Intelligence and Job Performance: Predictive Value and Limitations
Author: Angela Bogdanova · Published: October 12, 2026 · Editorial Policy
Research cluster: Intelligence, Reasoning & Cognitive Abilities · Written by Artificial Sapiens
Does IQ predict job performance? Cognitive ability is associated with work performance across many jobs, especially because reasoning and learning help people acquire job-relevant knowledge. The best contemporary synthesis, however, estimates a substantially smaller relationship than the value most often repeated in older summaries. An August 2026 meta-analysis by Demeke and colleagues integrated 212 samples including 109,396 participants and proposed a mean operational validity of approximately r = .20 for general mental ability (GMA) and overall job performance, with estimates of .18–.22 across analyses.
That result supports a measured conclusion: a well-developed cognitive ability assessment provides some information about expected work performance at the population level. It does not determine who will excel, does not capture the whole range of occupational competence, and cannot independently establish that a particular applicant is suitable for a job. This article explains what intelligence and job performance mean, how the predictive evidence has changed, why different studies report different numbers, when the relationship matters, and how to interpret fairness and measurement limitations.
What Does “Intelligence Predicts Job Performance” Mean?
In occupational psychology, prediction refers to a statistical relationship between a score collected at one point and a relevant work outcome, often measured later. A higher ability score can be associated with higher average performance across a population without allowing an assessor to infer any individual employee’s eventual output with certainty. Correlation coefficients summarize how two measurements covary in a specified sample; they are not probabilities of success, hiring guarantees, or percentages of a person’s ability.
General mental ability refers to broad shared variance among cognitive performances involving reasoning, learning, comprehension, and complex problem solving. The construct is often associated with the psychometric general factor, g. An intelligence test score is an observed result influenced by the particular instrument, its norms, the conditions of testing, and measurement error. An IQ score is a standardized score within a specific reference population; the work-performance evidence usually concerns GMA measures rather than one universal IQ instrument. Our guide to general intelligence explains the psychometric construct in depth.
A general intelligence measure, an aptitude test, a knowledge examination, a numerical reasoning subtest, and a work simulation can overlap while measuring different combinations of learned knowledge and cognitive abilities. Performance prediction requires specifying the actual predictor and the actual outcome. The comprehensive SIOP Principles for the Validation and Use of Personnel Selection Procedures emphasize that the validity of inferences depends on the proposed use of a selection procedure.
The Main Research Finding: A Real but Modest Contemporary Association
Why the traditional .50 estimate became so influential
For decades, influential personnel selection reviews reported an intelligence–job-performance validity near .50. The widely cited Schmidt and Hunter (1998) review summarized older evidence and placed general mental ability among the strongest available predictors of performance and training success. Readers often encountered the .51 value as though it described every occupation, test, work outcome, era, and applicant population. The underlying estimates were based on psychometric correction procedures and historically assembled datasets, not a direct observation of a universal coefficient.
The methodological question is consequential. A correlation in employed workers may be lower than one in all applicants because hiring restricts score variation; supervisor ratings also contain measurement error. Statistical corrections attempt to recover a population relationship from incomplete information. Their defensibility depends on the design, known selection process, and actual reliability data. They may increase a coefficient, but a corrected coefficient is still an estimate conditional on modeling assumptions.
The 2022 and 2024 reassessments
Sackett, Zhang, Berry, and Lievens (2022) reexamined widely used personnel selection validity estimates and argued that certain historical corrections for restriction of range had been applied too liberally. Their recalculation reduced the operational estimate for GMA to approximately .31 in the older database, and many other selection methods also showed lower corrected validities. They found structured interviews particularly competitive. The precise correction model remains a topic of technical debate, but the central discovery was that the apparent accuracy of several hiring predictors had been overstated.
A later meta-analysis by Sackett and colleagues (2024) deliberately concentrated on research conducted in the current century. Its 153 samples and 40,740 individuals produced a mean observed GMA–overall-performance correlation of .16 and a corrected value of .22, based on the study’s treatment of criterion unreliability and range restriction. The observed .16 and corrected .22 are different statistics. Reporting one in place of the other erases the methodological issue that the article was designed to investigate.
The integrated 2026 estimate
Demeke, Sackett, Nye, Li, and Kuncel (2026) combined unique primary studies from three modern meta-analytic databases while removing overlapping samples. Across 212 samples involving 109,396 people, their operational validity estimates clustered between .18 and .22, leading to a recommended summary estimate of .20. The paper explicitly recommends treating the contemporary findings separately from data collected in earlier periods.
