Can People Fake a Big Five Personality Test? Response Bias and Social Desirability

7 days ago

20 min read

Author: Ukrainian Psychological Hub · Published: October 9, 2026 · Editorial Policy

A Big Five personality test can be faked because most widely used Big Five questionnaires ask people to describe themselves. A respondent can deliberately choose answers that create a preferred impression, especially when the assessment has consequences such as a hiring decision. Researchers have repeatedly demonstrated score changes under instructions to appear more desirable and have also observed differences between high-stakes and low-stakes administration. The size, direction, and consequences of those differences depend on the situation and the measure. Dunlop, Xia, and Anglim (2025) provide a current critical review.

The central scientific question is therefore more precise than whether faking is possible. It is whether the responses obtained on a particular Big Five instrument in a particular setting continue to support the interpretation and decision for which the scores are being used. Faking ability, actual faking behavior, detection accuracy, and predictive validity are different questions. A test can remain informative at the group level even when some respondents strategically manage impressions; an apparently favorable score cannot, by itself, reveal whether one individual has misrepresented their personality.

This guide examines the mechanisms behind distorted responses, what the major studies find, why high-stakes selection changes the problem, how test formats address it, and how individuals and organizations should interpret Big Five findings. For the broader five-domain framework, see our Big Five model guide.

Short Answer: Can You Fake a Big Five Personality Test?

Yes. People can intentionally select answers to look more conscientious, agreeable, emotionally stable, outgoing, or otherwise suited to a desired role. Because standard Big Five self-reports depend on the information supplied by the respondent, strategically favorable answers can change observed scores. An experimental ability to raise a score is not evidence that every job applicant does so or that all hiring scores are invalid.

A meta-analysis of 33 applicant-versus-nonapplicant studies found higher applicant averages on conscientiousness and emotional stability, with smaller average differences on extraversion and openness. These are differences between groups, not percentages of applicants caught lying. Birkeland and colleagues (2006) reported standardized differences of approximately 0.45 for conscientiousness, 0.44 for emotional stability, 0.11 for extraversion, and 0.13 for openness. The job being sought also influenced the pattern.

A later meta-analysis of 20 within-person comparisons, in which real applicants completed personality measures in both high- and low-stakes conditions, found moderately more favorable mean scores in the applicant condition, slightly less score variability, and substantial consistency in individuals' relative ordering. Hu and Connelly (2021) thus document both response distortion and the persistence of meaningful individual differences across contexts.

What Does Faking Mean in Big Five Assessment?

In personality assessment research, faking generally means deliberately altering self-presentation to obtain a desirable outcome. The respondent may know what impression an item creates and select an answer that serves a goal rather than best representing usual patterns of thinking, feeling, and behavior. Researchers also use terms such as intentional response distortion, motivated responding, and impression management, although particular studies define and operationalize these constructs differently.

The Big Five are continuous trait dimensions. They are not five personality types, diagnostic labels, virtues, or employment qualifications. A questionnaire estimates positions on measured domains or narrower scales from responses; it does not directly observe an enduring personality in every circumstance. Understanding these distinctions prevents a misleading leap from a test score to a judgment about a person's character. The Big Five hierarchy of domains, aspects, and facets describes levels of measurement, while our Big Five test guide explains what an instrument-specific result can establish.

Faking good, faking bad, and role-congruent self-presentation

Faking good is answering to appear more favorable according to a perceived criterion. In hiring, this often means emphasizing reliability, cooperation, and emotional steadiness. Faking bad means deliberately presenting oneself in an unusually unfavorable way; its motivations and consequences differ, and it is not the central pattern in ordinary hiring studies. Role-congruent impression management is more specific: a respondent adjusts answers to fit what they believe a particular role requires, rather than simply maximizing every seemingly positive item.

In a 2025 experiment, 231 participants completed the Big Five Inventory-2 honestly and then under instructions to present themselves favorably in either a job or dating context. Desirable responding appeared in both simulated contexts, but it was generally stronger for employment. This was an instructed-faking experiment; its numerical effects should not be treated as the prevalence of deception among actual applicants. See Rau, Schömann, and Grosz (2025).

