How Accurate Are Polygraph Tests Really?

Polygraph accuracy is one of the most searched and most misunderstood questions in forensic psychophysiology. Journalists often ask whether a polygraph is a “lie detector”. Lawyers ask whether a result can be relied upon. Therapists ask whether it can support disclosure work. Private clients ask whether it will finally resolve uncertainty.

The scientifically honest answer is that a polygraph does not literally detect lies. It records physiological activity associated with attention, emotional salience, threat appraisal, orienting responses and autonomic arousal while an examinee answers carefully formulated questions. The interpretation of those responses may provide useful decision-support information, but it is not infallible proof of truth or deception.

This article explains polygraph accuracy in a way that is useful for journalists, legal professionals, therapists, documentary researchers and high-intent clients. It covers sensitivity, specificity, positive predictive value, negative predictive value, base rates, Bayesian interpretation, false positives, false negatives, examiner effects, screening versus specific-issue testing, and why “95% accurate” claims can be misleading.

Key points for journalists

• A polygraph does not literally detect lies; it measures physiological responses that may be associated with deception under controlled questioning.
• Accuracy is not one number. Sensitivity, specificity, PPV and NPV answer different questions.
• Base rates matter. A result has different meaning in low-prevalence screening than in a focused specific-issue examination.
• False positives and false negatives are possible, even when a technique performs above chance in research.
• Meta-analyses are useful summaries of research performance, but they do not guarantee that every real-world test performs identically.
• Examiner skill, question construction, examinee suitability and context can materially affect the quality of an examination.
• Polygraph results should not be reported as proof, confession, diagnosis or legal determination.

Polygraphs do not literally detect lies

The phrase “lie detector test” is familiar, but imprecise. A polygraph instrument records physiological channels such as respiration, cardiovascular activity and electrodermal activity. These channels do not contain a direct marker for lying. There is no physiological signal that means, by itself, “this person is deceptive”.

Instead, a properly conducted polygraph examination uses a structured interview, suitability assessment, carefully reviewed questions and validated scoring methods to examine whether responses to relevant questions differ from responses to comparison questions. The result is an inference from a pattern of data, not a mechanical reading of truth.

This distinction matters. It is why polygraph results should be interpreted with professional caution, particularly in legal, safeguarding, therapeutic or media contexts.

Sensitivity and specificity in plain English

Two terms are central to any serious discussion of lie detector test accuracy: sensitivity and specificity.

• Sensitivity is the test’s ability to correctly identify people who are deceptive about the issue being tested. If a technique has high sensitivity, it has a relatively low false negative rate.
• Specificity is the test’s ability to correctly identify people who are truthful about the issue being tested. If a technique has high specificity, it has a relatively low false positive rate.

In everyday terms, sensitivity asks: if deception is actually present, how often does the test detect it? Specificity asks: if deception is not present, how often does the test avoid wrongly flagging the person?

These are important measures, but they are not the same as the probability that a particular result is correct. For that, we need positive predictive value and negative predictive value.

PPV and NPV: the numbers most people actually want

When people ask how accurate are polygraph tests, they often want to know: if someone fails, how likely is it that they were actually deceptive? If someone passes, how reassured should we be?

Those questions are answered by predictive values:

• Positive Predictive Value (PPV): among people who receive a result indicating deception, the proportion who are actually deceptive.
• Negative Predictive Value (NPV): among people who receive a result indicating no deception, the proportion who are actually truthful on the tested issue.

PPV and NPV depend not only on polygraph sensitivity and specificity, but also on polygraph base rates: how common the target behaviour is in the tested population. This is where many simplistic accuracy claims go wrong.

Why base rates change the meaning of a result

A base rate is the underlying prevalence of the behaviour or condition being tested. In a medical screening programme, it is the proportion of people who actually have the disease. In polygraph testing, it is the proportion of examinees who are actually deceptive about the issue being examined.

Base rates vary sharply by context. A specific-issue examination about a defined allegation may have a moderate or high prior probability because there is already a complaint, disclosure, inconsistency or evidential concern. By contrast, a broad screening polygraph may involve a population where the target behaviour is uncommon.

This is why the same sensitivity and specificity can produce very different PPV and NPV in different settings.

A simple worked example

Imagine a hypothetical polygraph technique with 90% sensitivity and 90% specificity. These are deliberately simple figures for illustration, not a claim about any specific case or technique.

Scenario A: low base-rate screening

Suppose 1,000 people are screened and only 5% are actually deceptive about the target issue. That means 50 are deceptive and 950 are truthful.

Even with strong hypothetical sensitivity and specificity, a low base rate means many positive results may be false positives. This is not unique to polygraph testing. It is a property of all imperfect screening tests.

Scenario B: focused specific-issue testing

Now imagine 1,000 specific-issue cases where the base rate is 50% because each case already involves a focused allegation or concern. That means 500 are deceptive and 500 are truthful.

The test characteristics have not changed. The interpretation has changed because the base rate changed. This is the Bayesian point: evidence updates a prior probability. It does not float free of context.

Bayesian interpretation: from prior probability to posterior probability

Bayesian reasoning provides a disciplined way to understand polygraph results. Before the test, there is a prior probability based on the case context, referral information, disclosures, collateral material and the plausibility of alternative explanations. The polygraph result then shifts that probability up or down.