The contemporary evidence therefore supports an association considerably below the old .51 benchmark. It does not establish that intelligence has ceased to matter. The relevant scientific claim is narrower and clearer: scores on validated general cognitive ability measures have an average, modest positive relationship with commonly assessed overall job performance, and the strength varies with study design, job setting, measures, and sampling.
How Should We Interpret r = .20?
A correlation of .20 is positive, but it represents considerable overlap in performance outcomes between people with higher and lower scores. For illustration, the square of .20 is .04, or 4%. This is the squared bivariate coefficient in a simple standardized linear relationship under the assumptions of that calculation; it is not a claim that an IQ test literally determines four percent of each individual’s productivity, nor that only four percent of intelligence matters in any occupation. Corrected meta-analytic validity and an employer’s observed predictive R-squared also should not be silently equated.
Suppose two applicants have different reasoning-test scores. The result may shift their expected average performance conditional on the relevant population and job criterion. Both applicants still have broad ranges of possible outcomes, influenced by experience, training quality, opportunities, work conditions, motivation, support, professional judgment, and the requirements of the role. A moderate difference in test scores can coexist with an inverse difference in eventual work performance.
A useful predictive model should be judged against the alternatives and the added information it contributes. If a company already knows candidates’ relevant work samples, structured interview results, licenses, and technical expertise, an ability score may add less than it would in a model with no other information. This is incremental validity: the gain in prediction when a new measure is added to an existing set of measures. The relative usefulness of a score is therefore inseparable from the selection context.
What Researchers Actually Measure as Job Performance
“Job performance” names a multidimensional family of outcomes. The work-performance framework reviewed by Campbell and Wiernik (2015) distinguishes role-relevant behaviors and multiple performance domains. A single omnibus rating may blend dimensions that have different predictors and may obscure differences across occupations.
Task performance and technical effectiveness
Task performance concerns completing core duties to an appropriate standard: accurate decisions, production quality, technical output, problem resolution, timeliness, and completion of assigned responsibilities. It might be assessed through supervisor ratings, behavioral examples, error rates, objective output, or standardized simulation. The appropriate criterion for a software engineer differs from one for a nurse, teacher, mechanic, or customer-support specialist. Even within an occupational label, jobs differ by organization and level.
Training success and acquisition of job knowledge
Learning a new procedure, integrating complex information, or mastering technical material creates a plausible route from cognitive ability to success at work. General ability may be particularly informative about the ease or speed of acquiring knowledge when tasks are unfamiliar. Training grades, examination results, and later real-world job performance are distinct outcomes. A predictor that forecasts training performance well cannot automatically be assumed to forecast teamwork, leadership, safety, or customer satisfaction equally well.
Contextual, adaptive, and interpersonal performance
Coworker assistance, reliable cooperation, professional communication, and contribution to a team represent contextual performance. Adaptive performance includes adjusting to changed tools, procedures, or operational demands. Intelligence may help with parts of adaptation and difficult decisions, while the expression of these abilities also depends on learned skills, incentives, behavior, and collaborative processes. The 2026 integrated meta-analysis discusses how the composition of modern work and the outcomes employers evaluate have changed.
Performance ratings are measurements, too
Supervisor ratings may reflect access to observable work, different standards, rater perspectives, performance opportunities, and criteria that were not fully specified. Zhou, Sackett, Shen, and Beatty’s updated meta-analysis of supervisory rating reliability (2024) addresses a measurement problem central to occupational validity research. When the outcome itself is noisy, predictor–criterion correlations require interpretation against that uncertainty. Using multiple appropriate sources of performance evidence can make a validation study more informative.
Why Might Cognitive Ability Predict Work Outcomes?
The leading psychological account proposes an indirect route through learning and application of job knowledge. Reasoning can support understanding instructions, recognizing patterns, connecting new information with prior knowledge, resolving unfamiliar situations, and building expertise. In turn, knowledge and practiced skills support effective task behavior. This account is consistent with established models of individual performance, including Campbell and Wiernik (2015), but statistical correlations alone do not identify the strength of every causal pathway.
Fluid reasoning is relevant to solving unfamiliar problems, crystallized knowledge to drawing on accumulated concepts and vocabulary, and domain-specific skills to using these capacities appropriately in work. Processing speed and working memory are related cognitive constructs, but each differs from overall intelligence. A general factor can summarize correlated differences while narrower abilities matter for particular tasks. The CHC intelligence framework provides one influential model for understanding those levels.
