What Is SCAN Statement Analysis, and Does It Work?

By Murder Book Editorial TeamPublished by Gungrounds, the studio making Murder BookPublished 3 October 2026Updated 3 October 2026

SCAN (Scientific Content Analysis) is a statement analysis technique developed by former Israeli polygraph examiner Avinoam Sapir. It claims that truthful and deceptive people use language differently, and that an analyst can spot deception by examining specific linguistic features in a written statement. SCAN is used by law enforcement agencies in several countries, but the scientific evidence does not support its claims. Multiple empirical studies have found that SCAN criteria cannot reliably distinguish truthful from fabricated statements, and researchers have called for its use to be discontinued in its current form. It lacks a clear theoretical foundation, a standardised list of criteria and adequate empirical validation.

Key takeaways

SCAN was developed by Avinoam Sapir and is used by some law enforcement agencies worldwide, but it has not been validated by independent scientific research.
A 2016 study found that SCAN could not distinguish between truthful and fabricated accounts of a negative event.
SCAN lacks a well-defined, standardised list of criteria and a consistent scoring system.
Researchers have described SCAN as lacking the empirical support needed for forensic use and have recommended against its application.
Evidence-based alternatives such as criteria-based content analysis and the verifiability approach have stronger research backing, though they too have limits.

How does SCAN work?

SCAN asks a subject to write a statement about the event in question, usually beginning with "What I did on [date]." The analyst then examines the written text for specific linguistic features that Sapir claims differ between truthful and deceptive accounts.

Examples of SCAN criteria include:

Pronouns: changes in pronoun use (shifting from "we" to "I" or dropping pronouns altogether) are interpreted as signs of distancing or deception.
Tense changes: shifts from past tense to present tense within a narrative are treated as significant.
Missing time: gaps in the chronological account are flagged.
Social introductions: how the writer introduces other people in the statement is examined.
First person singular: the presence or absence of "I" at key points.
Emotions: where feelings are mentioned in the narrative, and where they are absent.
Unnecessary connections: words like "so," "therefore" and "because" are scrutinised for what they might reveal about the writer's reasoning.

Sapir developed these criteria from his experience as a polygraph examiner. He did not publish them in peer-reviewed journals, and the list of criteria has never been formally standardised. Different sources describe different numbers of criteria, and there is no consistent scoring system.

What does the research say about SCAN?

The short answer is that SCAN has not held up under scientific testing.

Bogaard, Meijer, Vrij and Merckelbach (2016) published a study in Frontiers in Psychology titled "Scientific Content Analysis (SCAN) Cannot Distinguish Between Truthful and Fabricated Accounts of a Negative Event." They asked 117 participants to write one truthful and one fabricated statement about a recent negative event. All statements were analysed using 11 SCAN criteria. The result: SCAN was not able to correctly classify the statements as true or fabricated. The authors concluded that "lacking empirical support, the application of SCAN in its current form should be discouraged."

Earlier research pointed in the same direction. Studies have consistently found that SCAN criteria do not reliably separate truthful from deceptive statements. The U.S. National Institute of Justice has funded research on statement analysis, but the results have not supported SCAN's claims.

A key problem is the lack of theoretical grounding. Unlike reality monitoring, which builds on Johnson and Raye's memory theory, or criteria-based content analysis, which rests on the Undeutsch hypothesis, SCAN does not offer a clear explanation of why the linguistic features it examines should differ between truth tellers and liars. Sapir developed the criteria based on practical experience, not systematic research, and he has not provided the kind of theoretical framework that would allow other researchers to test and refine the approach.

Why is SCAN still used?

Despite the lack of scientific support, SCAN has been adopted by law enforcement agencies in the United States, Canada, the United Kingdom, the Netherlands and other countries. Several factors may explain this:

It is intuitive. The idea that you can spot deception by carefully reading someone's words has strong surface appeal. Many of the criteria, such as pronoun shifts and missing time, feel plausible to practitioners.

It is practical. SCAN requires only a written statement and an analyst's attention. No special equipment, no laboratory, no lengthy coding procedure. This makes it attractive to investigators working under time pressure.

Training is commercially available. Sapir offers SCAN training courses, and agencies have invested in sending officers to them. Once an organisation has invested in training, there is institutional momentum behind continued use.

The distinction between validated and unvalidated tools is not always clear in practice. Jupe and Denault argued in a 2019 article in Psychiatry, Psychology and Law that some people have given police officers, lawyers and judges information about nonverbal communication and verbal cues that has not been empirically tested or supported by scientific theory. SCAN fits this pattern.

How does SCAN compare with evidence-based methods?

The contrast with validated methods is instructive.

Feature	SCAN	CBCA (part of SVA)	Verifiability approach
Theoretical basis	None published	Undeutsch hypothesis	Strategic self-presentation theory
Number of criteria	Varies, not standardised	19, well-defined	Two main measures (verifiable count, proportion)
Meta-analytic support	No supporting meta-analysis	g = 0.98 across 55 studies	d = 0.49 across 28 comparisons
Scoring system	No standardised system	Structured coding	Structured coding
Peer-reviewed validation	Studies show it does not work	Extensive, with known limits	Growing, with known limits

CBCA and the verifiability approach are far from infallible. Both have limits, and neither can determine with certainty whether a specific person is lying. But both have a research base that shows they detect real differences between truthful and fabricated accounts. SCAN does not have this evidence.

For a broader look at the difficulty of detecting deception, see Can You Actually Tell When Someone Is Lying?.

Does using an unsupported method matter?

It matters a great deal. When an investigator uses a tool that cannot reliably distinguish truth from lies but believes that it can, the risk of tunnel vision increases. A statement flagged as "deceptive" by SCAN may lead investigators to focus on that person while ignoring other leads. A statement that "passes" SCAN may lead them to overlook genuine inconsistencies.

In criminal justice, the stakes are high. Decisions about whether to pursue or abandon a line of investigation, whether to charge a suspect, and how to weigh evidence at trial can all be affected by an investigator's belief about whether someone is telling the truth. Using an unvalidated tool to make those judgments introduces a source of error that evidence-based methods, for all their imperfections, are designed to reduce.

How does Murder Book approach statement analysis?

Murder Book: The Art of Interrogation is an upcoming narrative interrogation game from Gungrounds, an independent studio in Zagreb, Croatia. Its press kit names Fabrication, Avoidance, Narrative Gaps and Emotional Leakage among the behaviours players examine. Development screenshots show an evidence board linking statements to contradictions, such as "I NEVER met her!" linked to a fabricated description of a gun.

The game does not present any single technique as a reliable detector of deception. Instead, it asks players to weigh multiple kinds of evidence, compare statements against known facts, and decide how far to press. This is closer to how validated research methods work than to the binary "truthful or deceptive" judgments that SCAN promises. In Murder Book, as in the research, behaviour and language are reasons to ask more questions, not final answers.

Murder Book is in development for PC (Steam) and mobile, with release currently planned for the end of Q2 2027.