Return to Gateway to Intelligence page
Mr. Chairman, members of the Judiciary Committee, I want to thank you for having me here to testify on this matter of great importance to our country’s security.
I wish to begin by making clear that there is no distinctive lie response. Polygraph operators try to determine if someone is lying by comparing responses to Relevant vs. “control” or comparison questions. Relevant questions deal with issues related to being a traitor to your country while the control questions deal with possible misdeeds from your past. For example, the physiological response to the relevant question “Have you committed espionage?” is compared to the response elicited by the control question “Did you ever violate a traffic law?” Both questions are answered “no.” These tests are based on the idea that spies will be more physiologically aroused by the espionage question. But scientific research suggests that sophisticated criminals or spies can pass the test by artificially augmenting their response to the control question. When asked whether he violated a traffic law, a spy need only curl his toes inside his shoe, lightly bite his tongue, or commence mental arithmetic exercises to pass this test. Polygraph operators have no way of detecting the use of these countermeasures, and detailed instructions regarding how to employ them can be found in libraries and on the internet. Someone who is clever enough to be a spy should be clever enough to learn these simple techniques to beat a polygraph.
For an innocent person to pass, he must be more worried about answering a question about a traffic violation than about espionage. However, it is obvious to everyone which of these two questions is more important. Even though innocent, being asked about espionage is likely to be upsetting because your patriotism is being challenged and because your response to this question determines your future employment. The consequences of being physiologically aroused to a question about espionage are grave even for innocent people; that is why they would be expected to fail polygraph tests in large numbers when the tests are scored in the standard way. In fact, the best studies of polygraph tests, using real-life cases and published in top scientific journals, find that innocent people fare little better than chance on these tests, with 40% or more failing on average.
This explains why large numbers of FBI applicants fail pre-employment polygraph exams even after they have been judged to satisfy most of the FBI qualifications to be an agent. Surprisingly, however, when polygraphs are given to those already employed by the government, almost all of whom can be presumed to be innocent, very few individuals fail. This outcome is obtained because examiners understand that failing more than a handful of those with security clearances would be embarrassing to themselves and have potentially catastrophic consequences for government programs that depend on having a stable, expertly trained work force. In other words, post-employment polygraphs are not scored following standard procedures; examiners make subjective judgments that find few workers deceptive. In the absence of a scientific basis for their program, the government has turned to other arguments to justify polygraphs, claiming they have utility because they generate important admissions from examinees. This argument has not been supported by data. Sworn statements of significant wrongdoing, firings, arrests, convictions, and lists of spies uncovered as a result of polygraph tests have not been forthcoming. The government also argues polygraphs have a deterrent effect, a claim with no empirical support that is certainly unlikely to be true as employees learn virtually no one fails polygraphs.
At their invitation, I recently met with scientists at Los Alamos National Laboratory. I found them very concerned about protecting nuclear secrets. However, these scientists do not believe polygraph testing will accomplish this objective. They are worried because valued senior staff are retiring early and talented young prospects are turning away from lab employment rather than subject their careers and integrity to the polygraph equivalent of Russian roulette.
To conclude, polygraphs are unlikely to catch spies and are likely to have deleterious effects on the recruitment and retention of the best employees. To the extent that their value derives from admissions made during testing, these admissions will only be forthcoming if examinees believe the tests work. The success of the government program thus depends on examinees being ignorant of the procedure, an unsafe assumption, especially when those tested are the kinds of smart people we want as intelligence officers and weapons scientists. Even if polygraph testing were as accurate as the government claims, long-term harm to national security may outweigh any benefits. Thank you.
Polygraph Screening of Federal Employees and Job Applicants
Addendum to Oral testimony of William G. Iacono1
Submitted to the U.S. Senate Judiciary Committee Hearing
“Issues Surrounding the Use of Polygraphs”
25 April 2001
National Security Screening
In view of the federal Employee Polygraph Protection Act of 1988 (29 US Code, Chapter 22), which prohibits requiring employees or job applicants in the private sector to submit to polygraph testing, it is ironic that the federal government is the principal employer of polygraph examiners. Applicants for positions with the FBI, CIA, NSA, Secret Service, and similar agencies are required to undergo lie detector tests intended to supplement or substitute for background investigations. Current employees of some of these agencies, military personnel who hold high security clearances, and civil employees of defense contractors doing classified work may be required to undergo periodic tests for screening purposes. The Department of Defense conducted some 17,970 such tests in 1993.2 Most of these tests are referred to as counterintelligence scope polygraph tests by the government.