The mechanism also depends on opportunities. A worker can have strong reasoning skills and still lack necessary job-specific preparation, usable equipment, clear objectives, access to mentoring, or authority to implement a solution. Well-designed tasks and training can allow a broader range of people to demonstrate competence. Occupational outcomes are produced within workplaces, not only within individuals.
Is IQ More Predictive in Complex Jobs?
Older reviews commonly concluded that intelligence predicts performance more strongly in cognitively complex positions. There are plausible reasons: complex work often requires learning novel relationships, integrating information, and responding to unfamiliar problems. Yet this broad claim has become less secure when tested against contemporary data. The relationship depends on what “complexity” means and how work is classified.
In their 2026 integrated synthesis, Demeke and colleagues examined O*NET-derived occupational demands grouped into Data, People, and Things. They found little relationship between the Data-based measure of cognitive complexity alone and variation in the validity of GMA, a positive association with the Things dimension, and higher validity where Data and Things demands were jointly elevated. These findings challenge a simple rank-order story in which GMA inevitably predicts success most strongly wherever a job seems most intellectual.
One possible interpretation is that earlier complexity classifications captured several demands at once and relied heavily on older industrial job categories. The authors also consider shifts in work toward interpersonal interdependence. Their moderator analyses remain observational analyses of occupations and study estimates, rather than experimental proof of why validity differs. Contemporary evidence supports evaluating actual job demands rather than assigning a universal “high complexity equals high IQ validity” rule.
General Ability Versus Specific Cognitive Skills
A GMA score summarizes broad cognitive performance. Specific scores may describe verbal comprehension, numerical reasoning, spatial abilities, mechanical reasoning, or learned occupational knowledge. Two people with similar general scores may have different profiles of particular abilities. The value of a specific assessment depends on whether it measures a capability genuinely relevant to the work and adds useful information beyond general tests.
Nye, Ma, and Wee (2022) synthesized the links between cognitive abilities and several work criteria and examined whether narrower abilities could supply predictive information above GMA. Their findings indicate that some narrower abilities can add useful information under particular conditions. This does not imply that every narrow subtest predicts every job or that building a large battery of tests automatically improves accuracy.
For example, spatial visualization may be relevant to a demanding technical task, while a language-intensive position requires specific comprehension and communication skills. A valid professional assessment must follow the task demands and the evidence. The general question is how the ability is operationalized, whether the test has adequate reliability and validity for that population, and whether the added prediction justifies the time and consequences of testing.
IQ Versus Experience, Personality, Interviews, and Work Samples
Intelligence is one member of a larger network of performance-related characteristics and circumstances. Prior education, occupation-specific knowledge, practical skills, experience, interests, conscientiousness, interpersonal behavior, and workplace support can all matter. Which measure deserves priority depends on whether the purpose is to predict initial learning, complex task work, sustained performance, promotion, or behavior under changing demands.
The recalibration of selection-method estimates has affected comparisons. Sackett and colleagues (2022) identified structured interviews among the strongest methods in their updated analysis. Highly structured interviews use consistent questions and predefined scoring criteria, helping employers gather comparable evidence. Interviews can assess job-relevant constructs that cognitive tests do not capture well. Findings about the success of one kind of interview should not be exported to improvised conversations or informal impressions.
Work samples and realistic exercises may measure demonstrated ability to carry out relevant tasks, but their predictive value depends on sampling representative job behaviors and constructing meaningful scores. Such exercises can also privilege applicants who have already had access to a particular tool or procedure. The choice between testing, training, and direct work demonstration requires an explicit definition of what the role demands at entry and what can reasonably be learned after hire.
Personality research supplies a complementary example. A meta-analysis by Watrin and colleagues (2023) found a modest association between conscientiousness and job performance while warning that much evidence relied on incumbent and concurrent designs rather than actual applicant prediction. For the separate question of traits, see our published Big Five and job performance review. The present article retains its own focus on intelligence as a predictor.
Selection Accuracy and Fairness: What the Evidence Says
A predictor can be statistically useful and still demand careful scrutiny before it is used to allocate employment opportunities. Technical issues include whether the construct is relevant to essential tasks, whether test scores operate comparably across populations, whether the process is accessible, and whether a selection rule has adverse impact. Predictive validity, equitable treatment, and legal justification are separate questions that must be considered together.
