Polygraph Sensitivity vs. Specificity: What Numbers Mean

Defining Sensitivity: The True Positive Rate

Sensitivity — also called the true positive rate — answers one fundamental question: when someone is actually lying, how often does the polygraph correctly flag them as deceptive? In clinical medicine, sensitivity is the measure used to evaluate a screening test's ability to detect disease when disease is present. In polygraph science, the 'disease' is deception.

A test with 90% sensitivity will correctly identify 90 out of every 100 deceptive individuals as deceptive. The remaining 10 will escape detection — these are called false negatives.

The mathematical formula is straightforward: Sensitivity = True Positives / (True Positives + False Negatives). For example, if a validated polygraph technique is used on 200 people who are confirmed to be deceptive, and the test correctly identifies 180 of them, the sensitivity is 180 / 200 = 0.90, or 90%.

The APA's 2011 meta-analytic survey of validated polygraph techniques reported an aggregated sensitivity rate for deception during diagnostic polygraphs of.84 (95% confidence range.73 to.93) [1]Verified Meta-Analytic Survey of Criterion Accuracy of Validated Polygraph Techniques
Confirms 89% accuracy for single-issue diagnostic testing, 85% for multi-issue, 87% overall, with inconclusive rates of 11-13%. Reports sensitivity of.84 and specificity of.77 for diagnostic polygraphs.. This means that properly conducted diagnostic polygraph examinations correctly identify approximately 84% of deceptive examinees, with some validated techniques achieving sensitivity above 90% [2]Verified Frequently Asked Questions about Validated Techniques and the Meta-Analysis
Confirms five validated techniques produce mean accuracy rates over 90% with inconclusive rates lower than 20% for evidentiary use..

Why High Sensitivity Matters

In contexts where missing a deceptive person carries severe consequences — national security screening, law enforcement hiring, or post-conviction sex offender testing (PCSOT) — high sensitivity is critical. An examiner conducting a counterintelligence polygraph at the NSA or CIA needs confidence that deceptive applicants will be caught. A test that lets threats through defeats the entire purpose of screening.

One field study of CQT polygraph examinations demonstrated a sensitivity of 0.94 in detecting deceptive examinees with a false negative rate of only 0.06 [3]Verified Estimated Accuracy of CQT Polygraph Examination
Confirms CQT field sensitivity of 0.94 and specificity of 0.835 using a novel ground-truth-free estimation method with Israel Police data.. Similarly, a study of CQT polygraph accuracy in criminal cases found a true positive rate of 87% [4]Verified Polygraph Accuracy of Control Question Test in Criminal Cases
Confirms CQT criminal case true positive rate of 87% and true negative rate of 82.2% with no significant effect of education level on detection accuracy.. These results demonstrate that, when conducted by competent examiners using validated techniques, polygraph testing achieves impressive detection rates.

However, increasing sensitivity often comes at a cost — and that cost is measured in specificity. Understanding this tradeoff is essential to interpreting any accuracy claim you encounter. For a comprehensive look at the instruments used in polygraph testing, our technical guide covers all the physiological channels involved.

Defining Specificity: The True Negative Rate

Specificity answers the opposite question: when someone is actually telling the truth, how often does the polygraph correctly clear them? A test with 85% specificity will correctly classify 85 out of every 100 truthful individuals as No Deception Indicated (NDI). The other 15 will be incorrectly labeled deceptive — these are false positives.

The formula is: Specificity = True Negatives / (True Negatives + False Positives). If 300 truthful people take a polygraph and 264 are correctly cleared, the specificity is 264 / 300 = 0.88, or 88%.

The APA's 2011 meta-analysis reported an aggregated specificity rate for truth-telling during diagnostic polygraphs of.77 (95% confidence range.65 to.85) [1]Verified Meta-Analytic Survey of Criterion Accuracy of Validated Polygraph Techniques
Confirms 89% accuracy for single-issue diagnostic testing, 85% for multi-issue, 87% overall, with inconclusive rates of 11-13%. Reports sensitivity of.84 and specificity of.77 for diagnostic polygraphs.. A review of relevant field studies by the British Psychological Society highlighted specificity figures between 53% and 78% for correctly identifying innocence, while laboratory studies showed more positive findings [5]Verified A review of the polygraph: history, methodology and current status
Confirms NRC 2003 review included 37 laboratory and 7 field studies meeting minimum standards, with median CQT accuracy of.85. Reports BPS field study review found sensitivity 83-89% and specificity 53-78%.. The wide range reflects differences in technique, scoring method, and study design.

