Polygraph Accuracy: What 80+ Years of Research Shows

Why Multiple Studies Matter

The question of how accurate a lie detector test truly is has been investigated through decades of rigorous scientific inquiry, spanning controlled laboratory experiments, real-world field studies with confirmed outcomes, and comprehensive meta-analyses synthesising data from thousands of examinations. Rather than relying on any single study, the scientific community evaluates polygraph accuracy through the weight of accumulated evidence.

Understanding which studies matter most — and what they actually found — is essential for anyone trying to assess whether a polygraph test will be reliable in their specific situation. Below, we examine the landmark studies that form the evidence base for modern polygraph testing. For a deeper exploration of what research says about exam accuracy, see our detailed analysis in Are Polygraph Exams Accurate? Science, Data & Expert Analysis.

APA Meta-Analytic Survey of Criterion Accuracy — 2011

The American Polygraph Association commissioned a comprehensive meta-analysis that stands as one of the most important single documents in polygraph science. Covering 38 studies that met strict inclusion criteria, involving 32 different samples, 45 experiments, 295 scorers, and 11,737 separately scored physiological results from 3,723 examinations, this analysis provides the aggregated accuracy data the APA cites in its own Standards of Practice [1]Verified Meta-Analytic Survey of Criterion Accuracy of Validated Polygraph Techniques
Confirms 38 studies, 11,737 scored results, 89% single-issue accuracy (CI 83–95%), 87% overall accuracy (CI 80–94%), and 11–13% inconclusive rates.

Key findings from the meta-analysis include: event-specific (single-issue) diagnostic testing achieved 89% aggregated decision accuracy with a confidence interval of 83–95% and an 11% inconclusive rate [1]Verified Meta-Analytic Survey of Criterion Accuracy of Validated Polygraph Techniques
Confirms 38 studies, 11,737 scored results, 89% single-issue accuracy (CI 83–95%), 87% overall accuracy (CI 80–94%), and 11–13% inconclusive rates. Multi-issue screening techniques achieved 85% accuracy with a confidence interval of 77–93% and a 13% inconclusive rate [1]Verified Meta-Analytic Survey of Criterion Accuracy of Validated Polygraph Techniques
Confirms 38 studies, 11,737 scored results, 89% single-issue accuracy (CI 83–95%), 87% overall accuracy (CI 80–94%), and 11–13% inconclusive rates. The combination of all validated techniques, excluding outlier results, produced a decision accuracy of 87% with a confidence interval of 80–94% and a 13% inconclusive rate [1]Verified Meta-Analytic Survey of Criterion Accuracy of Validated Polygraph Techniques
Confirms 38 studies, 11,737 scored results, 89% single-issue accuracy (CI 83–95%), 87% overall accuracy (CI 80–94%), and 11–13% inconclusive rates.

These findings were described as "consistent with those of the National Research Council's (2003) conclusions regarding polygraph accuracy" [1]Verified Meta-Analytic Survey of Criterion Accuracy of Validated Polygraph Techniques
Confirms 38 studies, 11,737 scored results, 89% single-issue accuracy (CI 83–95%), 87% overall accuracy (CI 80–94%), and 11–13% inconclusive rates. A 2015 update by Raymond Nelson, published as "Scientific Basis for Polygraph Testing" in Polygraph 41(1), confirmed and extended these findings [5]Verified Scientific Basis for Polygraph Testing (2015 Update)
Confirms Nelson's 2015 update extending the APA 2011 meta-analysis findings; published in Polygraph 41(1). The significance of this meta-analysis is that it represents the profession's own published position — the numbers the APA itself cites.

National Research Council (NAS) — 2003

The National Academy of Sciences conducted what remains the most comprehensive independent review of polygraph science ever undertaken. Commissioned by the US Department of Energy following security concerns at nuclear weapons facilities, the NRC assembled a panel of distinguished scientists — chaired by Stephen E. Fienberg of Carnegie Mellon University — who reviewed the broad polygraph literature [2]Verified The Polygraph and Lie Detection
Confirms NRC used 57 studies to quantify accuracy; commissioned by Department of Energy; reviewed polygraph literature comprehensively.

