The validity of detection of deception for multiple issues

This 1989 Department of Defense Polygraph Institute study examined whether polygraph examinations can effectively detect deception across multiple issues. One hundred military subjects were assigned to innocent or guilty conditions involving zero to three mock espionage crimes, then tested using either multiple-issue or single-issue examination formats.

Laboratory analog study with 100 subjects in four conditions (innocent, guilty of one, two, or three mock crimes), tested by 13 DoDPI instructors using either multiple-issue tests covering all three crimes or series of three single-issue examinations, scored using standard DoDPI methods.

This Department of Defense Polygraph Institute study addressed the under-researched question of whether polygraph examinations are effective when multiple deception issues are present in a single test. The research compared single-issue and multiple-issue examination formats using 100 military subjects tested on realistic mock espionage scenarios. While 67% of guilty subjects were called deceptive to at least one crime and 93% were detected when inconclusives were excluded, the critical finding was that examiners could not reliably identify which specific questions subjects answered deceptively. This distinction between detecting that deception occurred versus identifying which issues involved deception has profound implications for security screening and criminal investigation practices. The study fundamentally challenged core assumptions underlying multiple-issue polygraph testing.

The finding that examiners cannot reliably pinpoint specific deceptive responses—even when detecting overall deception—undermines the practical utility of multiple-issue examinations for investigations requiring identification of particular false statements. This research suggests that current polygraph interpretation methods may not adequately support the follow-up questioning and evidence development that depend on accurate issue-specific identification, with significant implications for both security screening policy and criminal investigation protocols.

Background & Context

The Barland, Honts, and Barger (1989) study represents a landmark investigation into a critical but under-researched question in polygraph science: Can polygraph examinations effectively detect deception when multiple issues are being investigated within the same examination? This question has profound implications for both criminal investigations and security screening contexts where examiners routinely face situations involving potential deception about more than one matter.

Prior research by Raskin, Kircher, Honts, and Horowitz (1989) had found that accuracy declined significantly when subjects were truthful about some issues but deceptive about others within the same examination. The Department of Defense Polygraph Institute (DoDPI) undertook this research to address these concerns in a systematic laboratory analog study, comparing the effectiveness of single-issue versus multiple-issue examination formats—a comparison with direct relevance to field practice.

Research Design & Methodology

This analog study utilized 100 subjects who were assigned to one of four conditions: (1) innocent, (2) guilty of one mock crime, (3) guilty of two mock crimes, or (4) guilty of three mock crimes. The average age of subjects was 20.2 years, and participants were recruited from Fort McClellan military personnel.

Half the subjects were tested with a multiple-issue test format containing questions about all three crimes under investigation, while the remaining subjects were tested using a series of three single-issue examinations, each containing questions about only one crime. Scoring rules taught at DoDPI were used by the original examiners to evaluate the tests. Thirteen DoDPI instructors conducted the examinations, ensuring ecological validity through the use of experienced, trained polygraph professionals.

The mock crimes involved realistic espionage scenarios to simulate security screening contexts where multiple-issue testing is commonly employed.

Results & Key Findings

Guilty subjects who committed one or more crimes were called deceptive to at least one of the crimes 67% of the time, deceptive to none of the crimes 5% of the time, and 28% were reported as inconclusive. Excluding inconclusives, 93% of guilty subjects were classified as deceptive to at least one crime. This represents the overall detection rate when considering whether deception was identified anywhere in the examination.

However, the findings revealed a critical distinction between detecting that someone was deceptive versus detecting which specific issues they lied about:

• An important question addressed was whether polygraph-based judgments of deception were attributable to readings indicating deception on questions the examinee actually answered deceptively, or to false positive readings on other questions answered truthfully.
• Several studies, including this one, could not consistently show accuracy in identifying the specific questions that were answered deceptively, directly contradicting basic assumptions that guide polygraph use.
• The study demonstrated that overall differences in correct identification between single-issue and multiple-issue formats were measurable, though the ability to pinpoint specific deceptive responses remained problematic.

Innocent subjects fared considerably better, highlighting an asymmetry in test performance that would become a recurring theme in screening polygraph research.

Discussion & Significance

This research fundamentally challenged assumptions about multiple-issue polygraph testing. The finding that examiners cannot reliably distinguish truthful from deceptive responses—even when they can distinguish truthful from deceptive examinees—has practical implications: if examiners obtain evidence of deception and follow up by focusing on the question that triggered their judgment, they are no more likely to be focusing on the correct question than if guided by chance.

The study contributed to a growing body of evidence in the late 1980s and early 1990s that security screening polygraph examinations performed substantially worse than specific-issue criminal examinations. Overall results were only slightly better than chance, reflecting a high bias toward truthful outcomes, and subsequent research replicated and extended the finding that false negative outcomes predominated with national security screening tests. This work was instrumental in prompting reforms to government screening polygraph programs.

The research also highlighted the tension between operational needs (investigating multiple security concerns efficiently) and scientific validity (accurately identifying deception on specific issues). This remains a fundamental challenge in screening contexts where time constraints and multiple topics make single-issue testing impractical.

Limitations & Considerations

While the study mimicked one aspect of true screening by making examiners uncertain which item was "relevant," in other respects it remained far from normal screening: the mock transgression was highly salient to all guilty examinees, and everyone involved knew the critical event was a specific staged transgression. Real-world screening involves effectively infinite possible deceptive answers and less psychological salience for potential transgressions.

The laboratory analog design, while methodologically rigorous, cannot fully replicate the consequences, motivations, and emotional stakes of actual security screening or criminal investigations. The 67% detection rate for guilty subjects (including inconclusives) and 93% excluding inconclusives may overestimate field performance given the controlled nature of the scenario and the fact that all participants knew they were in a research study.

Practical Applications

This research has direct implications for polygraph examiners and agencies conducting multiple-issue examinations. Examiners should recognize that even when they correctly identify an examinee as deceptive overall, they may not be able to reliably determine which specific issues prompted the deceptive physiological responses. This argues for caution in post-test interrogations and in making conclusions about specific security violations based solely on polygraph chart patterns.

For security agencies, the findings suggest that multiple-issue screening examinations trade efficiency for accuracy. When possible, focusing examinations on fewer, more salient issues may improve diagnostic precision. The research also underscores the importance of using polygraph results as one component of a comprehensive evaluation rather than as a standalone determinant of security suitability.