DoDPI Racial Bias Study: Polygraph examination results by race

The 1990 Department of Defense Polygraph Institute study examined racial disparities in polygraph examination outcomes across 1,141 training examinations. The research found that only 23.5% of innocent African Americans passed polygraph tests compared to 36.9% of innocent Caucasians, revealing significant racial bias in false positive rates. DoDPI suppressed the study and never published these findings.

Analysis of 1,141 polygraph examinations from DoDPI training classes, focusing on 401 truthful examinees (81 African American, 320 Caucasian) tested using standard DoDPI protocols and numerical scoring procedures with verified ground truth.

This suppressed 1990 DoDPI study represents one of the earliest examinations of racial bias in polygraph testing, conducted during expansion of federal screening programs. Researchers analyzed training examinations with known ground truth to compare outcomes across racial groups. The findings revealed that innocent African Americans were substantially more likely to be incorrectly classified as deceptive than innocent Caucasians. Both groups experienced extremely high false positive rates exceeding 60%, but the racial disparity was statistically significant. DoDPI attempted to suppress these concerning findings that challenged polygraph validity and fairness.

The findings raise fundamental questions about the fairness and validity of polygraph screening programs, particularly for security clearances and employment decisions. The study's suppression highlights institutional resistance to research challenging polygraph reliability and the need for transparent examination of demographic disparities in detection technologies.

Background & Context

The 1990 Department of Defense Polygraph Institute racial bias study represents one of the earliest and most significant examinations of potential demographic disparities in polygraph examination outcomes. The research revealed that only 23.5% of innocent blacks passed polygraph examinations, compared to 36.9% of innocent whites—a finding so concerning that DoDPI attempted to suppress the study and never published it.

This research emerged during a critical period when federal agencies were expanding polygraph screening programs for security clearances and employment. The study addressed fundamental questions about test fairness and whether physiological detection of deception techniques affected different racial groups equally. When the DoDPI study suggested that innocent blacks were much more likely than innocent whites to be wrongly branded as liars by the polygraph, DoDPI covered it up.

The suppression of this research occurred within a broader pattern of institutional resistance to findings that challenged polygraph validity. Department of Defense Polygraph Institute director Michael H. Capps dissolved DoDPI's Curriculum and Research Guidance Committee, which included eminently qualified professors of psychology, after the committee advised against polygraph screening.

Research Design & Methodology

The study analyzed a dataset of 1,141 polygraph examinations given in DoDPI training classes. The examinations were conducted as part of the standard training protocol for polygraph examiners at the Department of Defense Polygraph Institute, located at Fort McClellan, Alabama. These training examinations used real examinees with known ground truth—some participants were instructed to commit mock transgressions while others remained truthful.

The research focused specifically on comparing outcomes for nondeceptive (truthful) examinees across racial groups. The sample included 81 nondeceptive African American examinees and 320 nondeceptive Caucasian examinees. Examiners scored the polygraph charts using DoDPI's standard numerical evaluation system, which assessed physiological responses across respiratory, electrodermal, and cardiovascular channels.

Key methodological elements included:

• 1,141 total examinations from DoDPI training classes with verified ground truth
• Focus on 401 truthful examinees (81 African American, 320 Caucasian)
• Standard DoDPI polygraph testing protocols and numerical scoring procedures
• Analysis of false positive rates by racial group

Results & Key Findings

Only 23.5% of innocent African Americans passed the polygraph examination, compared to 36.9% of innocent Caucasians—representing a 13.4 percentage point disparity in pass rates. This means truthful African American examinees were substantially more likely to be incorrectly classified as deceptive.

False positive results were significantly more common among the 81 nondeceptive African American examinees than among the 320 nondeceptive Caucasian examinees. The statistical significance of this finding indicated the disparity could not be attributed to chance variation alone.

Key statistical findings:

• 76.5% false positive rate for truthful African American examinees
• 63.1% false positive rate for truthful Caucasian examinees
• Statistically significant racial disparity in error rates
• Both groups experienced high overall false positive rates exceeding 60%

The extraordinarily high false positive rates for both groups—with more than three-quarters of innocent African Americans and nearly two-thirds of innocent Caucasians failing—revealed fundamental problems with polygraph accuracy that extended beyond racial disparities.

Discussion & Significance

These findings carry profound implications for the fairness and validity of polygraph screening in federal employment and security clearance contexts. The racial disparity in false positive results aligned with theoretical arguments about how social and physiological factors might differentially affect polygraph outcomes across demographic groups. Factors such as heightened anxiety in testing situations, cultural differences in authority interactions, and physiological response variability could all contribute to differential error rates.

The National Research Council's 2003 comprehensive review of polygraph science acknowledged this study's importance while noting the limited research base. Only two studies were found that compared polygraph accuracy by race of examinees, and neither reported significant effects of examiner's race, examinee's race, or their interaction on polygraphic detection of deception in later analyses. However, the 1990 dataset remained the largest examination of racial disparities in polygraph outcomes.

The institutional response to these findings—suppression rather than publication—raises serious questions about scientific integrity in polygraph research. The study's concealment prevented the polygraph community from addressing documented bias and denied examinees information about differential error rates that could affect their careers and livelihoods.

Limitations & Considerations

The study examined polygraph examinations conducted during training exercises rather than actual field examinations, which may limit generalizability to operational contexts. Training examinations involve student examiners still developing their skills, potentially producing different error patterns than experienced practitioners. Additionally, the examinees were volunteers participating in training scenarios rather than actual job applicants or security clearance candidates facing real consequences.

The study did not examine potential contributing factors such as examiner race, socioeconomic variables, prior polygraph experience, or specific physiological response patterns that might explain the observed disparities. Without controlling for these confounding variables, the precise mechanisms underlying the racial differences in outcomes remain unclear. The research also did not assess whether the disparity persisted across different polygraph test formats or question types.

Practical Applications

The findings demonstrate that polygraph screening can produce systematically different error rates across racial groups, with serious implications for employment fairness and civil rights. Organizations using polygraph testing for hiring, security clearances, or personnel decisions should recognize that the technology may disproportionately exclude qualified truthful candidates from underrepresented groups. The 76.5% false positive rate for truthful African Americans means that using polygraph results as a basis for adverse employment decisions creates substantial risk of discrimination.

For individuals facing polygraph examinations, these findings underscore the inherent unreliability of the technology and the particular risks faced by African American examinees. The study provides empirical support for challenging polygraph-based employment decisions and highlights the need for alternative assessment methods that do not exhibit racial bias. Policymakers should consider whether polygraph screening can be reconciled with equal employment opportunity principles given the documented disparities in error rates.