Can we trust counterintelligence polygraph tests?

This 2000 narrative review by V.V. MacLaren evaluates the validity and utility of the Test for Espionage and Sabotage (TES), the Department of Defense's updated counterintelligence polygraph screening format that replaced the older Counterintelligence Scope Polygraph. Drawing on laboratory simulation studies and five years of DoD operational data covering over 43,000 tests, MacLaren argues that TES demonstrates acceptable accuracy and that a repeat-testing protocol substantially reduces false positive risk, supporting the continued use of polygraph screening in national security contexts.

Narrative literature review and conditional probability analysis synthesizing two DoDPI laboratory simulation studies of TES accuracy, five years of DoD Congressional report data from 43,648 administered tests, and Bayesian modeling of repeated-testing screening scenarios.

Following high-profile espionage cases like Aldrich Ames, MacLaren evaluated whether the Test for Espionage and Sabotage (TES) could effectively replace the inadequate Counterintelligence Scope Polygraph for security screening. The review synthesized laboratory studies showing 83.3% sensitivity and 90.7% specificity, along with five years of operational DoD data covering over 43,000 tests. Using conditional probability modeling, MacLaren demonstrated that repeat-testing protocols could substantially reduce false positive rates while maintaining security effectiveness. The analysis concluded that TES represents a meaningful improvement over previous screening methods when properly implemented.

The findings suggest TES can provide reliable counterintelligence screening when combined with repeat-testing protocols, offering improved accuracy for personnel security decisions in sensitive government positions.

Background & Context

In the late 1990s, a cascade of high-profile espionage cases — most notably the Aldrich Ames scandal — exposed serious vulnerabilities in U.S. government security vetting procedures. Fear of increased espionage activity had already prompted systematic appraisal of polygraph screening in the 1980s, during which several studies found the Counterintelligence Scope Polygraph (CSP) technique to be inadequate as a means of identifying persons involved in espionage activity. These failures made institutional reform urgent.

Recent developments at America's national laboratories had drawn fresh public attention to polygraph security screening, particularly after allegations of penetration by the People's Republic of China prompted the Department of Energy to initiate polygraph counterintelligence screening of employees with access to sensitive information. Critics simultaneously challenged both the accuracy of the tests and their intrusiveness.

Against this backdrop, Vance MacLaren of the University of New Brunswick authored this 2000 review article to assess whether the Test for Espionage and Sabotage (TES) — the newer screening instrument developed to replace the CSP — could actually be trusted to protect national security. The paper reviews open-source literature on polygraph security screening procedures currently in use by the US Department of Defense.

Research Design & Methodology

This paper is a narrative literature review and conditional probability analysis, not a primary empirical study. MacLaren synthesized evidence from multiple sources: published laboratory validity studies, Department of Defense Polygraph Institute (DoDPI) annual reports to Congress, and prior conditional probability models developed by researchers such as Honts (1991).

Key design elements of the review include:

• Laboratory validity synthesis: Integration of two large-scale DoDPI simulation studies (1997; 1998) of TES accuracy across both "programmed guilty" and "programmed innocent" participant groups
• Field data extraction: Analysis of operational DoD polygraph statistics spanning five fiscal years and 43,648 administered tests
• Bayesian/conditional probability modeling: Application of prior-probability reasoning to simulate real-world screening scenarios, including a "two-strikes" repeated-testing model
• Critique of prior assessments: Re-evaluation of Honts' (1991, 1994) false negative estimates using corrected base-rate assumptions

Earlier polygraph screening techniques had been subjected to systematic appraisal concluding that the CSP was inadequate for identifying espionage suspects. MacLaren's methodology specifically targeted whether the updated TES procedure had overcome those documented shortcomings, drawing on both controlled laboratory findings and real-world administrative data.

Results & Key Findings

The headline finding is that TES demonstrates acceptable accuracy under laboratory conditions and that a repeat-testing protocol can sharply reduce false positive risk in field settings. MacLaren synthesized a range of quantitative outcomes across both laboratory and operational data.