The NRC used 57 studies to quantify polygraph accuracy, all involving specific incidents, typically mock crimes [2]Verified The Polygraph and Lie Detection
Confirms NRC used 57 studies to quantify accuracy; commissioned by Department of Energy; reviewed polygraph literature comprehensively. Their final report, titled The Polygraph and Lie Detection, acknowledged that polygraph testing produces accuracy rates above chance. The NRC's concerns were primarily directed at broad counterintelligence screening programmes — not the focused, event-specific diagnostic testing used in private and legal examinations.

The 1983 Office of Technology Assessment report for Congress similarly reviewed the research base and found field and analog studies producing average accuracy of 82–88% using control question tests [6]Verified Scientific Validity of Polygraph Testing: A Research Review and Evaluation
Confirms 1983 OTA report for Congress reviewing field and analog studies with average accuracy of 82–88%. Together, these government reports establish that polygraph testing has a consistent track record of above-chance discrimination that has been repeatedly confirmed across decades. For further context on the early government evaluations, see our article on Polygraph Science in the 1970s: NAS Debate & Congress.

Honts, Thurber & Handler — CQT Meta-Analysis, 2021

Published in Applied Cognitive Psychology (Vol. 35, Issue 2, pp. 411–427), this independent meta-analysis examined 138 datasets on the Comparison Question Test — making it one of the largest meta-analytic efforts focused specifically on CQT methodology [3]Verified A Comprehensive Meta-Analysis of the Comparison Question Polygraph Test
Confirms 138 datasets, effect size 0.69, motivation as positive moderator, no publication bias, CQT accuracy generalises from lab to field. The study was authored by Charles R. Honts of Boise State University, Steven Thurber, and Mark Handler.

The study found a meta-analytic effect size including inconclusive outcomes of 0.69 [0.66, 0.79], confirmed that CQT accuracy generalises from laboratory to field settings, and found no evidence of publication bias in the literature [3]Verified A Comprehensive Meta-Analysis of the Comparison Question Polygraph Test
Confirms 138 datasets, effect size 0.69, motivation as positive moderator, no publication bias, CQT accuracy generalises from lab to field. Particularly relevant for private testing, the study identified motivation level as a significant positive moderator of accuracy — meaning that real examinations with genuine consequences produce more accurate results than laboratory studies with mock scenarios [3]Verified A Comprehensive Meta-Analysis of the Comparison Question Polygraph Test
Confirms 138 datasets, effect size 0.69, motivation as positive moderator, no publication bias, CQT accuracy generalises from lab to field.

As the authors noted, "average people have a truth bias" and "are only about 54% accurate" at detecting deception interpersonally [3]Verified A Comprehensive Meta-Analysis of the Comparison Question Polygraph Test
Confirms 138 datasets, effect size 0.69, motivation as positive moderator, no publication bias, CQT accuracy generalises from lab to field, while professionals such as police officers show approximately the same poor accuracy [3]Verified A Comprehensive Meta-Analysis of the Comparison Question Polygraph Test
Confirms 138 datasets, effect size 0.69, motivation as positive moderator, no publication bias, CQT accuracy generalises from lab to field. The CQT represents a substantial improvement over these baselines.

Earlier Foundational Research

Several earlier research programmes established the empirical foundation for modern polygraph testing. Raskin and Podlesny (1979) reported polygraph accuracy exceeding.90 using the CQT format, providing early quantification that helped shape testing standards [7]Verified Accuracy of Polygraph Techniques
Confirms polygraph accuracy exceeding.90 using CQT format; foundational early quantification. Honts and Peterson (1997) reviewed accuracy research and reported polygraph studies as exceeding.90, consistent with the earlier Raskin findings [8]Verified Review of Polygraph Accuracy Research
Confirms reported accuracy of polygraph studies as exceeding.90, consistent with Raskin findings.

Abrams (1989) conducted a comprehensive survey of published literature and reported an overall accuracy level of.89, with his earlier 1973 review finding laboratory accuracy of 83% and field accuracy of 98% [9]Verified Survey of Published Polygraph Literature on Accuracy
Confirms overall accuracy level of.89 from comprehensive survey of published literature. The 1983 Office of Technology Assessment report for the United States Congress analysed 10 field studies and 14 analog studies using the CQT, finding average unweighted laboratory accuracy of.83 [6]Verified Scientific Validity of Polygraph Testing: A Research Review and Evaluation
Confirms 1983 OTA report for Congress reviewing field and analog studies with average accuracy of 82–88%.