Why High Specificity Matters

Specificity is the metric that matters most to truthful examinees. If you are honest and are about to take a polygraph test, specificity tells you how likely it is that the test will correctly confirm your truthfulness. Low specificity means a higher chance of being falsely accused — an outcome that can derail a career, damage a relationship, or complicate a legal case.

Encouragingly, certain polygraph techniques demonstrate notably strong specificity. The Concealed Information Test (CIT), for example, has shown exceptionally high specificity — one meta-analysis of CIT studies using electrodermal measures found an average false alarm rate of only 0.04, meaning 96% of innocent examinees were correctly cleared [6]Verified The validity of psychophysiological detection of information with the Guilty Knowledge Test: a meta-analytic review
Confirms CIT meta-analysis found average hit rate of 0.83 and false alarm rate of 0.04 across 80 laboratory studies with 5,198 participants.. Research has also shown that CIT sensitivity can be improved from 86% to 97% hit rate when target-irrelevant response patterns are used to predict individual detection efficiency [7]Verified Predicting the sensitivity of the reaction time-based concealed information test
Confirms CIT sensitivity improvement from 86% to 97% hit rate using target-irrelevant response patterns, with ROC AUC of 0.95..

For consumers scheduling a private lie detector test, understanding specificity helps set realistic expectations. It also explains why choosing a qualified, experienced polygraph examiner matters — proper technique, appropriate question formulation, and validated scoring algorithms all contribute to higher specificity. Learn more about why examiner quality matters in our guide on unqualified polygraph examiners.

Understanding the Fundamental Tension

In every diagnostic test — whether it is a medical blood test, a COVID antigen test, or a polygraph examination — there exists a fundamental tension between sensitivity and specificity. Improving one typically degrades the other.

Imagine a polygraph scoring algorithm with an adjustable threshold for calling deception. If you set that threshold very low (flagging even mild physiological responses as deceptive), you will catch more liars (higher sensitivity), but you will also flag more truthful people (lower specificity). If you set the threshold very high (requiring strong, consistent physiological responses before calling deception), you will correctly clear more truthful people (higher specificity), but some deceptive people with mild responses will slip through (lower sensitivity).

This tradeoff is not a flaw of polygraph testing specifically — it is a mathematical reality of all binary classification systems. The same tension exists in cancer screening, airport security scanners, and fraud detection algorithms. The NRC's 2003 review noted that the concept of a decision threshold had been used for decades in signal detection research but had received little attention in U.S. government-supported polygraph research [8]Verified The Polygraph and Lie Detection
Confirms NRC 2003 review of 57 studies, median accuracy of.85, distinction between screening and diagnostic testing, and CPS discriminant function analysis methodology..

How Examiners Navigate the Tradeoff

Professional polygraph examiners do not arbitrarily choose between sensitivity and specificity. Instead, the APA Standards of Practice require the use of validated techniques — scoring systems that have been tested in controlled research to achieve an acceptable balance of both measures [9]Verified APA Standards of Practice
Confirms APA requirements for validated techniques including 90% accuracy for evidentiary testing and 20% maximum inconclusive rate.. Techniques like the Utah Zone Comparison Test and the Directed Lie Screening Test have published sensitivity and specificity data that examiners can reference.

The choice of technique often depends on the context of the examination. For evidentiary examinations, the APA requires techniques with at least two published empirical studies demonstrating an unweighted average accuracy rate of 90% or greater, with inconclusive results not exceeding 20% [9]Verified APA Standards of Practice
Confirms APA requirements for validated techniques including 90% accuracy for evidentiary testing and 20% maximum inconclusive rate.. Investigative techniques need at least 80% accuracy under the same inconclusive threshold [9]Verified APA Standards of Practice
Confirms APA requirements for validated techniques including 90% accuracy for evidentiary testing and 20% maximum inconclusive rate.. Five validated techniques currently meet the evidentiary standard, including the Federal You-Phase technique scored with the ESS, the Utah ZCT, and ZCT variants scored with the ESS [10]Verified Scientific Basis for Polygraph Testing
Confirms event-specific diagnostic polygraphs provide mean accuracy of.89 (CI.83-.95) and multi-issue screening mean of.85 (CI.77-.93)..

