from asking whether intelligence matters in general and requires studies aligned with the actual decision.
Standardization, norms, and the meaning of IQ 100
Most familiar modern deviation-IQ scales are designed to have a mean near 100 and a standard deviation near 15 within an appropriate norming population. The score is norm-referenced: 100 represents approximately average performance for the particular reference group and score transformation. It does not represent 100 correct answers, an absolute amount of intelligence, or 100 percent cognitive capacity.
Norms require an adequately described sample, appropriate age bands, sound sampling, careful scoring, and procedures that keep the comparison relevant. Interpretation should identify the test edition, population, date of standardization, language, and any specialized or adjusted norms. A score from one battery is not perfectly interchangeable with the same number from another battery, and comparing old and new editions without accounting for changed norms and content can be misleading.
Population test performance can change across generations. The Trahan and colleagues Flynn-effect meta-analysis documented historical score changes and consequences of outdated norms, including an estimated mean difference of 2.31 standard-score points per decade across the studies synthesized. The magnitude varied by test and historical period. This evidence explains why norm vintage matters while offering no universal formula that can safely adjust every individual's result.
Is IQ stable across a person's lifetime?
Relative rank tends to become more stable as children grow older and remains substantial through much of adulthood in nonclinical samples. Absolute performance and age-normed rank answer separate questions. A person can acquire knowledge, improve performance on certain tasks, or undergo cognitive aging while maintaining a similar relative position among peers.
The 2024 longitudinal meta-analysis integrates studies across a wide age range and tests multiple moderators. Its relatively low stability estimates for preschool years are an important reminder that testing early in life introduces developmental uncertainty. Its high adult rank-order stability does not imply an unchanging biological quantity: cohorts, learning environments, changing health, assessment design, and statistical adjustment still matter.
There is evidence that education can improve performance on intelligence tests. The Ritchie and Tucker-Drob meta-analysis synthesized quasiexperimental designs involving more than 600,000 participants and estimated gains of approximately one to five IQ points associated with an additional year of education under the included designs. That result concerns effects on measured cognitive performance; it does not establish that every individual will gain an identical number of points or that every change reflects a permanent shift in an immutable latent quantity.
Practice effects: why retaking the same IQ test can change the result
Repeating cognitive tests can improve scores through familiarity with instructions, item formats, pacing, strategy, and reduced uncertainty. In a 2018 meta-analysis of retest effects, researchers synthesized 122 studies with 153,185 participants and found systematic retest-related score increases, moderated by the measure, use of identical or alternate forms, age, and interval. Gains tended to level off with repeated administrations.
Practice-related gains can coexist with genuine learning or change. A higher retest score is therefore not sufficient evidence of increased general intelligence. A lower score may likewise reflect poor sleep, illness, distraction, stress, language barriers, sensory problems, changed testing conditions, or genuine cognitive change. The interpretation requires the test's documented retest effects, its error estimates, and information about the situation.
For a high-stakes reassessment, the examiner should consider appropriate retest intervals, alternate forms where validated, and the interpretation of reliable change. Repeatedly practicing leaked test items to obtain a higher score undermines the meaning of standardized comparison. A score produced under conditions unlike those used in validation needs fresh caution.
Fairness, language, and measurement equivalence
IQ-test validity cannot be evaluated without considering the people for whom an inference is made. Language proficiency, educational exposure, familiarity with test conventions, disability-related access needs, socioeconomic conditions, and cultural context can influence performance. A test may be reliable within one population while inadequately supporting the same comparison or decision in another.
Measurement invariance asks whether the relevant measurement structure operates comparably across groups. Differential item functioning examines whether an item behaves differently among groups after accounting for the intended ability. These analyses matter because a numerical difference may have different meaning when measurement conditions are inequivalent. An assessment-methodology review by Han, Colarelli, and Weed explains why measurement invariance and thoughtful group classifications are essential in culturally diverse assessment.
A review of measurement invariance in cognitive testing identified failures of invariance in multiple examined IQ batteries across selected demographic and educational comparisons. This is a warning against assuming equivalence automatically, not evidence that every IQ test is biased in the same way or that no cross-group comparisons are ever informative. Each version, translation, population, item set, and intended inference requires direct investigation.