Researchers at the University of Utah — including David C. Raskin, Charles R. Honts, and John C. Kircher — conducted groundbreaking work on the comparison question technique and developed the Utah Zone Comparison Test. Their field studies used real criminal cases with confirmed outcomes established through confessions, DNA evidence, or other independent confirmation.

Computerised Scoring Validation — OSS and ESS

Research validating computer-assisted scoring algorithms has demonstrated that combining trained examiner analysis with computerised numerical scoring achieves strong accuracy rates. The Objective Scoring System version 3 (OSS-3), developed by Raymond Nelson, Donald Krapohl, and Mark Handler and published in Polygraph 37(3) in 2008, provides an open-source, objective method for analysing polygraph data [10]Verified Brute Force Comparison: A Monte Carlo Study of the Objective Scoring System version 3 (OSS-3) and Human Polygraph Scorers
Confirms OSS-3 developed by Nelson, Krapohl, and Handler; published in Polygraph 37(3), 2008; open-source algorithm for objective chart analysis. The Empirical Scoring System (ESS), validated by Nelson and colleagues in 2011, produced a mean accuracy rate of 90.1% (95% CI: 83.8–95.8%) when used by trainee examiners evaluating confirmed criminal investigation cases [11]Verified Empirical Scoring System: A Cross-Cultural Replication and Extension Study
Confirms ESS mean accuracy rate of 90.1% (95% CI: 83.8–95.8%) with trainee examiners using confirmed criminal investigation cases.

These algorithms apply consistent mathematical analysis across all physiological channels simultaneously, removing subjective interpretation bias and providing objective, repeatable results. The development of both OSS and ESS represents a milestone toward evidence-based polygraph practice [10]Verified Brute Force Comparison: A Monte Carlo Study of the Objective Scoring System version 3 (OSS-3) and Human Polygraph Scorers
Confirms OSS-3 developed by Nelson, Krapohl, and Handler; published in Polygraph 37(3), 2008; open-source algorithm for objective chart analysis.

Physiological Channels Recorded

One of the most persistent misconceptions about polygraph testing is the idea that the instrument somehow detects lies directly. In reality, the polygraph measures specific physiological responses controlled by the autonomic nervous system — responses that occur involuntarily when a person experiences psychological conflict associated with deception.

Modern polygraph instruments record four primary physiological channels simultaneously. The APA Standards of Practice require all channels to be recorded independently on every examination [4]Verified APA Standards of Practice (Amended August 23, 2024)
Confirms current APA Standards effective date, required physiological channels, result categories, and technique validation requirements.

Respiratory Activity: Two pneumograph tubes record thoracic (chest) and abdominal breathing separately. Deception typically produces suppressed, irregular, or altered breathing patterns — particularly a reduction in breath amplitude and changes in the inspiration-to-expiration ratio. Learn more about this critical measure in our guide on Respiratory Suppression in Polygraph Testing.

Electrodermal Activity (EDA): Skin conductance sensors on the fingertips measure sweat gland activity controlled by the sympathetic nervous system. EDA is involuntary and highly sensitive to psychological stress — it cannot be consciously suppressed. Increased electrodermal activity following relevant questions is one of the strongest indicators of deceptive arousal in the polygraph record.

Cardiovascular Activity: A blood pressure cuff records relative blood pressure changes and pulse rate throughout the examination. Deceptive responses frequently produce increases in diastolic blood pressure and changes in pulse amplitude and rate.

Seat Activity Sensor: A movement sensor in the examination chair detects physical countermeasures — deliberate muscle tensing, toe-pressing, or shifting that could artificially inflate comparison question responses.

Why Differential Response Is the Key

The Comparison Question Technique — the methodology used in virtually all private, legal, and law enforcement examinations — works by comparing responses to relevant questions (the specific issue being investigated) against comparison questions (broad, mildly threatening questions unrelated to the issue). A deceptive examinee shows stronger physiological responses to relevant questions because they represent the greatest psychological threat. A truthful examinee shows stronger responses to comparison questions.