The updated selection meta-analytic matrix by Berry, Lievens, Zhang, and Sackett (2024) challenged the assumption that removing general cognitive tests from a particular selection composite necessarily entails a large loss in predictive accuracy. Their modeling found that excluding GMA from evaluated selection systems often had little or no effect on estimated composite validity while substantially reducing adverse impact. This conclusion concerns their modeled procedures, predictors, subgroup differences, and assumptions. It is not proof that every real hiring process improves when all cognitive tests are removed.
The EEOC guidance on employment tests and selection procedures explains that selection tools may raise discrimination concerns if they disproportionately exclude people in protected groups and lack adequate justification under applicable law. Employers should validate tools for the roles and purposes for which they are used, examine less discriminatory alternatives when relevant, and revise procedures as job requirements change. These are legal and practice considerations; this article is an evidence interpretation resource rather than individualized legal advice.
Disability-related accessibility is essential. The U.S. Department of Justice guidance on algorithms and disability discrimination in hiring explains that digital tests and automated screening may measure sensory, motor, or communication barriers rather than the intended workplace ability. Appropriate accommodations and accessible assessment methods help distinguish what the test purports to assess from irrelevant obstacles introduced by its delivery.
Why Cognitive Test Results May Mislead
Sampling and restriction of range
People who are already employed in a profession may have passed several screening stages. Their ability-score variation can be smaller than that of the original applicant pool. If scores are restricted, a simple correlation calculated only among hired workers can misrepresent what would have been observed across a broader population. Correcting this restriction requires details about how applicants were selected; different statistical methods can yield different estimates. The disagreement around historical corrections is described in Sackett et al. (2022) and investigated across approaches by Demeke et al. (2026).
The meaning and reliability of the performance criterion
A manager’s overall rating may partly measure a worker’s contextual behavior and reputation, whereas an error-rate metric captures something narrower. Outcomes may be confounded by team resources, task allocation, customer mix, and opportunities to demonstrate skill. A study that reports a strong association with training grades does not thereby demonstrate a matching association with coworker support or career satisfaction. Researchers and employers should separate those criteria when the decision requires it.
Timing, test quality, and local validity
An observed score also depends on instrument reliability, language, testing conditions, test security, practice effects, and the fit between the test’s reference population and the person assessed. A score from an unvalidated online IQ quiz is particularly weak evidence for employment decisions. Even well-developed instruments have finite precision. See IQ: What Intelligence Quotient Measures and What It Means for the distinct question of standardized IQ interpretation.
Generalization across occupations and eras
Meta-analyses combine heterogeneous tasks, predictors, and methods. A grand mean is useful for orienting the field but does not automatically transport to a particular company or emerging occupation. The 2026 synthesis explicitly highlights how historical samples differ from contemporary jobs, including the growing interpersonal dimensions of work. Evaluations should consider whether a cited study resembles the actual decision being made.
Does a High IQ Guarantee Career Success, Earnings, or Leadership?
No particular IQ score guarantees promotion, excellence, leadership, favorable earnings, or job satisfaction. These are separate outcomes shaped by educational pathways, social opportunities, industry, institutions, experience, chance, occupational choices, working conditions, and individual preferences. Correlations between intelligence and one career outcome do not license substituting a different outcome without direct research. A statistically favorable mean expectation still leaves wide variation among people with the same score.
Performance, career success, compensation, and psychological well-being should be modeled separately. An excellent technical employee may choose a career that offers autonomy rather than income growth. Someone may perform exceptionally well in a position requiring strong social coordination and domain mastery without unusually high general cognitive test scores. Workplaces should assess demonstrated qualifications and actual outcomes rather than constructing personality narratives around a number.
Do Training, Education, and Workplace Tools Change the Relationship?
Training improves knowledge and task proficiency when it targets useful skills and provides adequate practice and feedback. Employees acquire job-specific expertise, and organizations can improve instructions, reduce unnecessary complexity, establish clear checklists, supply decision aids, and build effective peer support. These interventions can affect performance even if a person’s standardized IQ score remains unchanged. Improving work results is a distinct goal from raising psychometric g.
Digital tools and AI systems increasingly redistribute portions of work among people, software, and automated decision processes. That shift changes which behaviors organizations observe and reward. As the 2026 research review notes, changes in occupational tasks are relevant when interpreting the past and future validity of GMA. Specific claims that new AI tools will increase or erase the role of intelligence in every job are preliminary until tested directly across occupations and time.