Modern computerized scoring algorithms further enhance consistency. The OSS-3 (Objective Scoring System version 3), developed by Nelson, Krapohl, and Handler, is a form of 7-position scoring where values are derived from measurement ratios of physiological features, eliminating subjectivity in chart interpretation [11]Verified Objective Scoring System – Version 3
Confirms OSS-3 is a computerized scoring algorithm with demonstrable validity across multiple validation samples.. The CPS (Computerized Polygraph System), originally developed at the University of Utah, uses linear discriminant function analysis to classify physiological data, focusing on three key features: skin conductance amplitude, cardiograph baseline increases, and a composite respiration measure [8]Verified The Polygraph and Lie Detection
Confirms NRC 2003 review of 57 studies, median accuracy of.85, distinction between screening and diagnostic testing, and CPS discriminant function analysis methodology.. These systems serve as decision-support tools that enhance, rather than replace, the examiner's expert interpretation.

For a detailed comparison of scoring approaches, see our guide on the 3-position scale in polygraph scoring.

The Four Possible Outcomes of Every Polygraph Test

Every polygraph examination produces one of four possible outcomes. Understanding these four categories is the foundation of interpreting any accuracy statistic.

True Positive (Correct Detection): A deceptive person is correctly identified as deceptive. The system worked as intended — the liar was caught.

True Negative (Correct Clearance): A truthful person is correctly identified as truthful (NDI — No Deception Indicated). The innocent person was properly cleared.

False Positive (Type I Error): A truthful person is incorrectly classified as deceptive. This is the scenario honest examinees fear most. It occurs when specificity is imperfect.

False Negative (Type II Error): A deceptive person is incorrectly classified as truthful. The liar passes the test undetected. It occurs when sensitivity is imperfect.

For a full explanation of what happens when a test returns a positive result, see our guide on Deception Indicated (DI) results.

Real-World Impact of Each Error Type

The consequences of false positives and false negatives depend entirely on context. In a police pre-employment polygraph, a false positive means a qualified, honest candidate is rejected — a career is derailed and the department loses a good recruit. A false negative means a dishonest applicant enters the force, potentially creating liability and public safety risks. Research into corporate polygraph screening has shown that both employers and employees increasingly support polygraph use in recruitment when properly conducted [12]Verified Use of special technical means (polygraph) on personnel recruitment in corporate structures
Confirms polygraph screening is increasingly accepted in corporate recruitment with employer and employee support for its use..

In an infidelity polygraph test, a false positive means a faithful partner is wrongly accused of cheating — potentially destroying a relationship. A false negative means a cheating partner receives a clean bill of health, deepening deception.

Neither error type is harmless. This is why understanding the specific accuracy profile of the technique being used — not just a generic accuracy percentage — matters enormously. Research has demonstrated that different screening tools applied to the same population can produce referral rates varying by over 50 percentage points, underscoring the critical importance of choosing validated methods [13]Verified Prison Mental Health Screening tools: choices and sensitivity
Confirms validated screening tools can produce referral rates varying by over 50 percentage points on the same sample, demonstrating the critical importance of tool selection..

What Causes Each Type of Error?

False positives can be caused by extreme nervousness or anxiety in truthful examinees, medical conditions that alter physiological baselines (cardiovascular issues, thyroid disorders), poor question formulation that triggers emotional responses unrelated to deception, inadequate pre-test interview that fails to address an examinee's concerns, or examiner bias in chart interpretation. Research has confirmed that cognitive processing alone — such as mental arithmetic — can generate physiological reactions detectable by polygraph instrumentation, highlighting the importance of proper baseline establishment [14]Verified Tests with Numbers
Confirms cognitive processing (mental arithmetic) generates observable EDA responses detectable by polygraph instrumentation.. Our guide on preventing intimidation during a lie detector exam offers practical strategies for managing test anxiety.