This differential pattern — the relative difference between response types, not the absolute magnitude of any individual response — is what the examiner scores using numerical analysis. This is precisely why general nervousness does not cause false results: since the test measures relative contrast between question types rather than absolute arousal levels, a nervous but truthful person will still show stronger comparison question responses. The anxiety elevates all responses equally without distorting the differential pattern.

For more on the psychological mechanisms underlying polygraph responses, see our article on Salience in Polygraph Testing: The Psychology of Responses. Understanding the role of the polygraph baseline test is also important for context.

Why Examiner Qualification Is the Largest Variable

If there is one finding that emerges from every major polygraph study — from the NRC in 2003 to Honts et al. in 2021 — it is that examiner qualification is the single largest variable determining accuracy. The instrument is validated. The methodology is scientifically supported. But the results of any specific examination depend overwhelmingly on whether the person administering it has been properly trained, uses validated techniques, and follows established standards.

The Honts et al. (2021) meta-analysis explicitly noted problems within the polygraph profession that lower field accuracy, including inconsistent licensing requirements and advisory-only professional standards [3]Verified A Comprehensive Meta-Analysis of the Comparison Question Polygraph Test
Confirms 138 datasets, effect size 0.69, motivation as positive moderator, no publication bias, CQT accuracy generalises from lab to field. Only 26 of 50 US states license polygraph examiners [3]Verified A Comprehensive Meta-Analysis of the Comparison Question Polygraph Test
Confirms 138 datasets, effect size 0.69, motivation as positive moderator, no publication bias, CQT accuracy generalises from lab to field. This makes verifying your examiner's credentials critically important.

APA-Accredited School Requirements

APA-accredited polygraph schools require a minimum of 400 classroom hours of structured instruction covering psychophysiology, instrument operation, question formulation, chart analysis, and professional standards [12]Verified PEAK Credibility Assessment Training Center — Basic Training Course
Confirms APA-accredited basic course consists of minimum 400 hours. The curriculum includes supervised internship conducting real examinations under the oversight of an experienced examiner before practising independently, validated technique training specifically in techniques with published empirical validation, and seven-position numerical scoring instruction.

The APA requires practising examiners to complete a minimum of 30 continuing education hours every two years in coursework related to polygraphy, as specified in APA By-Laws §1.2.1 [13]Verified APA Policy for Continuing Education Hours
Confirms APA By-Laws §1.2.1 requiring 30 continuing education hours every two years for practising examiners. Member credentials are searchable in the APA public member directory at polygraph.org.

For details on professional training standards, see PEAK CATC Online Polygraph Training: Course Catalog 2026 and PEAK Interviewing Course: 40-Hour Nonconfrontational Training.

Red Flags: Unqualified Examiners

Examiners who lack proper accreditation typically have short online or weekend courses of 40–80 hours with no supervised practice requirement, no supervised internship, non-validated techniques with no published accuracy data, no numerical scoring (entirely subjective chart analysis), no ongoing CPD, and no verifiable credentials. Booking with an unqualified examiner means your results cannot claim the accuracy rates supported by the research literature.

For guidance on verifying credentials and avoiding unqualified providers, see our guides on Avoiding Polygraph Scams in NC: License Verification and Polygraph Walk-In Services: Convenience or Consumer Trap?.

The Comparison Question Test (CQT)

The CQT is the workhorse methodology of professional polygraph testing. It measures relative physiological arousal between relevant questions about the specific issue being investigated and comparison questions that are broad, mildly threatening, and unrelated to the issue.

Validated CQT variants include the Utah Zone Comparison Test (UZCT), the Federal Zone Comparison Test, the Directed Lie Screening Test (DLST/TES), and the Air Force Modified General Question Test (AFMGQT). Each has published accuracy data meeting the APA's research standards. The APA meta-analysis found CQT variants achieving 85–89% accuracy depending on whether the test was single-issue or multi-issue [1]Verified Meta-Analytic Survey of Criterion Accuracy of Validated Polygraph Techniques
Confirms 38 studies, 11,737 scored results, 89% single-issue accuracy (CI 83–95%), 87% overall accuracy (CI 80–94%), and 11–13% inconclusive rates.