Ordinary response bias is broader than deliberate deception

Not every difference between a person's self-report and an external observer's account is deliberate faking. Respondents may misunderstand wording, recall distinctive episodes instead of typical behavior, compare themselves with different reference groups, prefer extreme or middle response categories, or genuinely hold a more positive self-view. Acquiescence is a tendency to agree with items regardless of their meaning; extreme responding is a tendency to select scale endpoints. These are response styles, and the available evidence does not justify automatically calling them dishonesty.

Social desirability can also refer to the attractiveness of an item, a tendency to describe oneself favorably, or a score on a dedicated impression-management questionnaire. Those are connected but nonidentical operational definitions. They should be kept separate when reading research or evaluating an assessment report.

How Can Someone Distort Big Five Scores?

Typical Big Five items ask how accurately a short description reflects the respondent. Someone who wants to appear highly organized could endorse favorable descriptions of planning and follow-through more strongly than their routine behavior warrants. Someone hoping to appear socially energetic might downplay situations in which they prefer solitude. An assessor cannot distinguish these possibilities from genuinely high trait standing by inspecting a single favorable answer.

A distortion can involve changing a handful of salient responses, adopting an unrealistically consistent image across many questions, or choosing the most job-relevant response to each item. The effect on a domain score depends on item wording, whether the items are reverse-keyed, the scoring method, facet coverage, and the test's design. Different Big Five instruments are not interchangeable even when their five broad domains overlap.

Why the same trait can be presented differently in different roles

A sales applicant may infer that interpersonal energy is valued, while an applicant for a role emphasizing detailed procedural work may infer that orderliness and dependability are most important. These are examples of perceived requirements, not claims that any given trait predicts performance in every role. The pattern reported by Birkeland and colleagues varied by the target occupation, illustrating why one universal faking correction or fixed profile of the ideal applicant is inadequate. Original meta-analysis.

Why the Big Five is easier to influence than a knowledge test

Many cognitive or technical tests contain answers that can be objectively scored as correct or incorrect. A standard self-report personality scale instead asks for a judgment about the self. The respondent is the principal information source, so the administration context and incentive to present oneself favorably become part of the measurement problem. This does not make all self-reports inaccurate: carefully developed Big Five scales can capture systematic individual differences, particularly when interpretation fits the evidence and assessment purpose. Soto and John (2017) describe the construction and validation of the BFI-2.

Does Social Desirability Affect All Five Traits Equally?

No single adjustment applies across all five domains, and socially desirable answers depend partly on the evaluated role. Conscientiousness and emotional stability have shown notable mean shifts in applicant comparisons; the sizes and patterns for extraversion, openness, and agreeableness differ across studies, jobs, instruments, and instructions. Birkeland et al. (2006) and Hu and Connelly (2021) summarize two important kinds of evidence.

It is also too simple to label one pole of every Big Five dimension intrinsically better. High conscientiousness might be valued in some role profiles, but tests should still represent typical behavior and assess job relevance. High agreeableness or extraversion is not automatically appropriate for every task or working environment. Traditional Neuroticism and the BFI-2 domain Negative Emotionality have related conceptual content, while instrument-specific nomenclature and scoring must be respected. A low score on a particular trait never establishes that a respondent is dishonest, psychologically disordered, or morally inferior.

What Does the Scientific Evidence Actually Show?

1. Instructed faking studies show capability

In instructed-faking experiments, participants respond honestly and then are asked to look as favorable as possible, or they are assigned simulated stakes. Such designs reveal what respondents can do when motivated and help compare formats under controlled conditions. They can overestimate what routinely occurs when actual applicants balance impression management with plausibility, self-knowledge, instructions, and the uncertainty of real selection.

A meta-analysis comparing matched forced-choice and single-statement assessments reported average honest-to-faked score increases of d = 0.41 and d = 0.75, respectively. These are pooled standardized shifts under the included study conditions; they are neither individual probabilities nor universal adjustment constants. The authors also reported that forced-choice formats retained benefits under faking conditions, without becoming invulnerable. Speer et al. (2023).

2. Applicant studies show motivated response shifts in the field

Actual applicant comparisons are more relevant to operational testing, but they have other limitations. People entering different applicant and nonapplicant samples can differ in genuine traits as well as motivation. Within-person studies reduce some between-person confounding by comparing the same individuals across contexts, yet they still depend on administration order, timing, stakes, and the assumption that the trait has not materially changed. Hu and Connelly (2021) review 20 such within-person comparisons.