Short-term cognitive exercises may produce improvements on practiced tasks without strong evidence of broad transfer to unrelated work outcomes. For the distinct evidence question concerning far transfer and intelligence, see our brain training and intelligence analysis. Hiring organizations can learn more by validating actual task performance than by assuming that a generalized cognitive-training product will transform occupational aptitude.
What Should an Applicant Take From This Research?
A cognitive screening score represents performance on a bounded set of tasks at a particular time. It should not become an identity label or a verdict about professional potential. If a job uses testing, applicants can reasonably seek information about which capabilities are measured, how the score is used, whether accessible administration is available, and how the assessment relates to the job’s essential responsibilities. Demonstrated work, experience, learning, judgment, and professional growth remain relevant.
Preparing for a real occupational demand is more useful than interpreting an IQ-style number as a destiny. Learn the tools and concepts required by the role, practice meaningful work samples, develop communication and documentation habits, and request feedback aligned with the position. When performance is currently difficult, examine instructions, training opportunities, workload, and resources alongside personal skill requirements.
What Should an Employer or Researcher Take From This Research?
Start with a defensible description of essential work activities and outcomes. Choose selection evidence that reflects those activities, specify whether success means training proficiency or later performance, and compare a cognitive test with structured interviews, relevant credentials, work samples, and other established predictors. Then test whether the added measure improves prediction enough to justify its consequences.
A careful validation plan examines reliability, predictive or concurrent design, the applicant population, correction assumptions, transportability across roles, accessibility, and possible differential outcomes. Report observed and corrected associations separately. Treat subgroup fairness and legal obligations as part of the decision rather than as a footnote after implementation. The SIOP/APA selection principles and EEOC guidance provide authoritative starting points for that work.
No single test should be presented as a complete model of an employee. The appropriate question is whether a documented, job-relevant inference from an assessment has enough evidence to support a particular decision, considering practical alternatives and the people affected.
Common Misconceptions About Intelligence at Work
“IQ explains half of job performance.” The famous historical validity near .50 was a correlation estimate, not “50% of performance explained.” Contemporary meta-analysis provides a much lower summary estimate near .20, and its interpretation still depends on corrections and outcomes.
“Smart people always perform better.” Group-level relationships describe averages across samples. Employees with different ability scores have extensively overlapping outcomes, and job success depends on multiple interacting conditions.
“Intelligence is useful only in highly complex white-collar jobs.” Modern moderator findings do not support a simple monotonic rule based on cognitive complexity alone. Occupations differ in the relevant combination of data, people, and physical demands.
“If a test predicts performance, its use is automatically fair.” Criterion-related validity and fairness are distinct requirements. Accessibility, applicable law, consequences, measurement equivalence, and less exclusionary alternatives need explicit consideration.
“A lower score indicates a clinical problem.” Workplace cognitive scores are occupational assessment data. They do not constitute a diagnosis of intellectual disability, ADHD, dementia, or any other disorder. Clinical evaluation has separate standards and purposes.
“The latest finding proves IQ never matters.” The modern average remains positive, while its magnitude is smaller than traditionally assumed. Researchers should update quantitative beliefs without converting revised estimates into categorical claims.
Frequently Asked Questions
Does IQ predict job performance?
Yes, cognitive ability scores predict some variation in average work performance across people and jobs. The newest integrated meta-analysis estimates the contemporary general mental ability–overall job performance validity at about .20. A correlation of that size is meaningful for research yet offers limited certainty about individual success.
What percentage of job performance does IQ explain?
In an elementary one-predictor linear model, squaring a correlation of .20 gives .04, equivalent to 4% of variance in the modeled outcome. This is a mathematical property under specified assumptions, not a universal claim about an individual’s productivity, a guarantee of test accuracy, or a causal percentage. Estimates also differ depending on whether a correlation is observed or psychometrically corrected.
Is intelligence the best predictor of employee success?
The answer depends on the outcome, available alternatives, and whether “best” refers to a single test or a multivariable selection system. Updated meta-analyses rank structured interviews competitively, and modern selection research questions the older assumption that GMA must dominate every predictive model.
Is IQ more important than hard work or experience?
That comparison combines different constructs. IQ measures standardized cognitive-test performance; effort is behavioral and context-dependent, while experience may represent both accumulated opportunities and acquired expertise. Their predictive contributions depend on the role and criterion. Research does not provide a universal exchange rate among them.
Does job complexity increase the predictive value of IQ?
Older findings suggested that it often did. Contemporary analyses have found little systematic moderation by data-oriented complexity alone, with more complicated results for manual and combined demands. The most relevant evidence comes from job-specific validation, not an occupational stereotype.