All private relationship, infidelity, and personal issue examinations use CQT variants. These are by definition single-issue diagnostic tests, which the research shows produce the highest accuracy of any polygraph application. For a complete walkthrough, see our Specific Incident Polygraph Test: Complete Guide.

The Concealed Information Test (CIT / GKT)

The CIT takes a completely different approach. Rather than measuring deception directly, it tests whether an examinee shows elevated recognition responses to specific details of an event that only someone with knowledge of it would recognise. MacLaren's (2001) comprehensive meta-analysis of CIT research confirmed strong discrimination between informed and uninformed examinees, with data later incorporated into the APA 2011 meta-analytic review [14]Verified A Comprehensive Meta-Analysis of CIT Research Using MacLaren's Data
Confirms strong CIT discrimination between informed and uninformed examinees; data incorporated into APA 2011 meta-analysis.

Ben-Shakhar (2012) provided a comprehensive overview of CIT research highlighting its strong theoretical foundation in orienting response theory, noting that the CIT's false positive rate remains extremely low (approximately 5%) across studies [15]Verified Current Research and Potential Applications of the Concealed Information Test: An Overview
Confirms CIT false positive rate of approximately 5% across studies and strong theoretical foundation in orienting response theory. Japan is the only country where the CIT is widely applied in criminal investigations, with 80–100 examiners conducting about 5,000 tests per year [3]Verified A Comprehensive Meta-Analysis of the Comparison Question Polygraph Test
Confirms 138 datasets, effect size 0.69, motivation as positive moderator, no publication bias, CQT accuracy generalises from lab to field. For more on Japan's system, see Japan's CIT Polygraph System: How NRIPS Standardized It.

The CIT is less applicable to private relationship examinations where there are no specific concealed details to test against.

APA-Defined Result Categories

The APA Standards of Practice define four categorical outcomes for comparison question examinations [4]Verified APA Standards of Practice (Amended August 23, 2024)
Confirms current APA Standards effective date, required physiological channels, result categories, and technique validation requirements: DI (Deception Indicated) — the physiological data indicates the examinee was deceptive; NDI (No Deception Indicated) — the data indicates the examinee was truthful; INC (Inconclusive) — the data was insufficient for a definitive determination; and NO (No Opinion) — the examiner cannot render an opinion due to procedural issues, countermeasure detection, or other factors.

Recognition tests use RI (Recognition Indicated), NRI (No Recognition Indicated), and NO [4]Verified APA Standards of Practice (Amended August 23, 2024)
Confirms current APA Standards effective date, required physiological channels, result categories, and technique validation requirements. These standardised labels are mandatory for APA-compliant reporting. Vague or non-standard conclusions such as "probably truthful" or "likely deceptive" are not APA-compliant and should be treated as a red flag. Understanding how to read your report is covered in our guide on What Your Polygraph Report Should Contain: 15-Point Checklist.

False Positives vs. False Negatives

A false positive occurs when a truthful person is incorrectly classified as deceptive (DI). A false negative occurs when a deceptive person is incorrectly classified as truthful (NDI). The majority of non-conclusive outcomes in validated techniques are classified as inconclusive — meaning the system correctly identifies ambiguous data rather than making an incorrect decision.

The APA Standards require that techniques for investigative testing demonstrate an unweighted average accuracy rate of 80% or greater, excluding inconclusive results, which shall not exceed 20% [4]Verified APA Standards of Practice (Amended August 23, 2024)
Confirms current APA Standards effective date, required physiological channels, result categories, and technique validation requirements. The 2011 meta-analysis found actual inconclusive rates of 11–13% across validated methodologies [1]Verified Meta-Analytic Survey of Criterion Accuracy of Validated Polygraph Techniques
Confirms 38 studies, 11,737 scored results, 89% single-issue accuracy (CI 83–95%), 87% overall accuracy (CI 80–94%), and 11–13% inconclusive rates. For a broader perspective on test limitations, see Polygraph Limitations: Technology, Science & Legal Issues.