Because these research designs answer related but different questions, apparently conflicting effects can reflect differences in design rather than a simple disagreement about whether response distortion exists. A convincing interpretation considers the comparison condition, who was sampled, which inventory was used, what outcome was assessed, and how faking was operationalized.

3. Predictive usefulness can decline without disappearing

A major concern is whether motivated responding weakens the relation between personality scores and external outcomes such as performance. In a 2025 meta-analysis of 80 paired samples, Speer and colleagues found that criterion-related validity was lower in motivated than honest conditions on average across five-factor traits. The reported average differences between correlations ranged from approximately 0.05 to 0.08 in the examined comparisons, with variation by trait and study. The result supports caution about transporting low-stakes validity estimates into hiring without additional evidence; it does not imply that tests lose all predictive value. Speer and colleagues (2025).

A group-level validity coefficient describes an average statistical relationship under specific conditions. It does not permit an employer to calculate an individual's future conduct with certainty, and a score inflation effect alone does not establish how a particular hiring decision would have changed.

Can a Big Five Test Detect Whether Someone Is Lying?

There is no general-purpose Big Five lie detector. Researchers have developed social-desirability scales, improbably favorable response indicators, consistency checks, response-pattern methods, and model-based anomaly detection. Each method has to be validated for a defined instrument, respondent population, and decision context. A statistical flag is evidence about a response pattern, not a verified statement about motive or honesty.

Social-desirability or impression-management scales

Some inventories include items about behaviors that seem unusually flawless, seeking to identify exaggerated virtuous self-presentation. However, an unusually favorable pattern may reflect genuine self-perception, cultural norms, item interpretation, or an unusual but truthful individual. Equally, a strategic respondent might avoid extreme claims. This creates both false positives and false negatives.

In a 2024 empirical study using Big Five measures and the Marlowe–Crowne Social Desirability Scale, the authors found that individuals classified as suspect by the desirability scale often differed from those identified as having changed their Big Five responses under experimental faking conditions. They cautioned against interpreting the desirability scale as a precise detector of individual faking. Liu, Yan, and Zhang (2024).

Consistency and unusually polished profiles

A pattern that is uniformly positive can justify inspecting the testing process, but it is not an adequate basis for accusing a person of deceit. Consistency checks can be influenced by attention, literacy, fatigue, accessibility needs, item redundancy, and stable self-beliefs. Response time can likewise be affected by reading fluency, device, disability, and language proficiency. Algorithmic classification requires known sensitivity, specificity, false-positive rates, and independent validation before use in consequential decisions.

Why a high score cannot prove faking

Conscientiousness, agreeableness, and other traits are distributions of individual differences, so high scores are expected for some people. Without validated instrument-specific normative data, an assessor cannot apply a universal cutoff separating honest and dishonest responses. A percentile tells where an observed score falls in a reference distribution; it does not tell how truthfully the person responded. See our guide to Big Five scores and percentiles for the interpretation of reference groups.

Do Forced-Choice Personality Tests Prevent Faking?

Forced-choice assessments ask respondents to select or rank statements within a block rather than independently rate the desirability of every statement. When options have been carefully balanced on attractiveness, the respondent has less ability to endorse every socially favorable statement. However, forced-choice is a format, not a guarantee of integrity, and different forced-choice scoring models have distinct psychometric requirements.

A 2019 meta-analysis found comparatively small average score inflation across included forced-choice studies and reported that desirable-statement balancing and normative scoring were important moderators. Cao and Drasgow (2019). A subsequent meta-analysis found evidence of faking resistance but highlighted differences between experimental and real-world selection settings and among ipsative and quasi-ipsative designs. Martínez and Salgado (2021).

The matched-measures meta-analysis by Speer et al. (2023) showed lower average inflation in forced-choice than single-statement instruments when both were measured under comparable conditions. It also demonstrated that forced-choice responses could still be manipulated. Researchers and employers should examine evidence for the exact instrument, its scoring procedure, its construct coverage, and its population rather than assuming a format label settles validity.