As a consequence of Public Law 106-65 (S. 1059) passed as part of the National Defense Authorization Act of 2000, potentially thousands of scientists and security personnel employed at U.S. weapons labs at Lawrence Livermore, Sandia, or Los Alamos must submit to polygraph tests as part of an effort to improve nuclear security. A relatively new procedure, the Test for Espionage and Sabotage (TES), or a nearly identical variant of this procedure, the Test for Espionage, Sabotage, and Terrorism (TEST), is used.
As outlined in the recently promulgated Department of Energy (DOE) Rule 7093 these counterintelligence polygraph examinations are to be limited to coverage of six topics:
4) intentional unauthorized disclosure of classified information,
5) intentional unauthorized foreign contacts, and
6) deliberate damage or malicious use of a U.S. government or defense system.
Rule 709 has a number of interesting features that are similar to those governing the use of polygraph tests by other federal agencies and that are likely to stimulate law suits.4 These include the following:
Prospective employees of the DOE or its contractors who refuse to take a polygraph cannot be hired and incumbent employees must be denied access to secret information.
Using the results of a polygraph test as an “investigative lead” can result in an administrative decision that denies or revokes an employee’s access to classified information and may lead DOE to “reassign the individual or realign the individual’s duties within the local commuting area or take other actions consistent with the denial of access.”
These tests will be conducted at least every five years and also on an aperiodic basis.
Public comment on the proposed regulations revealed widespread opinion that “that polygraph examinations have no theoretical foundation or validity.” DOE decided, however, that “as a matter of law,” the agency is mandated to conduct polygraph examinations, and “is no longer free to act favorably on comments arguing against establishment of a counterintelligence scope polygraph examination program because of information and claims about deficiencies in polygraph reliability.”
The TES5 includes four irrelevant questions (e.g., “Do you sometimes drink water?” “Is today ____?”) and the following four relevant questions: “Have you committed espionage?” “Have you given classified information to any unauthorized person?” “Have you failed to notify, as required, any contact with citizens of sensitive countries including China?” “Have you been involved in sabotage?” The responses to the relevant questions are compared to the responses to four “directed lie” questions that serve as “controls” or comparisons by providing an example of a response to a known lie. The directed lies are questions that both the examiner and the examinee know will be answered falsely. These four questions are chosen from a list of acceptable alternatives, but may include any of the following, which the examinee is directed to answer “No”: "Did you ever violate a traffic law?" “Did you ever say something that you later regretted?” “Did you ever lie to a co-worker about anything at all?” Examinees who show greater autonomic disturbance following the questions about espionage and sabotage, than they show following these directed lies, are classified as deceptive.
The field validity of counterintelligence scope polygraph examinations, including the TES, is unknown. However, the Department of Defense Polygraph Institute (DoDPI) has reported two laboratory studies of the validity of the TES.6 These both employed paid volunteers, 115 of whom were innocent while 60 others were each required to enact simulated acts of espionage or sabotage. Of the innocent subjects, 14 or 12.5% responded in the deceptive direction. Of the "guilty" subjects, 10 or 17% were misclassified as innocent.
It is obviously likely that innocent scientists or other persons with high security clearances would be more disturbed by the TES relevant questions asked during an official screening test than were these volunteers for whom the test carried no threat to their reputations or careers.7 The disturbance produced by the directed-lie questions, on the other hand, might be expected to be no greater in real-life than in simulated conditions of testing. Therefore, when innocent, loyal government employees with top secret classifications are subjected to the TES, one might expect many more to be classified as deceptive than the 12.5% suggested by the DoDPI studies. The actual rate of false-positive diagnoses is probably close to the 43% level indicated by the real-life studies referred to in footnote 7.
When DOE scientists are subjected to the planned TES (or TEST), these data indicate that large numbers of innocent employees would be classified as deceptive if the test scores were relied upon. DOE's polygraph examiners avoid any such disastrous result because they know that the base rate of spying (the proportion likely to be spies) among such a highly screened and dedicated group is likely to be tiny. Consequently, they cannot fail 43% or even 12.5% of scientists without undermining their own credibility, creating a personnel management nightmare, and wreaking havoc on employee morale.
Therefore, subjects who are more troubled by “Have you committed espionage?” than by “Did you ever say something that you later regretted?” are invited by the examiner to explain why they might have responded in this way. If the respondent’s answer and demeanor satisfy the examiner, his “fail” is converted to a “pass.” Thus, by permitting the polygraph operator to be the ultimate arbiter, relying on whatever clinical skills or intuitions s/he may (or may not) possess, the frequency of false-positive diagnoses is kept to a low value. Nevertheless, if as few as 2% of the 10,000 workers potentially covered by Rule 709 receive final diagnoses of “deception indicated,” 200 highly trained but probably innocent scientists would be implicated as spies in the first round of testing.8
Although the controversy surrounding the DOE polygraph screening program has been focused on the high likelihood that innocent individuals will be judged to be spies, there is little evidence that the program will actually catch spies. The laboratory studies of the TES, which reported only 83% accuracy in identifying persons “guilty” of committing mock-espionage, overestimate accuracy for the real-life guilty in two important ways.
