FAS | Secrecy | Polygraph || Index | Search | Join FAS

Polygraphs and Security

A Study by a Subpanel of Sandia’s Senior
Scientists and Engineers

October 21, 1999

Executive Summary

Because of concerns raised in Congress and the Executive Branch about inadequate security in the national nuclear weapons laboratories, the Department of Energy (DOE) plans to institute polygraph screening for some employees and applicants. These tests are intended to identify subversives and deter potential ones. This policy seemingly assumes that polygraph tests, test interpretation, and any follow-up processes will accurately identify subversives and nonsubversives. We conclude that there is no adequate scientific basis for this assumption. No specific polygraphic or behavioral response has been directly linked to the act of deception and there are too many subjective factors involved in the administration and interpretation of polygraph tests to be able to predict and control their effectiveness and limitations.

A review of the scientific literature on polygraph testing revealed substantial concern about polygraph accuracy for screening, and Federal law for most situations bars such usage. A summary of scientific opinion from a recent survey concludes that most psychology experts do not consider polygraphy to be technically sound and even more believe that skilled subversives can defeat polygraph tests.

Two general uses of polygraph testing are specific-incident investigations (as when an individual has been accused of a crime) and general screening (where a target population is tested to see if any of them have committed any crime). Published estimates of polygraph accuracy for specific-incident situations, based on the agreement of polygraph results with known facts, vary depending on the context in which data were obtained and the quality of data collection, selection, and analysis. A 90% accuracy rate is a reasonable expectation for adequately controlled specific-incident tests. It is, however, unwarranted to assume these accuracy rates apply to screening applications of polygraphy. Adequate studies have not been done for screening applications. Thus, it is impossible to predict what error rates (false negative—subversive passes polygraph test; false positive—innocent person fails polygraph test) and inconclusive results would occur in the proposed DOE screening. But, the costs and consequences of such errors need to be considered before the DOE policy is implemented. False positive results subject individuals to increased scrutiny and unwarranted suspicion. Even if a suspect is eventually exonerated, the process can damage that person’s career and job performance. Such possibilities can make it more difficult to recruit and keep personnel with the high professional qualities on which the nuclear weapons program relies.

Issues resulting from false positive results have influenced agencies to "tune" polygraph tests (reduce the number of positive indications for screening). In fact the DOE has stated that a 2% positive indication is anticipated. Tuning polygraph tests to decrease positive results increases the probability of false negative results, thus reducing the intended effectiveness of the tool. Consequently, 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.

Polygraph testing could drive away existing innocent, talented workers who have provided value to national security programs and deter prospective, talented employment candidates from considering a career in the national laboratories. Resources that could have been applied directly to national security programs or to finding more effective ways to enhance security may be wasted in administering a polygraph screening program and dealing with the consequences of false identifications.

We believe that the entire national laboratory security system should be improved using a systems approach in which the cost and benefits of changes can be plausibly estimated. A full systems evaluation is necessary because computer technology has fundamentally changed threats to national security. We doubt that polygraph screening of employees will provide value to an integrated security system.

Table of Contents

Preface
Charter
Acknowledgments
1.0 Introduction
2.0 Polygraphy
2.1 Theory 
2.2 Applications
2.3 Accuracy
2.4 Countermeasures
2.5 False Results
2.6 Examiner Influence
3.0 DOE Implementation
3.1 Improvements Needed
4.0 National Security Concerns
5.0 Alternative Measures
6.0 Conclusions
Appendix I: Acronyms
Appendix II: References

Preface

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.

This document is the report of the subpanel studying polygraphs and security at Sandia. Members of the subpanel are:

• Bob Benner, 9224
• Larry Bertholf, 4103
• Earl Boebert, 5901
• Dick Damerow, 2567
• Rob Easterling, 9800
• Lawrence Larsen, 15300
• Carl Melius, 8130
• Dana Powers, 6400
• Al Zelicoff, 5335

Charter

"I believe that the question of polygraphs is a central one that will occupy more and more of our time before it’s over. … The crux of the issue is that, while few if anyone really advocates the use of more extensive polygraphs as a screening tool (because of the false positive problems), a large body of opinion suggests that polygraphs are a useful investigatory tool. I confess to not knowing where we as a management team should stand on this issue … I would appreciate your thoughts and inputs as to where we would like this issue to come out … ." C. Paul Robinson