False negatives can be caused by certain countermeasures (though detection methods exist), certain personality traits that dampen physiological arousal, medication effects that suppress autonomic nervous system responses, or poor examiner technique. Research has demonstrated that physiological responses exist on a continuum rather than in discrete categories, which means that individual variation in physiological profiles is an inherent challenge for binary classification [15]Verified Physiological Reactions Profiling in Polygraph Testing: Insights from Fuzzy C-Means Clustering Analysis
Confirms physiological responses exist on a continuum rather than discrete categories, with individual variation in response profiles.. Understanding the psychophysiological basis of the CQT technique helps explain why these variations matter.

Experienced examiners mitigate these risks through proper pre-test procedures, validated question formulation, and use of multiple physiological channels including the cardiograph channel for heart data analysis. Our guide on common mistakes by novice polygraph examiners details how training and experience reduce error rates.

What Is Base Rate and Why Should You Care?

Base rate refers to the proportion of individuals in a tested population who are actually deceptive. This is arguably the most important — and most misunderstood — variable in polygraph accuracy discussions.

Consider two scenarios with the exact same polygraph test (90% sensitivity, 90% specificity):

Scenario A — Specific-Issue Test (High Base Rate): A couple comes in for an infidelity polygraph. Based on the circumstances, there might be a roughly 50% chance the examinee is deceptive (high base rate). With 90% sensitivity and 90% specificity, the test performs well in both directions.

Scenario B — Pre-Employment Screening (Low Base Rate): An agency screens 1,000 job applicants. Perhaps only 5% are actually deceptive about disqualifying behavior. That means 950 are truthful and 50 are deceptive. With 90% sensitivity, the test correctly catches 45 of 50 deceptive applicants. But with 90% specificity, it also falsely flags 95 of the 950 truthful applicants. The result: 45 true positives and 95 false positives — meaning more than twice as many innocent people are flagged as guilty ones.

As the 1983 Office of Technology Assessment report demonstrated, even assuming a very high (95%) polygraph validity rate with a 5% base rate of guilt, the predictive value of screening would only be 50% — approximately 47 of 50 guilty persons would be correctly identified, but 47 of 950 innocent persons would also be falsely flagged [16]Verified Scientific Validity of Polygraph Testing: A Research Review and Evaluation
Confirms that with 5% base rate and 95% validity, screening predictive value is only 50%, with approximately equal numbers of true and false positives..

Base Rate Mathematics in Practice

Here is a simplified breakdown for 1,000 people tested with a 90% sensitivity / 90% specificity polygraph at different base rates:

50% base rate (500 deceptive, 500 truthful): 450 true positives, 50 false positives. Of those flagged as deceptive (500 total), 90% are truly deceptive. Excellent predictive value.

10% base rate (100 deceptive, 900 truthful): 90 true positives, 90 false positives. Of those flagged (180 total), only 50% are truly deceptive. Half the deceptive results are wrong.

2% base rate (20 deceptive, 980 truthful): 18 true positives, 98 false positives. Of those flagged (116 total), only about 15% are truly deceptive. The vast majority of positive results are false.

This is not a failure of the polygraph. It is a mathematical certainty that applies to any diagnostic test used in a low base-rate population — the same reason mass screening for rare diseases produces many false positives. The 2003 NRC report specifically flagged screening polygraph programs on these grounds, noting that its accuracy in low base-rate security screening populations is insufficient for standalone decision-making [8]Verified The Polygraph and Lie Detection
Confirms NRC 2003 review of 57 studies, median accuracy of.85, distinction between screening and diagnostic testing, and CPS discriminant function analysis methodology.. One study using Israel Police data calculated that the posterior probability of a truthful (NDI) outcome being correct was 0.98, while the confidence in a deceptive (DI) outcome being correct was only about 0.65, directly reflecting base rate effects [3]Verified Estimated Accuracy of CQT Polygraph Examination
Confirms CQT field sensitivity of 0.94 and specificity of 0.835 using a novel ground-truth-free estimation method with Israel Police data..

Beyond Sensitivity and Specificity: What You Really Want to Know

When you receive a polygraph result — either Deception Indicated (DI) or No Deception Indicated (NDI) — you do not actually care about sensitivity or specificity in the abstract. What you want to know is:

If the result says 'deceptive,' what is the probability I actually am deceptive? This is the Positive Predictive Value (PPV).

If the result says 'truthful,' what is the probability I actually am truthful? This is the Negative Predictive Value (NPV).

PPV and NPV are what matter to the individual examinee. And unlike sensitivity and specificity (which are fixed properties of the test), PPV and NPV depend on the base rate of the population being tested.

