Polygraph Field Validity

Background & Context

The 1985 Elaad and Schahar study represents one of the earliest and most significant attempts to validate polygraph testing under real-world field conditions. Conducted on polygraph tests from 1973-74 at the Scientific Interrogation Unit of the Israel Police, this research addressed a critical gap in polygraph validation: moving beyond laboratory experiments to assess actual criminal investigative polygraph examinations.

At the time, polygraph field validity research was scarce and methodologically challenging. The difficulty lay in establishing "ground truth"—definitive knowledge of guilt or innocence in real criminal cases. Critics later noted that ground truth information in studies like Elaad's was "not systematically collected and is inadequately described," with verification sometimes by confession. Despite these limitations, the study provided rare empirical data from an operational law enforcement polygraph unit during a period when such research was critically needed.

This Israeli field study became frequently cited in subsequent reviews and meta-analyses, though it also faced scrutiny. The study was eventually excluded from the influential 1983 OTA analysis alongside studies by Peters and Widacki, highlighting ongoing debates about what constitutes acceptable field validation evidence in polygraph research.

Research Design & Methodology

Out of 2,540 polygraph examinations conducted during 1973-74, only 184 cases (7%) had definite feedback concerning the innocence or guilt of the subject. This dramatic attrition—from thousands of tests to a verified sample of less than two hundred—illustrates the fundamental challenge of field validation: obtaining independent verification of ground truth.

The criteria for evaluation were subsequent confessions or convictions of the subjects and/or others who cleared the subjects from involvement. This methodology relied on post-polygraph developments rather than pre-existing knowledge of truth status. The study examined conclusive polygraph decisions and compared them to these criterion outcomes to calculate accuracy rates.

Key design elements included:

• Real-world criminal investigations conducted by professional Israeli Police examiners
• Verification through confessions, convictions, or exoneration by others' confessions
• Classification of results as conclusive without reservation, conclusive with reservation, or inconclusive
• Analysis of error types (false positives vs. false negatives)

Results & Key Findings

Of the 184 verified cases, 10 (5%) produced inconclusive polygraph results. The remaining 174 conclusive cases formed the basis for accuracy analysis. The headline finding: 96.6% overall accuracy when combining both categories of conclusive results.

Detailed breakdown of the 174 conclusive cases:

• 145 cases (79%) were conclusive without reservation: 143 correct (78%) and 2 incorrect (1%)
• 29 cases (16%) were conclusive with reservation: 25 correct (14%) and 4 incorrect (2%)
• Total errors: 6 cases out of 174 conclusive decisions (3.4% error rate)
• Five errors were false negatives (guilty classified as truthful) and one false positive (innocent classified as deceptive)

The error pattern suggested a tendency to clear the subject when in doubt, with false negatives outnumbering false positives 5 to 1. This bias toward exoneration may reflect examiner decision-making under uncertainty or institutional pressure to avoid wrongly accusing innocent suspects.

Discussion & Significance

The study's high accuracy rate initially appeared impressive and was cited by polygraph proponents. Reviewer Norman Ansley adjusted the accuracy rate upward to 96.6% using a methodology that divided the difference between reported accuracy and 100%, reflecting assumptions about errors favoring either examiners or criterion measures. However, this statistical approach proved controversial.

The research highlighted a critical methodological concern that would plague field validity studies for decades. The authors cautioned that validation criteria—confessions and convictions—could have been influenced by the polygraph results themselves, and that confessions and convictions are not infallible since false confessions and convictions could occur. This creates potential circularity: if polygraph results influence whether suspects confess, using those confessions to validate the polygraph inflates apparent accuracy.

The study's exclusion from rigorous meta-analyses proved significant. The OTA review characterized Elaad and similar studies as "anecdotal reports" using "samples of cases collected from police files" with inadequately described verification, sometimes by confession. This critique established that merely having field data was insufficient—systematic collection and independent criterion measures were essential for valid field research.

Limitations & Considerations

The study faced substantial methodological limitations that compromised its validity as scientific evidence:

• Selection bias: Only 7% of cases could be verified, likely representing easier or more extreme cases rather than a representative sample
• Criterion contamination: Confessions used to validate polygraph decisions may have been influenced by polygraph outcomes
• Lack of independence: No blind evaluation or separation between operational decisions and research verification
• Inadequate documentation: Insufficient detail about case selection, examiner qualifications, or polygraph techniques used

Validity studies require "adequately described and systematically determined criterion of truth," and reports with "unclear or unsystematic" verification criteria fail to meet scientific standards. The Elaad and Schahar study's reliance on post-hoc, potentially contaminated criteria exemplified these problems.

Practical Applications

Despite its limitations, this study provided important practical insights for polygraph examiners and policymakers. The finding of asymmetric errors—overwhelmingly favoring false negatives over false positives—suggested that examiners in operational settings may be appropriately conservative, preferring to let guilty parties pass rather than falsely accuse the innocent. This operational philosophy, while protecting civil liberties, means guilty individuals may evade detection.

The research also demonstrated the severe practical challenges of conducting field validation studies. The 93% loss of cases due to inability to verify ground truth showed why laboratory studies with known truth status remain methodologically superior for validity research, even if they sacrifice ecological validity. For agencies considering polygraph implementation, this study underscored that operational polygraph work rarely produces the kind of systematic feedback necessary to evaluate or improve accuracy in real-world practice.