Acknowledgments

Dianna Blair (9811) made major contributions to this report. Parts of the background, approach, text, references, and philosophy came from her. The Seniors on this subpanel are indebted to her and gratefully acknowledge her contributions. Jerry Allen (4100) encouraged us to consider alternatives to improve security. Dan Garber (4141) provided several editorial improvements. Julie Kesti (4915) provided literature searches and references. Larry Greher (11200) provided references to court cases, Bob Park (11300) helped us interpret the Employee Polygraph Protection Act, and Paul Shoemaker (5002) provided good "leads" and direction.

1.0 Introduction

Because of concerns raised in Congress and the Executive Branch about inadequate security at the national weapons laboratories, the Department of Energy (DOE)¹ plans to institute polygraph screening for employees who have access to the most sensitive categories of classified information and materials, as well as applicants for such positions. These tests are intended to identify actual subversives and deter potential ones. This policy seemingly assumes that polygraph tests, test interpretation, and any follow-up processes will accurately identify subversives and nonsubversives.

The best summary of polygraphy that we found is the Office of Technology Assessment (OTA) report. The OTA concluded that "while there is some evidence for the validity of polygraph testing as an adjunct to criminal investigations, there is very little research or scientific evidence to establish polygraph test validity in screening situations."^{2, 3}

Although the accuracy of polygraph screening is very questionable, congressional legislation is mandating such screening by DOE: The proposed legislation requires a polygraph examination for all persons in "… positions with access to the most sensitive categories of classified information and materials, as well as applicants for such positions."⁴

Senator Domenici has made it clear that he is concerned about mandatory polygraph testing:

US Senator Pete Domenici today urged DOE Secretary Bill Richardson to carefully consider the implementation of mandatory polygraph tests for agency employees, contending that ‘a polygraph cannot be the sole determinant of the fitness for duty of national security workers.’

‘Loyal workers threatened by false positives must have rapid and sure recourse before their careers and work are ruined and critical national security programs are impacted through incorrect loss of key researchers,’ Domenici wrote Richardson. ‘Large numbers of such false positives may overload any system you devise to handle them. Complete plans to address this issue should be in place before large numbers of tests begin.’ ⁵

Senator Bingaman has expressed clear opposition to DOE’s plans for implementing polygraphs:

I am writing to express my opposition to plans by the Department of Energy (DOE) for implementing counterintelligence polygraphs, as proposed in the Federal Register in August 18, 1999. This rule goes far beyond what I envision as being an appropriate use of polygraphs, which would be as a limited investigative tool in cases where other evidence suggests the possibility of espionage. My opposition is based on five factors.

1. The proposed rule’s basic premise, that screening polygraphs offer a specially effective tool for detecting guilty individuals, is not supported by scientific evidence.

2. The provisions of the proposed rule are unacceptably vague on key issues, such as who would be subject to requirements of the rule, and overboard in the potential categories of individuals who might be affected.

3. The proposed rule, in my view, does not give sufficient consideration to the privacy and other legal issues that will result from DOE’s proposed program.

4. The proposed rule takes what I believe to be an unrealistic view of the problem of false positives. I am concerned that persons who are judged to have "failed" a polygraph screening will not be easily cleared, as this would involve proving a negative. The latter will, in my opinion, be particularly difficult to do, judging from the partisan atmosphere in which DOE security issues have been treated over the last year.

5. As a result of the proceeding four factors, I believe that the proposed counterintelligence polygraph program will make it much more difficult for the DOE laboratories to attract and retain the best and brightest scientific and technical talent.⁶

This report addresses an essential question: In a full systems context—as one of many security and counterintelligence tools—will polygraph testing add to or subtract value from the quality and security of the nuclear weapons program?