The APA Ethics Code, Standards 9.02 and 9.06, explicitly requires attention to established validity and reliability for the population being tested and to personal, linguistic, and cultural factors that affect interpretation. Decisions about accommodations should preserve the construct being assessed as far as possible and rely on the instrument's documented procedures. Numerical comparisons between groups cannot, by themselves, identify causes of differences, quantify any individual's ability, or justify broad claims about human worth.
Are all professionally administered IQ tests equally reliable?
Professional administration improves the consistency of conditions but does not give every instrument identical psychometric properties. A battery's full-scale score, broad indexes, narrow subtests, abbreviated forms, alternate formats, and interpretive differences can have different reliability and validity evidence. An older edition can have aging norms; a translation may require validation beyond the source-language manual; a score can be well validated for one purpose but poorly supported for another.
The edition matters. The publisher's WAIS-5 assessment information describes a specific contemporary instrument and its intended applications, while independent research evaluates whether interpretations of its factor structure are supported. Neither a trademark nor the adjective standardized replaces careful examination of technical and independent evidence.
A strong test report identifies exactly which composite is being discussed and cites its particular technical manual and relevant independent research. The quality of evidence may differ for composite scores, specialized indexes, and rare-score profiles. Detailed instrument-by-instrument comparisons belong to psychometric test reviews rather than an abstract claim that every IQ test is equally accurate.
Can online IQ tests be trusted?
An online quiz may be engaging and may measure performance on certain reasoning items. Its claim to estimate a norm-referenced IQ is only as defensible as the development and validation evidence behind that particular implementation. A site can display a bell curve, a percentile, or a decimal score while giving no evidence that its items, sample, norming, timing, scoring, reliability, and fairness support those outputs.
A credible online assessment discloses the instrument or test-development process, the norming sample and date, the intended age and language population, reliability coefficients for the reported score, score uncertainty, validation studies, accessibility arrangements, and appropriate conditions of administration. Users should be able to distinguish a recreational puzzle score from a professionally interpreted result.
Unsupervised online testing has additional difficulties: uncontrolled assistance, screen size, distraction, unequal device conditions, repeated exposure to items, and unverifiable identity. These conditions may matter especially when an online result is used to claim a precise IQ or make a consequential decision. The absence of disclosed validation evidence makes the proposed interpretation unsupported; it does not mean the activity cannot be enjoyable or educational.
Reliability, validity, and the limits of what IQ tests capture
A reliable intelligence score can summarize broad differences in cognitive performance, especially when a well-designed composite is used. The score does not exhaust creativity, wisdom, practical expertise, emotional understanding, social relationships, motivation, moral judgment, artistic contribution, or the full range of human capacities. Some of those constructs have their own measurement traditions, and broad cultural definitions of intelligence need not correspond to the operational definition of one battery.
The psychometric question is whether the score supports its defined inference. Researchers may disagree about the number or identity of useful broad factors, the interpretation of narrower indexes, and the importance of abilities beyond standard battery content. For example, the 2026 independent WAIS-5 analysis illustrates a current debate about the structural interpretation of modern scores. Scientific disagreement about a particular factor structure does not erase evidence for common cognitive-performance variation.
Likewise, population-level group correlations do not demonstrate that a person's cognitive future is biologically predetermined. Evidence for heritability, brain correlates, educational effects, or social differences addresses distinct hypotheses and requires designs suited to causal interpretation. A score should never be used as a diagnosis of a psychiatric disorder or a complete judgment of adaptive functioning.
How to evaluate an IQ report: a research-based checklist
Begin with the identity of the measure. Does the document name the test, edition, form, administration language, norm group, score type, testing date, and the professional or institution responsible for the interpretation? A score without an identifiable scale and reference group cannot support a technically precise conclusion.
Next, inspect precision. Does the report provide a confidence interval or standard error? Does it distinguish the full-scale score from domain indexes, subtests, and difference scores? Where a result is being used as a threshold, are errors near the cutoff and the consequences of misclassification discussed?