Inconclusive Is Not an Error

An inconclusive (INC) result means the physiological data was insufficient to make a definitive determination in either direction. This is the system working correctly — refusing to guess when the evidence is ambiguous. The inconclusive rate in CQT is typically between 10–20% [5]Verified Scientific Basis for Polygraph Testing (2015 Update)
Confirms Nelson's 2015 update extending the APA 2011 meta-analysis findings; published in Polygraph 41(1). The appropriate response to an INC result is retesting, not treating it as either a pass or a failure.

What Compliance Requires

The APA Standards of Practice, most recently amended August 23, 2024, set mandatory requirements for every polygraph examination [4]Verified APA Standards of Practice (Amended August 23, 2024)
Confirms current APA Standards effective date, required physiological channels, result categories, and technique validation requirements. These define the methodological framework within which the published accuracy data was generated. An examination that deviates from these standards cannot claim the accuracy rates the research supports.

Key requirements include: the use of evidence-based validated testing techniques [4]Verified APA Standards of Practice (Amended August 23, 2024)
Confirms current APA Standards effective date, required physiological channels, result categories, and technique validation requirements; recording of thoracic and abdominal respiration patterns separately, electrodermal activity, cardiovascular activity, and a seat activity sensor [4]Verified APA Standards of Practice (Amended August 23, 2024)
Confirms current APA Standards effective date, required physiological channels, result categories, and technique validation requirements; proper pretest interview procedures including review of all test questions prior to recording physiological responses [4]Verified APA Standards of Practice (Amended August 23, 2024)
Confirms current APA Standards effective date, required physiological channels, result categories, and technique validation requirements; and the use of standardised result categories (DI, NDI, INC, NO) [4]Verified APA Standards of Practice (Amended August 23, 2024)
Confirms current APA Standards effective date, required physiological channels, result categories, and technique validation requirements.

Examiners are also required to obtain informed consent and make reasonable efforts to determine examinee suitability, including basic inquiries into medical and psychological conditions [4]Verified APA Standards of Practice (Amended August 23, 2024)
Confirms current APA Standards effective date, required physiological channels, result categories, and technique validation requirements. For information about your rights during the process, see Can I Refuse a Lie Detector Test? Your Legal Rights.

Factors That Maximise Accuracy

Research consistently shows several factors that maximise polygraph accuracy: single-issue testing (one specific topic per examination), a qualified examiner who graduated from an APA-accredited school with a minimum of 400 hours of training [12]Verified PEAK Credibility Assessment Training Center — Basic Training Course
Confirms APA-accredited basic course consists of minimum 400 hours, the use of a validated technique with published accuracy data meeting APA research standards, seven-position numerical scoring rather than subjective global analysis, computerised scoring algorithms (such as OSS-3 or ESS) used alongside manual analysis [10]Verified Brute Force Comparison: A Monte Carlo Study of the Objective Scoring System version 3 (OSS-3) and Human Polygraph Scorers
Confirms OSS-3 developed by Nelson, Krapohl, and Handler; published in Polygraph 37(3), 2008; open-source algorithm for objective chart analysis [11]Verified Empirical Scoring System: A Cross-Cultural Replication and Extension Study
Confirms ESS mean accuracy rate of 90.1% (95% CI: 83.8–95.8%) with trainee examiners using confirmed criminal investigation cases, proper pretest interview establishing psychological set, and an appropriate testing environment free from distractions.

The Honts et al. (2021) meta-analysis identified motivation level as a significant positive moderator — real examinations where the outcome genuinely matters to the examinee produce more accurate results than laboratory simulations [3]Verified A Comprehensive Meta-Analysis of the Comparison Question Polygraph Test
Confirms 138 datasets, effect size 0.69, motivation as positive moderator, no publication bias, CQT accuracy generalises from lab to field. This means private diagnostic examinations, where the examinee has genuine stakes in the outcome, operate under conditions that research shows enhance accuracy.

Factors That Reduce Accuracy

Conversely, accuracy is compromised by: multi-issue testing (attempting to address too many topics in one examination), examiners lacking accredited training or using non-validated techniques, subjective chart analysis without numerical scoring, inadequate pretest interview, and broad screening contexts where there is no specific event to investigate.