First, consistent with real life screening test practices that help to keep the number failing these tests low, these studies did not conclude that deceptive polygraph tests were in fact failed if, during a post-test interview, an examinee offered information that reasonably justified why the test might be a false positive outcome. However, the design of the studies allowed only innocent test subjects this opportunity to “talk their way out of” a failed test because guilty people were instructed to confess as soon as the examiner confronted them with their deceptive test results. We do not know how many guilty individuals would have been mistakenly judged “false positives” had they been allowed to try to “explain away” the outcome of their examinations.
Second, these DoDPI studies did not account for the likelihood that real spies would use countermeasures to defeat the TES.9 DOE scientists are not simpletons: if one or two are in fact spies, surely both they and their foreign handlers would have sense enough to be prepared to bite their tongues after each directed-lie question. Thus it is to be expected that the only weapons-lab scientists, with their highly specialized skills, who fail the projected DOE polygraph screens, will be truthful, honorable people who cannot offer a plausible excuse for failing their polygraphs. The most likely result of Rule 709 will be their ruined reputations and the government’s loss of skilled, dedicated employees.
Besides the facts that these tests are not justified on scientific grounds and that they are clearly biased against truthful employees, there is no evidence that personnel screening tests have any true utility.10 No spy has ever been uncovered because of a failed polygraph test. Although the government has argued that the admissions individuals make when undergoing these tests provide valuable information, there is no evidence documenting that vital or even important information has been uncovered as a result of polygraph tests. It is possible that employee screening has a deterrent effect in that knowledge that one must pass such tests may discourage would be spies from seeking employment, and it may discourage the currently employed from entertaining thoughts about becoming a spy. However, there is no evidence to support such an assertion. Given the ease with which individuals can learn to defeat these tests coupled with the fact that almost no one is judged to have failed them,11 it is unlikely that they have any serious deterrent effect.
Opinions of DOE National Laboratory Senior Scientists Regarding Employee Screening
Concerned that national laboratory employees must submit to periodic lie detector tests, a panel of the more senior national laboratory scientists and engineers undertook a detailed appraisal of the existing literature relating to the nature and validity of polygraph screening methods. Sandia’s Senior Scientists and Engineers (“Seniors”) provide a service to the Laboratories as independent, experienced, corporate evaluators of technical issues. They are available as a group to assist Sandia management with technical reviews of particularly significant issues and programs. Implementation uses subpanels of the Seniors (helped as necessary by other Sandia staff) to conduct the initial, detailed review of issues or programs. The reports of the subpanels are then made available for review by all other Seniors prior to submission to management. The report of the subpanel studying polygraphs and security at Sandia was circulated in the fall of 1999.12
These Seniors, whose expertise is in physics, chemistry, and/or mathematics, do not pretend to be psychologists, psychophysiologists, or psychometricians. But they do know how to read research reports and to evaluate statistical evidence and probabilities. In their Executive Summary, they concluded that
There were no adequate studies to support polygraph screening
It is impossible to predict what error rates to expect
Polygraph testing could drive away existing innocent, talented workers who have provided value to national security programs, and it would deter prospective, talented employment candidates from considering a career in the national laboratories
Because few spies are likely to be detected, real subversives may be more likely to become insiders—particularly if over-reliance on polygraph testing leads to reduced emphasis on other security and counterintelligence methods.
Personnel screening cannot be scientifically justified. If the polygraph charts obtained from security screening tests were scored objectively, large numbers of innocent employees would be expected to fail. They do not because the tests are scored subjectively, with few failing. Claims that these tests have deterrent value are not supported and are unlikely to be true as government employees learn that virtually no one fails these tests (see footnote 8). Claims that they have utility due to admissions made during testing are not supported by empirical evidence (again, see footnote 8). There is no evidence they catch spies, and it is likely that spies can learn to use countermeasures to defeat them. When bright, talented government workers and employee prospects come face to face with the requirement that now or in the future they will have to pass repeatedly a test that is the equivalent of playing Russian roulette with their careers, they are likely to opt for other careers. Over the long term, employee morale is likely to suffer, as will the nation’s national defense as the best and brightest employees are lost to government employment.