The formulas are: PPV = True Positives / (True Positives + False Positives). NPV = True Negatives / (True Negatives + False Negatives).

As the base rate examples demonstrate, PPV drops dramatically in low base-rate populations. NPV, conversely, tends to be very high in low base-rate populations — because when most people are truthful and the test has decent sensitivity, very few deceptive individuals will escape detection relative to the large number of correctly classified truthful people.

Practical Implications for Consumers

For someone scheduling a private lie detector test about a specific incident — infidelity, theft, false accusation — the base rate is relatively high (one of the parties is likely deceptive about the specific issue). This means the PPV of a positive result is substantially higher than in a mass screening context. A Deception Indicated result from a specific-issue exam conducted by a qualified examiner carries meaningful weight.

For someone undergoing a pre-employment screening polygraph, the base rate is lower. A Deception Indicated result should prompt additional investigation — and the APA Standards of Practice reflect this by recommending a 'successive hurdles' approach for screening, where initial screening results are followed by additional validated testing methods [9]Verified APA Standards of Practice
Confirms APA requirements for validated techniques including 90% accuracy for evidentiary testing and 20% maximum inconclusive rate.. For more on the distinction between multi-issue polygraph tests and single-issue formats, our detailed guide breaks down the differences.

The global evaluation approach in polygraph provides examiners with additional tools to contextualize numerical scores alongside behavioral observations, further improving predictive accuracy in individual cases.

Two Fundamentally Different Contexts

Polygraph examinations fall into two broad categories, and the accuracy profile of each is substantially different.

Screening Polygraph — Used in pre-employment hiring for law enforcement, federal agencies, and security clearances. Questions cover broad categories. The tested population typically has a low base rate of deception. The APA's 2011 meta-analysis found multi-issue screening techniques produced an aggregated decision accuracy of 85% (confidence interval 77%–93%) with an inconclusive rate of 13% [1]Verified Meta-Analytic Survey of Criterion Accuracy of Validated Polygraph Techniques
Confirms 89% accuracy for single-issue diagnostic testing, 85% for multi-issue, 87% overall, with inconclusive rates of 11-13%. Reports sensitivity of.84 and specificity of.77 for diagnostic polygraphs..

Diagnostic Polygraph — Used in criminal defense, private investigations, infidelity testing, and PCSOT. Questions focus on a single, clearly defined issue. The base rate of deception is typically higher. The same APA meta-analysis found event-specific single-issue diagnostic techniques achieved an aggregated decision accuracy of 89% (confidence interval 83%–95%) with an inconclusive rate of 11% [1]Verified Meta-Analytic Survey of Criterion Accuracy of Validated Polygraph Techniques
Confirms 89% accuracy for single-issue diagnostic testing, 85% for multi-issue, 87% overall, with inconclusive rates of 11-13%. Reports sensitivity of.84 and specificity of.77 for diagnostic polygraphs..

The 2003 NRC report drew a clear distinction between these contexts, finding that specific-incident polygraph tests can discriminate lying from truth-telling at rates well above chance [8]Verified The Polygraph and Lie Detection
Confirms NRC 2003 review of 57 studies, median accuracy of.85, distinction between screening and diagnostic testing, and CPS discriminant function analysis methodology.. For screening, the committee noted that accuracy is almost certainly lower than what can be achieved by specific-incident testing [8]Verified The Polygraph and Lie Detection
Confirms NRC 2003 review of 57 studies, median accuracy of.85, distinction between screening and diagnostic testing, and CPS discriminant function analysis methodology..

Why This Distinction Matters for Consumers

This distinction is critical for consumers. If you are taking a specific-issue polygraph about whether your spouse was unfaithful, or whether you committed a specific theft, the accuracy literature is more favorable than the generic numbers sometimes cited by critics who focus exclusively on screening contexts.

The Honts et al. (2021) comprehensive meta-analysis of the Comparison Question Test, which coded data from 138 datasets, found that the CQT can achieve meaningful accuracy levels, with motivation having a positive linear relationship with detection accuracy [17]Verified A comprehensive meta-analysis of the comparison question polygraph test
Confirms meta-analysis of 138 CQT datasets found meaningful accuracy, no publication bias, and positive relationship between motivation and accuracy.. This means that in real-world specific-issue testing — where stakes are high and examinees are genuinely motivated — the test tends to perform at its best.

