The Control Question Test in Polygraphic Examinations with Actual Controls for Truth

Bradley, MacLaren, and Black (1996) tested a modified version of the Control Question Test in which traditional incriminating, deception-demanding control questions were replaced with unambiguous questions answered truthfully by all participants. Using a three-group mock-crime laboratory design — guilty, innocent-informed, and innocent-uninformed — the study found that this structural modification still yielded detection rates of 86% for guilty subjects and 83–89% for innocent subjects, suggesting that the deceptive framing of control questions is not a necessary condition for CQT validity.

Laboratory analog study using a three-group mock-crime design (guilty, innocent-informed, innocent-uninformed) with a modified CQT format featuring unambiguous, truthfully answered control questions, measuring galvanic skin response as the primary physiological indicator.

MacLaren's 1996 study challenged a fundamental assumption of the Control Question Test by replacing traditional ambiguous, deception-inducing control questions with unambiguous ones that could be answered truthfully. The laboratory study used three participant groups (guilty of mock crime, innocent but informed of crime details, and innocent uninformed) and measured galvanic skin response. Results showed the modified CQT achieved comparable detection rates to standard CQT formats. This finding suggested that the theoretical foundation requiring deceptive control questions may be flawed, and that physiological differentiation between guilty and innocent subjects could persist with truthful controls.

These findings challenge core theoretical assumptions about CQT design and suggest polygraph accuracy may not depend on embedding deception in control questions. The research opens possibilities for developing more methodologically sound polygraph formats while maintaining diagnostic effectiveness.

Background & Context

The Control Question Test (CQT) — also known as the Comparison Question Test — has been the dominant format in polygraphic examinations for decades. Its fundamental design pairs crime-relevant questions with "control" questions intended to serve as a physiological baseline. The defining feature of traditional control questions is that they are deliberately crafted to be incriminating, ambiguous, and worded in such a way that the subject is likely to answer deceptively — the theory being that innocent subjects will react more strongly to these vague, personal questions than to the specific crime-relevant ones.

This design assumption, however, had long attracted scientific scrutiny. Critics argued that building deception into the control question format introduced a fundamental methodological contradiction: how can a test claim to detect lies if it relies on participants telling lies as part of its baseline mechanism? Bradley, MacLaren, and Black (1996) set out to directly challenge this embedded assumption, asking whether the CQT could function effectively even when its control questions were unambiguous and answered truthfully.

Published in Perceptual and Motor Skills, this study arrived at an important moment in polygraph science — a period of growing debate about the theoretical underpinnings of the CQT. By testing a structurally modified version of the test, the research helped disentangle the question of whether the CQT's accuracy depends on the deceptive nature of its control questions, or whether the fundamental physiological differentiation between guilty and innocent subjects could persist regardless.

Research Design & Methodology

Subjects who were guilty of a mock crime, innocent and informed of the details of the crime, or innocent and uninformed of the details were examined on the polygraph with a modified version of the Control Question Test. This three-group design was particularly valuable because it distinguished not just between guilty and innocent, but also between informed-innocent and uninformed-innocent participants — a distinction rarely made in earlier CQT research, which allowed the investigators to assess whether knowledge of crime details (without participation) affected physiological reactivity.

The core methodological innovation was the transformation of the control questions themselves. The researchers challenged the necessity of using incriminating, ambiguous control questions which may demand a lie as an answer, and instead created and tested control questions that were unambiguous and answered truthfully. In this modified version of the CQT, the control questions were clear and direct questions, similar to those used in the Concealed Information Test — for example, if the relevant question was "Did you steal $100?" a possible control would be "Did you steal $150?"

Key design elements of the study include:

• Three participant groups: Guilty (mock crime perpetrators), innocent-informed (knew crime details), and innocent-uninformed
• Primary physiological measure: Galvanic skin response (electrodermal activity), consistent with standard polygraph methodology
• Modified CQT format: Truthfully answerable, unambiguous control questions replacing the traditionally deception-demanding controls
• Setting: Laboratory analog study conducted at the University of New Brunswick, Saint John, Canada
• Classification: Registered as a Randomized Controlled Trial in PubMed, indicating random assignment to conditions

Results & Key Findings

The modified CQT with truthful control questions achieved detection rates comparable to those reported for the standard CQT format. The results with the modified questions showed correct identification of 86% of the guilty subjects and 83% to 89% of the innocent subjects. This range for innocent subjects reflects differences between the informed and uninformed innocent groups, a distinction that carries its own theoretical importance.