Ipsative, quasi-ipsative, and normative interpretation

Some forced-choice methods primarily yield within-person comparisons: they express the relative prominence of one trait compared with another in the same profile. Normative approaches aim to support comparisons among people. These two interpretations answer different questions, and their scoring requirements matter especially in selection. A forced-choice inventory should not be described as producing directly comparable population scores unless the method's validation supports that interpretation.

Do Different Big Five Questionnaires Have Different Vulnerabilities?

Yes. A short classroom questionnaire, the Big Five Inventory-2, a NEO-based measure, an IPIP-based inventory, and a purpose-built personnel-selection assessment can use different items, item lengths, facet structures, response scales, administration conditions, and scoring rules. The fact that each aims to assess five broad trait domains says little by itself about vulnerability to impression management.

The BFI-2, for example, was developed as a hierarchical Big Five measure with five domains and fifteen facets, not the thirty facets associated with a NEO-PI-R/NEO-PI-3 formulation. Its psychometric design also addresses response style considerations. These facet structures are instrument-specific; no scientific rule requires all Big Five instruments to use exactly fifteen or thirty facets. Soto and John (2017).

A research finding obtained using a particular forced-choice employment inventory should not be transferred automatically to a generic free online quiz. Nor should a small facet-level difference on a long validated questionnaire be equated with the output of a two-minute quiz. An instrument's response bias evidence belongs alongside its reliability, validity, applicable norms, and intended use.

Does Faking Change Big Five Personality or Just the Test Score?

Changing questionnaire answers changes the reported measurement, not necessarily the underlying disposition. A trait concerns relatively enduring tendencies across situations and time, while a person's current motives, mood, role, and perceived consequences can alter responses in an assessment session. A higher conscientiousness score during an application could reflect genuine contextual behavior, deliberate embellishment, a different interpretation of the questions, or a mixture of these processes.

Repeated assessments can differ because of random measurement error, temporary conditions, practice, response strategies, and genuine trait change. These explanations need distinct study designs to separate. A sudden higher score is therefore neither proof of personal development nor proof of intentional distortion. Interpretation should use the instrument's standard error, test–retest evidence, administration context, and appropriately validated comparison data.

Can Respondents Answer Differently Without Realizing It?

Yes. People build self-judgments from memory and context. Someone answering as a student may consider deadlines and classwork; the same person answering as a team member may think of conflict management or collaboration. They may supply genuinely different answers without deciding to deceive. Item ambiguity, the reference period, and social comparison can also change the perceived meaning of a description.

A useful distinction is among intentional misrepresentation, self-deceptive or overly favorable self-perception, and ordinary response-style effects. Different studies assess these phenomena differently, and no single psychometric indicator cleanly assigns an individual to one category. Independent behavioral records or observer ratings can add information, although they have their own blind spots and contextual limitations.

What Happens When Applicants Fake Personality Tests for Jobs?

Hiring creates an incentive to choose answers that appear aligned with a job description. This can inflate some scale means, compress variation among applicants, alter rank ordering, and weaken correlations with later outcomes. Whether those effects meaningfully damage a particular decision procedure depends on the exact assessment, occupational criterion, applicant pool, selection ratio, scoring, and the way other evidence is combined.

The most responsible organizational question is whether a validated assessment adds useful information about a clearly job-related requirement, given the costs of bias and error. An employer should not treat one trait as a universal marker of competence and should not convert a questionnaire into a clinical judgment. Dunlop and colleagues (2025) review safeguards and trade-offs; Speer et al. (2025) quantify average validity attenuation under motivated responding.

Fairness, privacy, and individual harm

Faking countermeasures can themselves be problematic when they classify truthful respondents as deceptive or penalize people whose response styles differ across languages, cultures, disability contexts, or testing environments. Employers should specify the data collected and its purpose, minimize access, provide appropriate accommodations, and examine subgroup performance and error rates in the intended use population. The fact that an algorithm provides a risk score does not turn an uncertain inference into an established fact.

Personality tests in hiring should form part of a documented, job-related, proportionate assessment process, not an automated gatekeeper that substitutes a personality stereotype for evidence. The broader topic of assessment fairness belongs to workplace assessment and psychometrics; the specific focus here is how Big Five self-report distortion changes score interpretation.

Which Anti-Faking Strategies Have Scientific Support?