The DOE's historical use of polygraph testing provides instructive case studies in how screening programs have evolved to address the base rate challenge. Meanwhile, the Relevant-Irrelevant (RI) format represents one of the earliest approaches that has since been superseded by more accurate validated techniques.

What the Research Actually Reports

Published polygraph accuracy data varies by technique, study methodology, and context. Here are the key findings from major reviews and meta-analyses:

NRC 2003 Report: The National Research Council examined approximately 80 research studies that the American Polygraph Association had relied upon, finding that 57 of these had significant methodological limitations [18]Verified Polygraph - Wikipedia
Confirms NRC reviewed approximately 80 studies from APA, found 57 had significant limitations, and that specific-incident testing discriminates lying from truth at rates above chance.. Among the studies that passed minimum quality standards — 37 laboratory studies and 7 field studies — the NRC estimated a median CQT accuracy of.85 using ROC curve analysis [5]Verified A review of the polygraph: history, methodology and current status
Confirms NRC 2003 review included 37 laboratory and 7 field studies meeting minimum standards, with median CQT accuracy of.85. Reports BPS field study review found sensitivity 83-89% and specificity 53-78%.. An analysis of 30 polygraph data sets revealed an overall accuracy of 85%, and seven field studies showed a median accuracy of 89% [19]Verified The Scientific Status of Research on Polygraph Techniques: The Case for Polygraph Tests
Confirms Honts, Raskin, and Kircher (2002) chapter in Modern Scientific Evidence, pages 446-483..

APA Meta-Analytic Survey (2011): This landmark analysis, published in Polygraph, 40(4), reviewed 38 studies that satisfied qualitative and quantitative requirements, encompassing 3,723 examinations scored by 295 scorers [1]Verified Meta-Analytic Survey of Criterion Accuracy of Validated Polygraph Techniques
Confirms 89% accuracy for single-issue diagnostic testing, 85% for multi-issue, 87% overall, with inconclusive rates of 11-13%. Reports sensitivity of.84 and specificity of.77 for diagnostic polygraphs.. Single-issue diagnostic testing produced 89% decision accuracy (CI: 83%–95%), multi-issue techniques produced 85% (CI: 77%–93%), and the overall combination of validated techniques produced 87% decision accuracy (CI: 80%–94%) [1]Verified Meta-Analytic Survey of Criterion Accuracy of Validated Polygraph Techniques
Confirms 89% accuracy for single-issue diagnostic testing, 85% for multi-issue, 87% overall, with inconclusive rates of 11-13%. Reports sensitivity of.84 and specificity of.77 for diagnostic polygraphs..

The American Polygraph Association has compiled 80 research projects published since 1980, involving 6,380 polygraph examinations. Field examination studies showed an average accuracy of 98%, while laboratory simulations produced an average accuracy of 80%, yielding an overall average of 88% across all studies [20]Verified About Polygraph: Lie Detector Accuracy
Confirms APA compiled 80 research projects since 1980 involving 6,380 examinations with overall average accuracy of 88%..

Comparison Question Test (CQT): The 2021 Honts et al. meta-analysis captured 138 datasets and confirmed that the CQT can be accurate, that experimental studies are generalizable, and no publication bias was detected [17]Verified A comprehensive meta-analysis of the comparison question polygraph test
Confirms meta-analysis of 138 CQT datasets found meaningful accuracy, no publication bias, and positive relationship between motivation and accuracy.. Individual field studies have reported sensitivity as high as 94% and specificity of 83.5% [3]Verified Estimated Accuracy of CQT Polygraph Examination
Confirms CQT field sensitivity of 0.94 and specificity of 0.835 using a novel ground-truth-free estimation method with Israel Police data., while CQT studies in criminal cases have shown true positive rates of 87% and true negative rates of 82.2% [4]Verified Polygraph Accuracy of Control Question Test in Criminal Cases
Confirms CQT criminal case true positive rate of 87% and true negative rate of 82.2% with no significant effect of education level on detection accuracy..