The NRC's 2003 report specifically noted that their concerns were directed primarily at broad counterintelligence screening programmes — not focused diagnostic testing [2]Verified The Polygraph and Lie Detection
Confirms NRC used 57 studies to quantify accuracy; commissioned by Department of Energy; reviewed polygraph literature comprehensively. This distinction is crucial for understanding why the accuracy data is most favourable for the type of single-issue testing used in private examinations.

Interpersonal Deception Detection

Research consistently demonstrates that unaided human deception detection is remarkably poor. The Honts et al. (2021) meta-analysis noted that average people are only about 54% accurate at detecting deception — barely above chance [3]Verified A Comprehensive Meta-Analysis of the Comparison Question Polygraph Test
Confirms 138 datasets, effect size 0.69, motivation as positive moderator, no publication bias, CQT accuracy generalises from lab to field. Moreover, professionals such as police officers charged with making credibility judgments show approximately the same accuracy but with a lie bias (they tend to judge people as liars) [3]Verified A Comprehensive Meta-Analysis of the Comparison Question Polygraph Test
Confirms 138 datasets, effect size 0.69, motivation as positive moderator, no publication bias, CQT accuracy generalises from lab to field.

As Vrij (2023) documented in his paradigm-shifting review, meta-analyses show only 28% of studied behavioural cues significantly relate to deception, with small effect sizes (d =.25) [16]Verified A Paradigm Shift in 30 Years of Deception Research
Confirms only 28% of studied behavioural cues significantly relate to deception with small effect sizes (d =.25). Fisher, Vrij, and Leins (2013) demonstrated that the most commonly cited reason for discrediting witnesses — testimonial inconsistency — is scientifically problematic since truthful recollections naturally vary [17]Verified Does Testimonial Inconsistency Indicate Memory Inaccuracy and Deception?
Confirms inconsistency is the most commonly cited reason for discrediting witnesses but this reliance is scientifically problematic. Levine (2015) presented updated accuracy findings in deception detection research that further contextualise the limitations of non-instrumental methods [18]Verified New and Improved Accuracy Findings in Deception Detection Research
Foundational research relevant to deception detection accuracy benchmarks.

The CQT's demonstrated accuracy of 87–89% represents a substantial and meaningful improvement over these interpersonal baselines [1]Verified Meta-Analytic Survey of Criterion Accuracy of Validated Polygraph Techniques
Confirms 38 studies, 11,737 scored results, 89% single-issue accuracy (CI 83–95%), 87% overall accuracy (CI 80–94%), and 11–13% inconclusive rates [3]Verified A Comprehensive Meta-Analysis of the Comparison Question Polygraph Test
Confirms 138 datasets, effect size 0.69, motivation as positive moderator, no publication bias, CQT accuracy generalises from lab to field. For additional deception detection context, see our collection of Famous Quotes on Deception & Their Relevance to Lie Detection.

Emerging Technologies

Various alternative credibility assessment technologies have been proposed, including voice stress analysis, fMRI-based lie detection, and eye-tracking systems such as EyeDetect. While some show promise in laboratory settings, none currently matches the CQT's extensive validation history across both laboratory and field settings [3]Verified A Comprehensive Meta-Analysis of the Comparison Question Polygraph Test
Confirms 138 datasets, effect size 0.69, motivation as positive moderator, no publication bias, CQT accuracy generalises from lab to field. For a detailed comparison, see EyeDetect vs. Polygraph for PCSOT: Complete 2026 Guide.

What the Research Shows About Countermeasures

Research by Honts, Raskin, and Kircher (1994), published in the Journal of Applied Psychology, examined countermeasures in an experiment with 120 subjects recruited from the general community [19]Verified Mental and Physical Countermeasures Reduce the Accuracy of Polygraph Tests
Confirms 1994 study with 120 subjects finding countermeasures enabled approximately 50% defeat rate; published in Journal of Applied Psychology. Of 100 guilty subjects, 80 were trained in physical countermeasures (biting the tongue or pressing toes to the floor) or mental countermeasures (counting backward by 7). Each type of countermeasure enabled approximately 50% of subjects to defeat the polygraph test in that controlled laboratory setting [19]Verified Mental and Physical Countermeasures Reduce the Accuracy of Polygraph Tests
Confirms 1994 study with 120 subjects finding countermeasures enabled approximately 50% defeat rate; published in Journal of Applied Psychology. The countermeasures were described as "difficult to detect either instrumentally or through observation" in that study [19]Verified Mental and Physical Countermeasures Reduce the Accuracy of Polygraph Tests
Confirms 1994 study with 120 subjects finding countermeasures enabled approximately 50% defeat rate; published in Journal of Applied Psychology.