Forced-choice design and desirability-balanced items

These approaches have the strongest body of comparative evidence among commonly discussed format-level options, but results depend on item construction, forced-choice model, and scoring. Matched comparisons and meta-analyses support benefits under some motivated conditions rather than complete immunity. Cao and Drasgow (2019); Speer et al. (2023).

Instructions and faking warnings

Honesty instructions may signal the assessment's purpose, and some warning interventions reduce motivated score inflation in research samples. A 2025 meta-analysis of warning types reported an average reduction in faking of d = 0.31, with considerable heterogeneity and weaker results in some applied recruitment samples. The evidence does not justify a claim that warnings reliably detect dishonest individuals or remove all bias. Moon and colleagues (2025).

Multimethod assessment

When the decision stakes warrant it, combining a validated Big Five self-report with structured, job-related work samples or other appropriate information can avoid making one self-description the sole evidence source. Observer reports can reveal discrepancies worth interpreting, but differences may reflect access to different behavior rather than deception. Any integration of measures still requires validation for the decision being made.

Statistical corrections and automated detectors

Some studies investigate adjustments based on social desirability scales, or detection via unexpected response patterns. These approaches need external validation showing that they improve relevant outcomes and do not produce harmful errors in the specific setting. As the Liu et al. (2024) findings illustrate, an ostensibly objective suspicion score may track the wrong individuals. A correction that makes averages look more typical is not automatically more accurate for a particular applicant.

Can AI Help Someone Fake a Big Five Test?

Generative AI can produce answers that appear tailored to a requested persona or role, so AI assistance introduces a new source of intentional response construction. A 2026 experiment found role-sensitive shifts across Big Five inventory domains in model-generated responses; that result concerns how a tested language model responded under prompts, not the personality of the model or the detection of a human respondent's genuine traits. Treglown and Furnham (2026) The underlying distinction is among generated test responses, Big Five construct measurement, and claims about real-world human behavior.

A 2026 study directly stress-tested single-statement and forced-choice HEXACO inventories with humans and AI-generated response sets. Some tested AI models obtained highly favorable applicant-condition scores, and no examined format universally prevented distortion. This was HEXACO-specific evidence with a defined participant sample and model set; it should not be presented as proof that every Big Five instrument is equally affected or that AI assistance always succeeds. Robie and colleagues (2026).

The practical implication is methodological: test vendors should evaluate AI-assisted response conditions for the precise inventory, verify out-of-sample performance, assess false-positive risk, and maintain appropriate privacy safeguards. Neither generative fluency nor an automated detection score establishes that a system knows a person's inner personality.

How Should Someone Answer a Big Five Test Accurately?

Answer with your usual behavior across relevant situations and a realistic time frame, rather than choosing what appears most flattering. Think about representative patterns rather than a single remarkable day. If an item is unclear, follow the administration instructions instead of guessing what an employer wants. A conscientious respondent can legitimately admit occasional mistakes, and an outgoing respondent can legitimately enjoy solitary time; trait scores describe tendencies along continua.

A sensible approach is to complete the questionnaire in the intended language, without unnecessary interruptions, and with knowledge of how the results will be used. Avoid treating a self-report as a contest in which the highest score on every domain wins. If testing occurs in a formal process, ask for the test's purpose, relevant privacy information, and any appropriate accommodations.

How Should You Interpret a Suspicious or Unusually Favorable Big Five Profile?

Begin with the instrument. Establish which questionnaire was used, whether its scoring and norms apply to this respondent, whether the assessment was low- or high-stakes, and what external criterion the interpretation concerns. Examine response patterns and contextual information together. A seemingly ideal profile can be genuine; a profile with mixed scores can also have been strategically constructed.

Do not infer a mental health condition from a social desirability score or a personality domain score. Personality traits are descriptive dimensions, and response bias is an assessment issue. For a person under formal assessment, consequential interpretations require the competence, ethical procedures, and measurement evidence appropriate to that context.

Common Myths About Faking the Big Five

Myth: Everyone can obtain a perfect personality score

There is no universal perfect Big Five profile. Each domain is continuous; the relevance and trade-offs of trait tendencies depend on context, and specific tests use different items, norms, and score transformations. The notion of a single ideal score across all dimensions misstates the measurement model.

Myth: A validity scale can identify every liar

A response-validity indicator can sometimes signal a pattern requiring review, but its accuracy is conditional on the test and population. False positives and false negatives must be taken seriously. Directly labeling an individual a liar from one score exceeds the evidence.