Concealed Information Test (CIT): Meta-analyses of CIT research report large effect sizes — Cohen's d of 1.55 for skin conductance, 1.11 for respiration, 0.89 for heart rate, and 1.89 for P300 brain potentials [21]Verified Memory detection with the Concealed Information Test: A meta analysis of skin conductance, respiration, heart rate, and P300 data
Confirms CIT effect sizes: Cohen's d of 1.55 for SCR, 1.11 for RLL, 0.89 for HR, and 1.89 for P300.. The CIT using electrodermal measures has shown an average hit rate of 0.83 and a remarkably low false alarm rate of 0.04 [6]Verified The validity of psychophysiological detection of information with the Guilty Knowledge Test: a meta-analytic review
Confirms CIT meta-analysis found average hit rate of 0.83 and false alarm rate of 0.04 across 80 laboratory studies with 5,198 participants., demonstrating exceptional specificity. Eye-tracking measures have also shown promise as additional CIT detection channels [22]Verified Deception detection by means of eye-tracking in the concealed information test
Foundational research relevant to CIT detection using eye-tracking as an additional physiological channel..

Why Numbers Vary So Much

Several factors create the wide range of reported accuracy figures:

Study type: Laboratory (mock crime) studies have confirmed ground truth but may not fully replicate the emotional stakes of real exams. The NRC noted that laboratory accuracy estimates are almost certainly higher than actual field accuracy [8]Verified The Polygraph and Lie Detection
Confirms NRC 2003 review of 57 studies, median accuracy of.85, distinction between screening and diagnostic testing, and CPS discriminant function analysis methodology.. Field studies have higher ecological validity but often rely on confession-verified ground truth, which introduces selection bias.

Technique: Different polygraph techniques have different published accuracy rates. The APA recognizes 19 techniques as validated as of 2012 [5]Verified A review of the polygraph: history, methodology and current status
Confirms NRC 2003 review included 37 laboratory and 7 field studies meeting minimum standards, with median CQT accuracy of.85. Reports BPS field study review found sensitivity 83-89% and specificity 53-78%., each with its own accuracy profile. For a complete listing, see our guide to the APA validated polygraph technique list.

Scoring method: Manual (hand) scoring and automated computer algorithms can produce different results on the same charts. The OSS-3 eliminates subjectivity by deriving scores from measurement ratios [11]Verified Objective Scoring System – Version 3
Confirms OSS-3 is a computerized scoring algorithm with demonstrable validity across multiple validation samples., while the CPS system uses discriminant function analysis with three key physiological features [8]Verified The Polygraph and Lie Detection
Confirms NRC 2003 review of 57 studies, median accuracy of.85, distinction between screening and diagnostic testing, and CPS discriminant function analysis methodology.. Both approaches have demonstrated accuracy between 85–92% under laboratory conditions [23]Verified Modern Algorithms in Polygraph Data Analysis
Confirms OSS-3 and PolyScore have demonstrated accuracy rates between 85-92% under laboratory conditions..

Examiner skill: The quality of the pre-test interview, question formulation, and chart collection significantly affects outcomes. The NRC acknowledged that the examiner can affect test accuracy, underscoring the importance of proper training and expertise [19]Verified The Scientific Status of Research on Polygraph Techniques: The Case for Polygraph Tests
Confirms Honts, Raskin, and Kircher (2002) chapter in Modern Scientific Evidence, pages 446-483.. The digital workflow from test to report details how modern examiners maintain consistency.

Inclusion of inconclusives: The APA's meta-analysis found inconclusive rates of 11% for diagnostic testing and 13% for screening techniques [1]Verified Meta-Analytic Survey of Criterion Accuracy of Validated Polygraph Techniques
Confirms 89% accuracy for single-issue diagnostic testing, 85% for multi-issue, 87% overall, with inconclusive rates of 11-13%. Reports sensitivity of.84 and specificity of.77 for diagnostic polygraphs.. However, studies at the Department of Defense Polygraph Institute showed no-opinion rates ranging from 0% to 50%, demonstrating enormous variation [8]Verified The Polygraph and Lie Detection
Confirms NRC 2003 review of 57 studies, median accuracy of.85, distinction between screening and diagnostic testing, and CPS discriminant function analysis methodology.. Some studies exclude inconclusive results; others count them as errors. This dramatically affects reported accuracy rates.