Key numerical outcomes from the study include:

• Guilty detection rate: 86% — correctly classified as deceptive
• Innocent specificity: 83%–89% — correctly classified as truthful (range across informed vs. uninformed innocent groups)
• The informed-innocent group (who knew crime details) showed somewhat lower specificity than the uninformed-innocent group, consistent with the hypothesis that knowledge of crime-relevant information elevates physiological arousal even in truthful subjects
• Overall accuracy in the mid-to-upper 80% range — competitive with benchmark figures often cited for the standard CQT in laboratory settings

Cooperation instructions were incorporated into the altered CQT, and it was reported that, although informed innocent participants recognized the relevant questions as threatening, their cooperation (truthful answers to the relevant questions) was associated with lower physiological responses than those of the less cooperative guilty participants. This finding underscores that the mechanism driving CQT accuracy may be rooted less in the deceptive framing of control questions and more in the differential threat appraisal and motivational states of guilty versus innocent examinees.

Discussion & Significance

The study's most consequential implication is that the deceptive nature of traditional CQT control questions may not be a necessary condition for the test's diagnostic validity. For decades, the standard CQT had been defended — and simultaneously attacked — on the grounds that its control questions work precisely because they elicit lies. Bradley et al.'s findings suggest this assumption is empirically testable and potentially incorrect; physiological differentiation between guilty and innocent subjects appears robust enough to survive a structural overhaul of the control question format.

This work also has meaningful implications for the informed-innocent problem — the recognized risk that individuals who are genuinely innocent but possess detailed knowledge of a crime may generate deceptive-appearing physiological profiles. The three-group design revealed that informed-innocent subjects, while still predominantly classified correctly, did show elevated reactivity compared to uninformed innocents, confirming that knowledge contamination is a real and measurable source of error in CQT-based polygraph exams.

The approach pioneered by Bradley, MacLaren, and Black (1996) was further studied by Cullen and Bradley (2004), demonstrating that the modified CQT attracted subsequent replication and scholarly attention. The study thus served as a methodological springboard for exploring hybrid formats that might combine the structural simplicity of the Concealed Information Test with the broader applicability of the CQT format.

Limitations & Considerations

As a laboratory analog study, the research carries the limitations common to all mock-crime paradigms. Participants in a simulated theft scenario face fundamentally lower stakes than real-world examinees who may face criminal prosecution, job loss, or social stigma — factors known to amplify physiological reactivity in field settings. The generalizability of laboratory accuracy figures to operational polygraph examinations must therefore be treated with appropriate caution.

Additional considerations include:

• Sample characteristics: The study used adult participants recruited at a Canadian university, which may not represent the demographic diversity of real-world examinees
• Examiner effects: The laboratory context and administration by a professor (as noted in related work) may differ from the dynamics of a professional polygraph examination
• Limited physiological channels: Primary reliance on galvanic skin response, while a validated measure, does not capture the full multi-channel physiological profile used in standard polygraph practice
• The informed-innocent group's lower specificity raises unresolved questions about how knowledge-possession affects classification accuracy in investigative contexts where witnesses and suspects may share information

Practical Applications

For polygraph practitioners and test format researchers, this study offers a constructive challenge to longstanding assumptions about control question construction. The finding that truthfully answered, unambiguous control questions can sustain high detection rates opens the door to modified CQT formats that are more ethically defensible — formats that do not require examiners to engineer deception in examinees as a prerequisite for test functionality. This is relevant to ongoing efforts to improve the scientific standing of the CQT and reduce its vulnerability to critiques centered on its deception-dependent design.

For consumers and legal practitioners evaluating polygraph evidence, the study reinforces that correctly designed physiological comparison questions — even when answered truthfully — can yield meaningful diagnostic differentiation between deceptive and truthful subjects. The informed-innocent finding is a practical reminder that polygraph examiners must carefully consider examinee knowledge levels before drawing conclusions, as awareness of crime details can influence physiological profiles independent of actual guilt.