1 Distinguished McKnight University Professor, Professor of Psychology, Law, and Neuroscience, University of Minnesota, Director, Clinical Science and Psychopathology Research Training Program, recipient of the American Psychological Association's Distinguished Scientific Award for an Early Career Contribution to Psychology, the Society for Psychophysiological Research's Distinguished Scientific Award for an Early Career Contribution to Psychophysiology, Past-President of the Society for Psychophysiological Research (1996–97) and former Member, Department of Defense Polygraph Institute's Curriculum and Research Guidance Committee. This addendum was adapted from W.G. Iacono and D.T. Lykken, “The scientific status of research on polygraph techniques: The case against polygraph tests” in D. Faigman et al. (Eds.), Modern Scientific Evidence: The Law and Science of Expert Testimony (second edition; in press). St. Paul, MN: West Publishing.
2 Department of Defense Polygraph Institute, A Comparison of Psychophysiological Detection of Deception Accuracy Rates Obtained Using the Counterintelligence Scope Polygraph and the Test for Espionage and Sabotage question Formats. 26 Polygraph, 79-106 (1997), at 80. (hereafter DoDPI Study 1.)
3 Part 709 “Polygraph Examination Regulations”’ in Chapter III of Title 10 of the Code of Federal Regulations.
4 In anticipation of the DOE regulations, attorneys representing government employees and employee prospects have indicated a desire to sue the government based on adverse employee decisions made as a result of polygraph examinations.
5 Because the government has published information only on the TES, we will refer to this procedure in the remainder of this section.
6 DoDPI Study 1, supra note 33; and Department of Defense Polygraph Institute, Psychophysiological Detection of Deception Accuracy Rates Obtained using the Test for Espionage and Sabotage. 27 Polygraph, 68-73 (1998). Because inconclusive polygraph tests are typically repeated until they yield a conclusive verdict, inconclusive outcomes are not included in the calculation of accuracy rates in this study.
7 Field studies based on actual espionage cases are needed to determine how the TES works in real life when innocent persons reputations and careers are on the line. Such studies do not exist. However, there are studies of the accuracy of real life criminal polygraph tests that have been published in top journals such as Nature and the Journal of Applied Psychology. These journals routinely reject over 85% of submitted articles. The studies published in these journals indicate that approximately 43% of innocent subjects fail the polygraph when the decision is based on physiological responses to relevant and control questions. (for more details, see W.G. Iacono and D.T. Lykken, The Case Against Polygraph Tests in D. Faigman et al., Modern Scientific Evidence: The Law and Science of Expert Testimony. St. Paul, MN: West Publishing, 1997).
8 The Department of Defense Polygraph Program report to Congress for Fiscal year 2000 illustrates how polygraphers adjust the outcomes of their tests to minimize failing anyone [Department of defense Polygraph Program Annual report to Congress, Fiscal Year 2000, Office of the Assistant Secretary of Defense (2000); available at http://www.fas.org/sgp/othergov/polygraph/dod-2000.html]. For fiscal year 2000, 7,688 individuals were given counterintelligence scope polygraph tests but demonstrated “no significant physiological response to the relevant questions and provided no substantive information.” In other words, some undetermined number provided a substantial physiological response but passed because they did not make incriminating revelations. An additional 202 individuals produced significant physiological reactions and provided “substantive information.” Of these, 194 received “favorable adjudication” with the remaining 8 cases still pending decisions, with no one receiving “adverse action denying or withholding access” to classified information. These data confirm that the government goes to extreme lengths to ensure no one fails these tests, but the also demonstrate that the tests have no utility.
9 Honts et al. have shown that 50% or more of guilty subjects in laboratory studies can defeat a polygraph test by engaging in countermeasures such as lightly biting the tongue, curling toes inside one’s shoes, or doing mental arithmetic when control questions are asked. Skilled examiners could not determine when countermeasures were being used. Charles R. Honts et al., Effects of Physical Countermeasures on the Physiological Detection of Deception, 70 J. Applied Psychol. 177, 177–187 (1985); Charles R. Honts et al., Mental and Physical Countermeasures Reduce the Accuracy of Polygraph Tests, 79 J. Applied Psychol. 252, 252–259 (1994).
10 In the Clinton Administration’s Joint Security Commission Report ["Redefining Security," A Report to the Secretary of Defense and the Director of Central Intelligence, February 28, 1994, Joint Security Commission, Washington, D.C. 20505 ; available at http://www.fas.org/sgp/library/jsc/../index.cfm], it is noted that “the most important product of the polygraph process is more likely to be an admission made during the interview than a chart interpretation...While senior officials at the CIA and the NSA acknowledge the controversial nature of the polygraph process, they also strongly endorse it as the most effective information gathering technique available in their personnel security systems.”
11 See footnote 8 summarizing the DoD Fiscal year 2000 report.
12 Polygraphs and Security, A study by a Subpanel of Sandia’s Senior Scientists and Engineers, October 21, 1999, Sandia, NM; available at http://www.fas.org/sgp/othergov/polygraph/sandia.html.