Myth: If a test can be faked, it is scientifically useless

Susceptibility to strategic responding is an important limitation rather than a blanket invalidation. Meta-analyses find score shifts and some reduction in validity under motivated conditions while also finding that useful statistical relationships can remain. The practical assessment question is whether the evidence supports the intended decision. Speer et al. (2025).

Myth: Forced-choice questions make faking impossible

Forced-choice designs can reduce average inflation when well developed, but the method, scoring and stakes matter. The relevant meta-analyses and emerging AI-assisted stress tests do not support a universal immunity claim. Speer et al. (2023); Robie et al. (2026).

Frequently Asked Questions

Can a person intentionally raise their conscientiousness score?

Yes. In experiments, motivated respondents can endorse more desirable trait descriptions; applicant studies also report higher average conscientiousness scores in some comparisons. The amount varies across assessment settings and does not identify any one applicant as deceptive. Birkeland et al. (2006).

Which Big Five trait is easiest to fake?

No universal ranking is justified. Conscientiousness and emotional stability have shown notable average applicant shifts, but the apparent ease of distortion changes with the occupation, questionnaire, facet, item wording, and instructions. Comparing domains from different studies as if they were a single standardized contest would mislead.

Does a Big Five assessment have right or wrong answers?

Most self-report Big Five items have no knowledge-test answer key. Their meaning depends on how faithfully the responses represent the person's typical patterns and on whether an instrument supports the intended score interpretation. An answer can be misleading without being categorically right or wrong in the way a factual answer would be.

Will recruiters know when I have faked a personality test?

A recruiter or assessor may have evidence about the quality of a response pattern, but a flattering profile alone does not reveal intent. No universal method can reliably determine whether a particular individual deliberately distorted Big Five answers. A formal process should rely on validated interpretation and other appropriate evidence, not suspicion.

Can you fail a Big Five personality test?

In a descriptive Big Five assessment there is no universal pass or fail trait profile. A particular organization may use an assessment as one input in selection, but any decision threshold must be justified for the specific instrument, occupational purpose, and relevant validation evidence.

Are free online Big Five quizzes easier to fake?

Ease of distortion cannot be inferred from price. Free and commercial tests differ in content, format, length, and quality. Very short, transparent questionnaires may offer fewer cross-checks, but even lengthy paid measures are not automatically resistant. Inspect the published validation evidence for the exact instrument.

Can social desirability cause a high Big Five score without intentional lying?

Yes. Favorable self-beliefs, response styles, context, and deliberate impression management can overlap. Social desirability is not synonymous with conscious dishonesty, and it cannot be resolved simply by calling high scores invalid.

Do personality tests use trick questions to catch fakers?

Some assessments contain response-pattern or impression-management indicators, but these are test-design features with specific validation requirements, not universally reliable trick questions. Their presence should be documented and interpreted in the intended context.

Is it ethical for an employer to use Big Five tests?

Ethical use depends on a clear job-related purpose, defensible validation, appropriate privacy and fairness protections, competent interpretation, and the integration of other relevant evidence. The mere existence of faking research neither authorizes nor prohibits every use. High-stakes decisions demand higher scrutiny than educational self-reflection.

Could repeated answers change even if I was honest both times?

Yes. Variation can reflect differences in context, memory, transient state, measurement error, or real changes in dispositional behavior. Repeated assessments need instrument-specific error and stability evidence before changes are interpreted as substantive.

Does the Big Five diagnose mental illness or personality disorders?

No. The Big Five is a dimensional personality framework, and its questionnaires are not stand-alone clinical diagnostic tools. A respondent's high or low domain score and any response-bias flag do not determine a psychiatric diagnosis.

What the Evidence Supports

People can alter Big Five self-report scores, and incentives influence average response patterns. Field and experimental research identify meaningful variation in the magnitude of distortion and in its effects on criterion-related validity. Carefully designed forced-choice formats and some instructions may reduce average score inflation, but no method renders every test immune. Individual deception remains hard to establish from a questionnaire alone. These conclusions reflect a converging but methodologically diverse literature and should be applied to the actual instrument, setting, and population. Critical review (2025) and matched-sample meta-analysis (2025).