A Critical Reading Framework

The next time you encounter a claim like 'polygraphs are 95% accurate' or 'lie detectors are only 65% accurate,' apply this framework:

Ask: Sensitivity, Specificity, or Overall? A single accuracy number usually means overall correct classification rate — the total number of correct decisions divided by total tests. But this obscures critical detail. A test could be 90% accurate overall but have 95% sensitivity and only 80% specificity, or vice versa. Always try to find the breakdown. The APA's research standards now require that descriptions of criterion accuracy include calculations of test sensitivity, test specificity, and inconclusive rates [9]Verified APA Standards of Practice
Confirms APA requirements for validated techniques including 90% accuracy for evidentiary testing and 20% maximum inconclusive rate..

Ask: What Was the Base Rate? If a study tested equal numbers of deceptive and truthful subjects (50/50 base rate), the overall accuracy looks maximally favorable. Real-world screening populations rarely have 50% base rates. The relevance of the study to your situation depends on whether the base rate matches your context.

Ask: What Happened to Inconclusive Results? As the APA illustrates, if 10 examinations yield 7 correct decisions, 1 wrong, and 2 inconclusive, the accuracy is either 87.5% (excluding inconclusives) or 70% (including them as errors) [20]Verified About Polygraph: Lie Detector Accuracy
Confirms APA compiled 80 research projects since 1980 involving 6,380 examinations with overall average accuracy of 88%.. This single methodological choice can swing reported accuracy by nearly 20 percentage points.

Ask: What Technique Was Used? Validated techniques meeting APA standards produce accuracy above 80%, while non-validated techniques can score as low as 61% [9]Verified APA Standards of Practice
Confirms APA requirements for validated techniques including 90% accuracy for evidentiary testing and 20% maximum inconclusive rate.. A claim about 'polygraph accuracy' that does not specify the technique is nearly meaningless.

Ask: Who Published the Study and When? Peer-reviewed research in recognized journals carries more weight than commercial claims. The comprehensive Honts et al. (2021) meta-analysis found no publication bias across 138 CQT datasets [17]Verified A comprehensive meta-analysis of the comparison question polygraph test
Confirms meta-analysis of 138 CQT datasets found meaningful accuracy, no publication bias, and positive relationship between motivation and accuracy., providing reassurance that the positive findings are genuine rather than artifactual.

Putting the Numbers in Perspective

For consumers considering a polygraph examination, the key insights from the research are genuinely encouraging:

Specific-issue diagnostic polygraph exams — the type most commonly used for private testing — achieve the highest accuracy levels, with validated techniques consistently demonstrating 89% or better decision accuracy [1]Verified Meta-Analytic Survey of Criterion Accuracy of Validated Polygraph Techniques
Confirms 89% accuracy for single-issue diagnostic testing, 85% for multi-issue, 87% overall, with inconclusive rates of 11-13%. Reports sensitivity of.84 and specificity of.77 for diagnostic polygraphs.. When conducted by experienced examiners using APA-validated methods, these exams provide meaningful, reliable results.

The most common type of error truthful people worry about — false positives — is actively mitigated by proper technique selection, thorough pre-test interviews, and validated scoring algorithms. The CIT in particular demonstrates remarkable specificity, with false alarm rates as low as 4% [6]Verified The validity of psychophysiological detection of information with the Guilty Knowledge Test: a meta-analytic review
Confirms CIT meta-analysis found average hit rate of 0.83 and false alarm rate of 0.04 across 80 laboratory studies with 5,198 participants..

The polygraph's value extends beyond its accuracy statistics. Research shows that the mere presence of a polygraph makes individuals substantially more likely to disclose information than in standard interviews, making it a powerful investigative tool regardless of the specific accuracy rate for physiological detection alone [17]Verified A comprehensive meta-analysis of the comparison question polygraph test
Confirms meta-analysis of 138 CQT datasets found meaningful accuracy, no publication bias, and positive relationship between motivation and accuracy..

For consumers, the practical advice is clear: choose a qualified examiner who uses APA-validated techniques, ensure the examination format matches your situation (specific-issue diagnostic for targeted questions), and understand that the resulting accuracy profile will be substantially more favorable than the broad screening statistics sometimes cited by critics.

Learn about the historical development of polygraph techniques in our profiles of pioneers like Cleve Backster and Fred Inbau, whose work laid the groundwork for today's validated methods. For an overview of the essential science, our psychophysiology for polygraph examiners guide provides the physiological foundations.