Can an employer legally use an IQ test to hire people?
Employment tests are subject to applicable federal, state, and local requirements. In the United States, anti-discrimination and disability protections are relevant, and a procedure may require evidence that its use is job-related and consistent with business necessity. Consult the official EEOC selection guidance and qualified legal and assessment professionals for a particular process.
Can someone with average IQ be an exceptional employee?
Yes. Standardized intelligence scores do not set a ceiling on performance, professionalism, creativity, accumulated knowledge, collaboration, or specialized expertise. Population correlations allow extensive overlap among employees’ actual outcomes.
Do online IQ tests predict success at work?
An online quiz with undocumented norms, reliability, validity, and scoring procedures should not be treated as an occupational predictor. A commercially delivered cognitive test also requires evidence supporting the intended job-related inference, including accessible administration and responsible use.
Conclusion
Intelligence is a measurable contributor to differences in learning and some work outcomes. The central scientific development is quantitative: contemporary evidence places the average general cognitive ability–job performance association substantially below the historically dominant estimate. Demeke and colleagues’ 2026 integration offers a defensible present-day benchmark near .20, with meaningful variation by study and occupational conditions.
The strongest interpretation combines that predictive relationship with knowledge of real work, fair assessment, and the limits of inference from any single score. Intelligence research helps explain differences in cognitive performance; responsible occupational assessment asks what evidence is useful for a specified task, decision, and population.
Related Articles
References
Berry, C. M., Lievens, F., Zhang, C., & Sackett, P. R. (2024). Insights from an updated personnel selection meta-analytic matrix: Revisiting general mental ability tests’ role in the validity–diversity trade-off. Journal of Applied Psychology. PubMed.
Campbell, J. P., & Wiernik, B. M. (2015). The modeling and assessment of work performance. Annual Review of Organizational Psychology and Organizational Behavior, 2, 47–74. https://doi.org/10.1146/annurev-orgpsych-032414-111427.
Demeke, S., Sackett, P. R., Nye, C. D., Li, Y., & Kuncel, N. R. (2026). Is general mental ability still the best predictor of job performance? Integrating contemporary meta-analytic evidence. Journal of Business and Psychology. https://doi.org/10.1007/s10869-026-10146-8.
Nye, C. D., Ma, J., & Wee, S. (2022). Cognitive ability and job performance: Meta-analytic evidence for the validity of narrow cognitive abilities. Journal of Business and Psychology, 37, 1119–1139. https://doi.org/10.1007/s10869-022-09796-1.
Sackett, P. R., Demeke, S., Bazian, I. M., Griebie, A. M., Priest, R., & Kuncel, N. R. (2024). A contemporary look at the relationship between general cognitive ability and job performance. Journal of Applied Psychology, 109, 687–713. https://doi.org/10.1037/apl0001159.
Sackett, P. R., Zhang, C., Berry, C. M., & Lievens, F. (2022). Revisiting meta-analytic estimates of validity in personnel selection: Addressing systematic overcorrection for restriction of range. Journal of Applied Psychology, 107, 2040–2068. https://doi.org/10.1037/apl0000994.
Schmidt, F. L., & Hunter, J. E. (1998). The validity and utility of selection methods in personnel psychology: Practical and theoretical implications of 85 years of research findings. Psychological Bulletin, 124, 262–274. https://doi.org/10.1037/0033-2909.124.2.262.
Society for Industrial and Organizational Psychology. (2018). Principles for the validation and use of personnel selection procedures (5th ed.). Industrial and Organizational Psychology, 11(S1), 1–97. https://doi.org/10.1017/iop.2018.195.
U.S. Department of Justice. (2022). Algorithms, artificial intelligence, and disability discrimination in hiring. https://www.ada.gov/resources/ai-guidance/.
U.S. Equal Employment Opportunity Commission. (2007). Employment tests and selection procedures. https://www.eeoc.gov/laws/guidance/employment-tests-and-selection-procedures.
Watrin, L., et al. (2023). The criterion-related validity of conscientiousness in personnel selection: A meta-analytic reality check. International Journal of Selection and Assessment, 31, 286–301. https://doi.org/10.1111/ijsa.12413.
Zhou, Y., Sackett, P. R., Shen, W., & Beatty, A. S. (2024). An updated meta-analysis of the interrater reliability of supervisory performance ratings. Journal of Applied Psychology, 109, 949–970. https://doi.org/10.1037/apl0001174.