For a reader, the best use of a Big Five score is a carefully qualified description of self-reported tendencies. For a professional evaluator, the essential standard is evidence that the score supports a specific inference and decision under the conditions in which it was collected. This standard protects respondents while preserving the scientific value of appropriately developed personality assessment.

References

Birkeland, S. A., Manson, T. M., Kisamore, J. L., Brannick, M. T., & Smith, M. A. (2006). A meta-analytic investigation of job applicant faking on personality measures. International Journal of Selection and Assessment. https://doi.org/10.1111/j.1468-2389.2006.00354.x.

Cao, M., & Drasgow, F. (2019). Does forcing reduce faking? A meta-analytic review of forced-choice personality measures in high-stakes situations. Journal of Applied Psychology, 104(11), 1347–1368. https://doi.org/10.1037/apl0000414.

Dunlop, P. D., Xia, M. R., & Anglim, J. (2025). Faking on personality assessments in high-stakes settings: A critical review. Current Opinion in Psychology, 65, 102057. https://doi.org/10.1016/j.copsyc.2025.102057.

Hu, J., & Connelly, B. S. (2021). Faking by actual applicants on personality tests: A meta-analysis of within-subjects studies. International Journal of Selection and Assessment, 29, 412–426. https://doi.org/10.1111/ijsa.12338.

Lee, H. (2023). Reduction of faking with the use of a forced-choice personality test: Cross-cultural comparisons between South Korea and the United States. International Journal of Selection and Assessment, 31, 147–162. https://doi.org/10.1111/ijsa.12408.

Liu, J., Yan, C., & Zhang, J. (2024). Could faking on personality constructs be detected by social desirability scales? An example of the Marlowe–Crowne Social Desirability Scale. Psychologia, 66(2), 149–161. https://doi.org/10.2117/psysoc.2023-A219.

Martínez, A., & Salgado, J. F. (2021). A meta-analysis of the faking resistance of forced-choice personality inventories. Frontiers in Psychology, 12, 732241. https://doi.org/10.3389/fpsyg.2021.732241.

Moon, B., Daljeet, K. N., O'Neill, T. A., Harwood, H., Awad, W., & Beletski, L. V. (2025). Comparing the efficacy of faking warning types in preemployment personality tests: A meta-analysis. Journal of Applied Psychology, 110(1), 131–147. https://doi.org/10.1037/apl0001224.

Rau, R., Schömann, L. M., & Grosz, M. P. (2025). People fake-good on personality self-reports more strongly in a job context than in a dating context. Journal of Research in Personality, 116, 104596. https://doi.org/10.1016/j.jrp.2025.104596.

Robie, C., Feeney, J. R., Rasheed, S., Bourdage, J. S., Christiansen, N. D., Dunlop, P. D., & Daljeet, K. N. (2026). Faking resistance of forced-choice and single-statement personality measures: Stress-testing with humans and generative AI. International Journal of Selection and Assessment, 34(3), e70075. https://doi.org/10.1111/ijsa.70075.

Soto, C. J., & John, O. P. (2017). The next Big Five Inventory (BFI-2): Developing and assessing a hierarchical model with 15 facets to enhance bandwidth, fidelity, and predictive power. Journal of Personality and Social Psychology, 113(1), 117–143. https://doi.org/10.1037/pspp0000096.

Speer, A. B., Wegmeyer, L. J., Tenbrink, A. P., Delacruz, A. Y., Christiansen, N. D., & Salim, R. M. (2023). Comparing forced-choice and single-stimulus personality scores on a level playing field: A meta-analysis of psychometric properties and susceptibility to faking. Journal of Applied Psychology, 108(11), 1812–1833. https://doi.org/10.1037/apl0001099.

Speer, A. B., Delacruz, A. Y., Chawota, T., Wegmeyer, L. J., Tenbrink, A. P., Gibson, C., & Frost, C. (2025). Evaluating the impact of faking on the criterion-related validity of personality assessments. International Journal of Selection and Assessment. https://doi.org/10.1111/ijsa.12518.

Treglown, L., & Furnham, A. (2026). AI, social desirability, and personality assessments: Impression management in large language models. Personality and Individual Differences, 251, 113563. https://doi.org/10.1016/j.paid.2025.113563.

Psychological Encyclopedia