A result indicating deception in a moderate-base-rate, well-structured specific-issue examination may produce substantial information gain. The same result in a low-base-rate screening context may justify further enquiry, but it should not be treated as proof. This is explored further in the Centre’s articles on polygraph accuracy, base rates and predictive value and prior probabilities in screening polygraph.

False positives and false negatives

A false positive polygraph result occurs when a truthful person is incorrectly classified as deceptive. False positives can have serious consequences, especially in employment, relationships, safeguarding or legal contexts. They may lead to suspicion, disciplinary action, relationship breakdown or pressure to confess.

A false negative occurs when a deceptive person is incorrectly classified as truthful. False negatives can also be serious, because they may provide inappropriate reassurance or delay further investigation. Responsible interpretation requires acknowledging both error types.

False positives and false negatives are not simply mathematical abstractions. They are ethical and practical risks. This is why polygraph findings should be reported carefully and considered alongside the quality of the examination, collateral information and the wider professional context.

False confession risk and ethical caution

Polygraph testing can create pressure. When an examinee is told that a test indicates deception, some may feel compelled to confess, elaborate or agree with suggestions in order to escape distress, please authority figures, reduce conflict or avoid perceived consequences. This is especially relevant where there is coercion, domestic conflict, legal pressure, trauma, shame, neurodivergence, cognitive vulnerability or acute emotional distress.

Professionals should therefore avoid using polygraph results as a coercive lever. A failed result should not be used to force an admission. Where a disclosure follows a polygraph result, the disclosure still requires careful evaluation. In therapeutic contexts, additional attention should be paid to emotional stability, consent and suitability. The Centre discusses related issues in its guidance on safeguarding and polygraph examinations and therapeutic disclosure statement work.

Screening polygraph versus specific-issue polygraph

A screening polygraph is designed to look for possible undisclosed behaviours across a category of concern, such as compliance, security risk or post-conviction monitoring. Screening may be useful for risk management, but low base rates can reduce the PPV of positive results. In screening, a concerning result is often best treated as a reason for further review, not as a stand-alone conclusion.

A specific issue polygraph examines a defined allegation or event. The question is narrower, the behaviour is usually more operationally defined, and the base rate may be higher because there is already a specific concern. Specificity of the issue can improve interpretability, provided the examinee is suitable and the questions are well constructed.

These two contexts should not be treated as equivalent. Claims about polygraph sensitivity specificity from one type of setting do not automatically transfer to every other setting.

Examiner effects and question construction

Polygraph accuracy is not only about the instrument. Examiner effects matter. The examiner conducts the pre-test interview, assesses suitability, formulates questions, explains procedures, manages examinee understanding, collects the data and applies scoring rules. Poor practice at any point can degrade the value of the result.

Question construction is especially important. Questions must be clear, behaviourally defined, answerable with yes or no, and understood in the same way by the examinee and examiner. Ambiguous questions increase error risk. This is why the Centre places substantial emphasis on polygraph question formulation and suitability assessment before testing.

Meta-analyses are useful, but not a guarantee

Meta-analyses help summarise research findings across studies. They are useful because they reduce reliance on single studies, anecdote or marketing claims. They can provide estimates of average performance for validated techniques under defined conditions.

However, meta-analytic averages are not guarantees. Real-world performance can vary with technique, examiner competence, examinee characteristics, case complexity, question quality, testing environment, countermeasure risk, scoring method and referral context. A meta-analysis may support the conclusion that a method performs above chance and has practical value; it does not mean that every individual examination is “95% accurate”.

Why “95% accurate” claims are often misleading

Claims that polygraphs are “95% accurate” are often misleading because they compress several different concepts into one impressive-sounding number. They may fail to specify whether the figure refers to sensitivity, specificity, overall accuracy, a particular technique, a particular study design, confirmed cases, laboratory research or selected field data.

They also often omit base rates. A test can have high sensitivity and specificity while still producing a low PPV in low-prevalence screening. Conversely, a negative result may be less reassuring in a high-prevalence population. The scientifically responsible position is not to dismiss polygraph evidence, but to interpret it probabilistically and contextually.

When polygraph results can be useful

Polygraph results can be useful when the examination is properly conducted, the issue is suitable, the questions are specific, the examinee understands the process, and the result is interpreted as decision-support information rather than infallible proof.

In legal settings, polygraph findings may assist case strategy, investigation or risk assessment, but should not be treated as a verdict. In therapy, they may support disclosure, accountability and treatment planning, but should not replace clinical judgement. In documentary or journalistic contexts, they may be newsworthy, but should be explained with the caveats that accuracy depends on context, base rates and professional quality.

Conclusion

So, how accurate are polygraph tests really? The best answer is: more informative than unaided guesswork when properly conducted, but not a literal lie detector and not infallible proof. Accuracy depends on sensitivity, specificity, base rates, question formulation, examiner competence, examinee suitability and the context in which the test is used.

Polygraph results can provide useful decision-support information when properly conducted and interpreted cautiously. They should not be used as a stand-alone determination of truthfulness, deception, guilt or credibility. The most scientifically defensible approach is balanced: neither dismissing polygraph evidence nor overstating it, but placing it within a broader evidential, clinical or investigative framework.

This article is provided for general information and educational purposes. It is not legal advice, clinical advice or a substitute for professional consultation. Polygraph results should be interpreted cautiously and in context. Where safeguarding, legal, coercion or mental health concerns are present, appropriate professional advice should be sought.

Frequently Asked Questions