However, important context applies. This 1994 study used instruments and techniques available at the time — without modern seat activity sensors, computerised scoring algorithms, or current APA Standards requiring movement artifact detection. Modern polygraph examinations include mandatory seat activity sensors that detect physical countermeasures [4]Verified APA Standards of Practice (Amended August 23, 2024)
Confirms current APA Standards effective date, required physiological channels, result categories, and technique validation requirements, computerised scoring algorithms (OSS-3, ESS) that provide objective analysis less susceptible to manipulation [10]Verified Brute Force Comparison: A Monte Carlo Study of the Objective Scoring System version 3 (OSS-3) and Human Polygraph Scorers
Confirms OSS-3 developed by Nelson, Krapohl, and Handler; published in Polygraph 37(3), 2008; open-source algorithm for objective chart analysis, trained examiner awareness of countermeasure indicators developed through decades of research, and standardised protocols specifically designed to identify irregular response patterns.

For information on how Statement Verification Testing (SVT) provides additional layers of assessment, see our dedicated guide.

Why Nervousness Does Not Cause Failures

Because the CQT measures the relative difference between responses to relevant questions and comparison questions — not absolute physiological arousal — general anxiety affects all responses equally. A nervous but truthful person will still show the characteristic pattern of stronger comparison question responses. The differential pattern the examiner analyses remains intact regardless of baseline anxiety level.

The APA Standards require examiners to make reasonable efforts to determine that the examinee is a suitable candidate for polygraph testing, including basic inquiries into medical and psychological conditions [4]Verified APA Standards of Practice (Amended August 23, 2024)
Confirms current APA Standards effective date, required physiological channels, result categories, and technique validation requirements. Mental, physical, or medical conditions that are observable or reasonably known to the examiner should be considered when conducting and evaluating an examination [4]Verified APA Standards of Practice (Amended August 23, 2024)
Confirms current APA Standards effective date, required physiological channels, result categories, and technique validation requirements.

Certain conditions may warrant special consideration, including cardiovascular conditions affecting blood pressure readings, respiratory conditions, medications that significantly alter autonomic nervous system function, and severe anxiety disorders. A qualified examiner will assess suitability during the pretest interview and may issue a No Opinion result if they determine conditions could compromise the examination's validity.

Essential Verification Steps

Before booking any polygraph examination, verify that the examiner graduated from an APA-accredited polygraph school (minimum 400 classroom hours) [12]Verified PEAK Credibility Assessment Training Center — Basic Training Course
Confirms APA-accredited basic course consists of minimum 400 hours, maintains active APA membership with current continuing education compliance (30 hours biennially as required by APA By-Laws §1.2.1) [13]Verified APA Policy for Continuing Education Hours
Confirms APA By-Laws §1.2.1 requiring 30 continuing education hours every two years for practising examiners, uses validated testing techniques listed in the APA's meta-analysis and Standards of Practice, employs seven-position numerical scoring and/or computerised scoring algorithms, records all four required physiological channels [4]Verified APA Standards of Practice (Amended August 23, 2024)
Confirms current APA Standards effective date, required physiological channels, result categories, and technique validation requirements, and can provide credentials searchable through the APA public member directory at polygraph.org.

Licensing requirements vary dramatically from state to state — currently only 26 of 50 US states license polygraph examiners [3]Verified A Comprehensive Meta-Analysis of the Comparison Question Polygraph Test
Confirms 138 datasets, effect size 0.69, motivation as positive moderator, no publication bias, CQT accuracy generalises from lab to field. For state-specific guidance, see our resources on Colorado Polygraph Licensing and Your Rights: Employer Polygraph Requests in NC. Understanding the history of the profession also helps — see Marston, Keeler & Larson: Disputed Origins of the Polygraph.