Laboratory Accuracy (TES — two pooled studies):

• 83.3% sensitivity (guilty detection rate): In the two simulation studies, 50 of 60 guilty participants were correctly identified.
• 90.7% specificity (innocent correct rate): Of 108 participants in programmed innocent groups, 98 were correctly identified.
• Combined laboratory samples: 60 guilty participants and 108 innocent participants across both DoDPI studies

Field Operational Data (DoD, five fiscal years):

• The Department of Defense administered 43,648 counterintelligence screening tests, in which 902 people made admissions relevant to security issues.
• An additional 148 tests produced either "significant physiological responses" or "inconclusive" outcomes; of the 963 flagged cases, only 24 resulted in adverse action following subsequent investigation.
• An additional 97 cases were still pending investigation or adjudication at the time reports were presented to Congress.

The "Two Strikes" Repeated-Testing Model: MacLaren demonstrated mathematically that requiring individuals who fail one TES to be re-tested dramatically reduces false positive risk. On a second test of those who initially failed, 86 innocent people would fail again and 69 guilty ones would fail, yielding 155 cases of double-failures — of which 45% are foreign agents and 55% are not. This is a substantial improvement in positive predictive value compared to acting on a single test result.

Discussion & Significance

MacLaren's analysis makes a pointed argument that earlier critics of polygraph screening — particularly Honts (1991, 1994) — overstated the false negative problem by using inflated base-rate estimates. Honts had relied on the finding that roughly 20% of participants in a CSP study made admissions about security violations. MacLaren countered that the vast majority of such violations are trivial, and that only about 1% of screened employees represent genuine security threats — a rate consistent with DoD operational data showing admissions in roughly 2% of cases.

In the two-strikes scenario, use of the polygraph produces tangible benefits to the efficiency and cost-effectiveness of counterintelligence efforts. By applying the counterintelligence polygraph tests in an orderly and careful way, large numbers of individuals can be effectively screened, leading to reductions in the financial expenditures related to security investigations.

The paper also highlights a dimension often neglected in purely psychometric debates: the value of admissions elicited during examination. As MacLaren notes, some agencies — such as the NSA — use the polygraph primarily as an interrogation tool to encourage disclosure, rather than as a binary truth-detection device. This broadens the utility of the instrument well beyond its raw classification accuracy rates.

Limitations & Considerations

MacLaren himself acknowledges several important constraints on his conclusions:

• No published field validity study of TES existed at the time of writing — all accuracy estimates derived from controlled laboratory simulations with "programmed guilty" participants, not confirmed real-world spies
• The base rate of espionage (assumed at 1%) is a critical and uncertain input; if the true prevalence differs substantially, the conditional probability calculations shift accordingly
• DoD annual report statistics do not constitute a proper criterion-validity study — ground truth about who among the tested population was actually guilty remains largely unknown
• The test-retest reliability of TES under repeated administrations was explicitly flagged as "an important, and as yet unanswered question" by the author himself
• The paper was published in the Polygraph journal of the American Polygraph Association, raising potential publication bias considerations

The review is also limited by its reliance on open-source material — operational details of classified counterintelligence screening protocols, actual adjudication outcomes, and full DoDPI methodological documentation were not available to independent researchers, restricting the scope of any external critique or replication.

Practical Applications

When a two-test policy is applied, additional investigative techniques can be deployed to sort innocent from guilty among double-failures — but investigators then have only 155 cases to sift through rather than 10,000. This triage function — dramatically narrowing the field for resource-intensive follow-up investigation — is perhaps the most practically compelling argument MacLaren advances for retaining polygraph screening in national security contexts.

For policymakers and security professionals, the paper underscores that polygraph effectiveness should not be judged solely on stand-alone classification accuracy. Current polygraph security screening procedures make a valuable contribution to the maintenance of national security when deployed as one layer within a multi-method security architecture — combined with financial record review, background interviews, and repeated examinations — rather than as a singular gatekeeping mechanism. Examiners and program administrators should treat any single inconclusive or deceptive result as a trigger for further investigation, not a final determination.