A very recent study by Eli Lehrer points out that basic polygraph technology has not changed in the last 60 to 70 years:

Skeptics and polygraph professionals agree that the fundamental technology, which measures breathing, pulse, blood pressure and galvanic skin response (sweating) has remained unchanged since … the 1930s. American Polygraph Association President Richard Keifer says that computers have simplified the work but agrees that the measurements have not changed.⁷

Polygraphs are used in conjunction with many test protocols—such as the Control Question Test (CQT), Guilt Knowledge Test, Relevant/Irrelevant Technique, and Peak of Tension Test. Polygraphs are used by experienced and new examiners in direct and in "blind" tests.⁹ They are also used when facts are known and tests are controlled and in cases when the "facts" are determined from confessions, evidence, and judicial decisions.

In specific-incident applications with controlled conditions, polygraphy can be useful. The following psychology laboratory experiment is an example of conditions where reasonable accuracy may be achieved:

Prototypically, the experiment is a card test with one of 6 simple geometries on each card. The subject is shown one card, which one is unknown to the examiner. … The polygraph examiner then shows each card to the subject and asks if this is the seen card. The subject replies ‘no’ after each trial. After repeated trials, the differential polygraphic response to the guilty knowledge (the seen card) can be detected about 90% of the time based on simple Autonomic Nervous System (ANS) reactions to the ‘lie.’ ¹⁰

General screening applications of polygraphy are a totally different matter. The examinee is not naïve; the screening accuracies are much lower; and there is much more at stake than a card experiment (such as national security, clearances, jobs, and jail). Furthermore, measures to counter the effects monitored by the polygraph have been found and the use of countermeasures by a guilty party upsets the conditional probabilities of accurate detection and identification.

For both specific-incident and screening applications, many external variables can influence test results, including countermeasures, test protocol, test calibration, and the personalities, biases, and tactics of the interrogator and the subject. A summary of scientific opinion from a recent survey concludes that most psychology experts do not consider polygraphy to be technically sound and even more believe that skilled subversives can defeat polygraph tests.¹¹

A summary of more than 2000 specific-incident cases in the 1980s shows an accuracy of 98% for cases where the examiner was directly (or interactively) involved in the decisions. In more than 900 specific-incident cases during the same time period, the accuracy was 90% for evaluators performing blind tests.¹²

In 1983, the OTA provided the following summary of results for research on the CQT in specific-incident criminal investigations:

• Six previous reviews of field studies: average accuracy ranged from 64 to 98 percent.

• Ten individual field studies: correct guilty detections ranged from 70.6 to 98.6 percent and averaged 86.3 percent; correct innocent detections ranged from 2.5 to 94.1 percent and averaged 76 percent; false positive rate (innocent persons found deceptive) ranged from 0 to 75 percent and averaged 19.1 percent; and false negative rate (guilty persons found nondeceptive) ranged from 0 to 29.4 percent and averaged 10.2 percent.

• Fourteen individual analog studies: correct guilty detections ranged from 35.4 to 100 percent and averaged 63.7 percent; correct innocent detections ranged from 32 to 91 percent and averaged 57.9 percent; false positives ranged from 2 to 50.7 percent and averaged 14.1 percent; and false negatives ranged from 0 to 28.7 percent and averaged 10.4 percent.¹³

The OTA report also comments on polygraph accuracy:

A major reason why scientific debate over polygraph validity yields conflicting conclusions is that the validity of such a complex procedure is very difficult to assess and may vary widely from one application to another. The accuracy obtained in one situation or research study may not generalize to different situations or to different types of persons being tested.¹⁵

A great deal of information highlights the gulf between polygraph accuracies for specific-incident cases and for screening. Illustrative information is summarized below.

The OTA report expresses reservations about use of the polygraph for screening:

… while there is some evidence for the validity of polygraph testing as an adjunct to criminal investigations, there is very little research or scientific evidence to establish polygraph test validity in screening situations, whether they be preemployment, preclearance, periodic or aperiodic, random, or dragnet [emphasis added].¹⁶

Concerned by the lack of evidence for the validity of these procedures, the Subcommittee [US House Select Committee on Intelligence in 1979] urged the director of the Central Intelligence Agency (CIA) to institute research on the ‘accuracy of the polygraph in the pre-employment setting and to establish some level of confidence in the use of that technique.’ No credible research on the important topic, however, has as yet been published. … No one knows whether the screening test has some, slight, or no validity at all.¹⁷