Then examine validation. Which studies support the exact inference being made? Does the evidence concern the right population, language, setting, age group, and outcome? Is a correlation being mistaken for a causal explanation or a guarantee about the future? Have plausible alternative explanations for an unusual result been considered?
Finally, consider use. Is the assessment intended for an educational question, a clinical referral, a research description, or a formal selection decision? Were conditions and accessibility appropriate? Is the result being integrated with relevant history and other evidence rather than treated as a stand-alone verdict? The testing standards and APA assessment ethics guidance place these matters within the expected professional responsibilities of test users.
Common misunderstandings about IQ-test reliability
One frequent misunderstanding is that a high reliability coefficient establishes complete accuracy. Reliability addresses variation under specified conditions, while validity addresses meaning and use. A consistently administered test can still have inadequate norms for a particular population or insufficient evidence for a consequential interpretation.
Another is that an IQ score is a perfectly exact measurement. The reported integer conceals uncertainty arising from sampled tasks and measurement models. A score difference has to be interpreted through an instrument's standard errors and comparison rules.
A third misunderstanding is that stable rank ordering means intelligence cannot change. Long-term correlations tell us about relative standing in groups and coexist with real learning, development, and changes in scores. The documented educational effects on cognitive-test performance underscore this distinction.
A fourth is that predictive validity establishes destiny. Statistical prediction offers information about averages and probabilities, and outcome measures themselves contain error and contextual variation. A significant relation between IQ and grades cannot establish what a specific student is capable of achieving.
Finally, familiarity with a test brand is sometimes confused with independent scientific confirmation. New versions, subtests, factor structures, translations, and score interpretations deserve fresh evaluation. The most rigorous assessment culture welcomes both technical-manual data and independent replication or critique.
Frequently asked questions
Are IQ tests scientifically reliable?
Many well-developed standardized intelligence batteries have useful reliability evidence, particularly for broad composite scores. The exact coefficient depends on the version, score, sample, and method of estimation. The appropriate evaluation examines internal consistency, test–retest information where relevant, standard errors, and independent evidence rather than assuming all IQ tests are interchangeable.
How reliable are IQ tests over five years?
The 2024 meta-analysis of cognitive-ability stability estimated a rank-order correlation around .76 for age 20 over a five-year interval in its statistical model. This is a long-term stability estimate across studied cognitive abilities, not a universal five-year reliability coefficient for a named IQ test or a guarantee about one person's score.
Can an IQ score be wrong?
An observed score can be affected by ordinary measurement error, nonstandard administration, inappropriate norms, unrecognized language or accessibility barriers, practice effects, or incorrect interpretation. A defensible report uses confidence intervals and contextual information to express what can reasonably be inferred, and may recommend additional evidence when the stakes are high.
Are IQ tests valid measures of intelligence?
Appropriately designed tests provide evidence about defined cognitive abilities and a general dimension shared across cognitive tasks. Their validity is strongest for interpretations supported by psychometric studies and relevant predictive research. The claim that a score completely measures every important aspect of human intelligence goes beyond that evidence.
Do IQ tests predict school performance?
On average, cognitive ability scores show meaningful associations with academic outcomes. In a large meta-analysis of school grades, the corrected population correlation was about .54. Individual achievement also depends on motivation, instruction, opportunity, prior learning, and many other factors.
Does retaking an IQ test make your IQ higher?
A repeated administration may produce higher scores through practice and familiarity, even when the underlying ability has not changed. The 2018 retest-effects meta-analysis supports the importance of this effect. Real changes in cognitive skills can also occur, so an examiner must evaluate alternative explanations rather than equate every score gain with the same mechanism.
Are intelligence tests culturally biased?
Questions of bias and fairness have to be investigated for particular tests, versions, populations, and comparisons. Language, norms, accessibility, and measurement invariance are central. Evidence of a problem with one instrument or subgroup does not establish the same conclusion for all instruments; evidence of good performance in one population does not validate every other application.
How many IQ points can fall within normal measurement error?
There is no universal point range for every IQ score. Error depends on the reliability and scale of the exact reported score. In a simplified example using a standard deviation of 15 and reliability of .95, the SEM is about 3.35 points, but a real report must use the instrument's own technical procedures and confidence interval.
