Research: Racial disparity in lie detectors
The National Academies of Science released a comprehensive report in 2003 that examined the state of the science in lie detection. After reviewing nearly a century of scientific data and reports and visiting polygraph units run by government agencies, the NAS committee found that polygraphs were prone to both false positives and false negatives.
“Its accuracy in distinguishing actual or potential violators from innocent test takers is insufficient to justify reliance on its use.” In other words, polygraphs aren’t accurate enough to be relied on as evidence.
Wired magazine recently published an investigative article on the use of polygraphs in job screening, which is surprisingly common among law enforcement and federal agencies. The magazine concluded that not only does the polygraph inherently produce unreliable test results but it also enables racial bias.
Can polygraphs be used in criminal cases?
It’s important to note that lie detector tests are generally not admissible in court in criminal cases. However, some law enforcement agencies continue to use them during interrogations. Suspects cannot typically be compelled to take a polygraph, but police will often pressure the suspect to do so, implying that refusing the test is evidence of guilt. Yet taking the test may do nothing to dispel the police’s suspicions. When suspects pass the test, police often perceive the suspect to have “beaten the test.” Thus, criminal suspects have little to gain by agreeing to take a polygraph.
How are polygraph results tainted by bias?
Wired obtained polygraph records from police departments and lawsuits and discovered two important things. First, failure rates varied wildly from one examiner to the next. Second, African-Americans failed the test disproportionately often as compared with other test-takers.
How could that be? Wired determined that the conclusions examiners draw about the truthfulness of polygraph subjects are surprisingly subjective. The test itself is an attempt to correlate physiological markers such as increases in heart rate, blood pressure or perspiration, with lying. However, examiners are free to interpret the physiological data as they see fit. One examiner might conclude that a subject’s perspiration indicates lying, while another might conclude it merely shows the subject is nervous.
A neuroscientist told Wired that “the examiner’s decision is probably based primarily on the human interaction that the two people have.” In other words, the test could be no more than a way for examiners to justify and lend weight to their personal assessments of the subject. This allows the examiner’s biases in while giving the dangerously false impression of impartiality.
It seems that polygraphs may be no more reliable than other supposedly scientific techniques that have been debunked in the past few years, like hair, bite mark and blood spatter analysis.