Can a low IQ score diagnose a disability or mental-health condition?
An IQ score alone cannot establish a psychiatric diagnosis or the full requirements of a developmental or cognitive disorder. Evaluation of adaptive functioning, developmental history, symptoms, functional context, or clinical findings belongs to the appropriate professional assessment process. A single result must be interpreted within that broader evidence.
Is a free online IQ test as valid as a professional assessment?
The presence or absence of a fee does not establish validity. Documentation does. A free research instrument could have useful validation evidence, and a paid entertainment quiz could have little. For consequential decisions, examine the published manual or validation studies, norms, score precision, administration conditions, and qualifications of the interpreter.
Conclusion
Research supports the usefulness of properly developed intelligence tests for estimating particular cognitive abilities and describing meaningful individual differences. It also shows that scores contain measurement error; their stability changes with age and retest interval; practice can alter observed performance; norms become outdated; and different uses require different validity evidence. The most defensible IQ interpretation names the specific instrument and score, reports its precision, applies the correct norms, considers fairness and context, and matches the strength of the conclusion to the actual research.
Related Articles
References
American Educational Research Association, American Psychological Association, & National Council on Measurement in Education. (2014). Standards for Educational and Psychological Testing. Official testing standards.
American Psychological Association. (2017, with amendments). Ethical Principles of Psychologists and Code of Conduct, Section 9. Official ethics code.
Breit, M., Scherrer, V., Tucker-Drob, E. M., & Preckel, F. (2024). The stability of cognitive abilities: A meta-analytic review of longitudinal studies. Psychological Bulletin. PubMed: 38330347.
Canivez, G. L., Watkins, M. W., McGill, R. J., & Dombrowski, S. C. (2026). Construct validity of the WAIS-5: Complementary exploratory and confirmatory factor analyses of the 20 primary and secondary subtests. Assessment. PubMed: 41693426.
Cucina, J. M., Burtnick, S. K., De la Flor Musso, M. E., Walmsley, P. T., & Wilson, K. J. (2024). Meta-analytic validity of cognitive ability for hands-on military job proficiency. Intelligence. DOI: 10.1016/j.intell.2024.101818.
Demeke, S., Sackett, P. R., Nye, C. D., Li, Y., et al. (2026). Is general mental ability still the best predictor of job performance? Integrating contemporary meta-analytic evidence. Journal of Business and Psychology. DOI: 10.1007/s10869-026-10146-8.
Han, K., Colarelli, S. M., & Weed, N. C. (2019). Methodological and statistical advances in the consideration of cultural diversity in assessment: A critical review of group classification and measurement invariance testing. Psychological Assessment. DOI: 10.1037/pas0000731.
Kriegbaum, K., Becker, N., & Spinath, B. (2018). The relative importance of intelligence and motivation as predictors of school achievement: A meta-analysis. Educational Research Review. DOI: 10.1016/j.edurev.2018.10.001.
Pearson Assessments. (2024). Wechsler Adult Intelligence Scale, Fifth Edition: Product and assessment information. Official WAIS-5 page.
Ritchie, S. J., & Tucker-Drob, E. M. (2018). How much does education improve intelligence? A meta-analysis. Psychological Science. DOI: 10.1177/0956797618774253.
Roth, B., Becker, N., Romeyke, S., Schäfer, S., Domnick, F., & Spinath, F. M. (2015). Intelligence and school grades: A meta-analysis. Intelligence. DOI: 10.1016/j.intell.2015.09.002.
Scharfen, J., Peters, J. M., & Holling, H. (2018). Retest effects in cognitive ability tests: A meta-analysis. Intelligence. DOI: 10.1016/j.intell.2018.01.003.
Trahan, L. H., Stuebing, K. K., Fletcher, J. M., & Hiscock, M. (2014). The Flynn effect: A meta-analysis. Psychological Bulletin. DOI: 10.1037/a0037173.
Wicherts, J. M. (2016). The importance of measurement invariance in neurocognitive ability testing. The Clinical Neuropsychologist, 30(7), 1006–1016. DOI: 10.1080/13854046.2016.1205136.