Clearly the use of polygraph testing for screening is problematic. We return to the OTA report for a concluding statement about polygraph accuracy in general:

No overall measure or single, simple judgment of polygraph testing validity can be established based on available scientific evidence [emphasis added].¹⁹ There are two major reasons why an overall measure of validity is not possible. First, the polygraph test is, in reality, a very complex process that is much more than the instrument. Although the instrument is essentially the same for all applications, the types of individuals tested, training of the examiner, purpose of the test, and types of questions asked, among other factors, can differ substantially. ... For example, there are differences between the testing procedures used in criminal investigations and those used in personnel security screening. Second, the research on polygraph validity varies widely in terms of not only results, but also in the quality of research design and methodology. Thus, conclusions about scientific validity can be made only in the context of specific applications and even then must be tempered by the limitations of available research evidence.²⁰

Theoretically, polygraph testing—whether for personnel security screening or specific-incident investigations—is open to a large number of countermeasures, including physical movement or pressure, drugs, hypnosis, biofeedback, and prior experience in passing an exam [emphasis added]. The research on countermeasures has been limited and the results—while conflicting—suggest that validity may be affected. OTA concluded that this is particularly significant to the extent that the polygraph is used and relied on for national security purposes, since even a small false negative rate (guilty person tested as nondeceptive) could have very serious consequences [emphasis added].²¹

If polygraph testing is to be more widely employed in national security investigations, there is an urgent need for research on countermeasures. Particular priorities would be research on drugs, biofeedback training, and subject gullibility, and motivation. Such research needs to be carried out both in field situations and in the laboratory. There are a number of drugs that are suspected of lowering ANS arousal and that theoretically may be able to invalidate the results of a polygraph examination or compel an ‘inconclusive’ finding. A first priority is to extend … research on meprobamate (which reduced detectability) to other psychoactive drugs. Biofeedback training, as well as other forms of training have not been investigated, yet their effects on polygraph examinations may be substantial. Subjects’ beliefs about the accuracy of the polygraph may also be critical. As suggested by the research … individuals who believe their underlying thoughts are detectable are more likely to provide truthful responses. The reverse phenomenon seems feasible and it would seem possible to train individuals to believe that the polygraph is ineffective. Such training might be accomplished by providing individuals with false feedback on the polygraph as well as by specific instructions during simulated polygraph examinations. Similarly, subjects who can be easily trained to beat the polygraph may be more desirable as intelligence agents [emphasis added].²²

Effects of countermeasures on the CQT polygraph test were examined in an experiment with 120 subjects recruited from the general community. Subjects were given polygraph tests by an examiner who used field techniques. Twenty subjects were innocent, and of the 100 guilty subjects, 80 were trained in the use of either a physical countermeasure (biting the tongue or pressing the toes to the floor) or a mental countermeasure (counting backward by 7) to be applied while control questions were being presented during their examinations. The mental and physical countermeasures were equally effective: Each enabled approximately 50% of the subjects to defeat the polygraph test. … Moreover, the countermeasures were difficult to detect either instrumentally or through observation.²³

A 1999 article by Robert Park presents a similar opinion from a Federal Bureau of Investigation (FBI) expert:

‘There is almost universal agreement that polygraph screening is completely invalid,’ Federal Bureau of Investigation polygraph expert Dr. Drew Richardson asserts. (Richardson taught his 10-year-old son to beat the test.) In 1997 Senate testimony, Richardson warned, ‘To the extent that we place any confidence in the results of polygraph screening, and as a consequence shortchange traditional security vetting techniques, I think our national security is severely jeopardized.’ ²⁵

False negatives. False negative results (subversives who "pass" the polygraph test) pose an obvious increased threat to national security. This major issue seems to have been overlooked by the public, their elected representatives, and the rest of the bureaucracy.

James Matte comments on false negatives:

Perhaps the greatest danger is that a clever and convincing psychopath can talk a polygraph examiner into believing him even though the polygraph charts indicate deception.²⁷

The problem … was that the examiners in each case had failed to establish the proper psychological atmosphere of fear and intimidation. Unless the subject is afraid of detection the experts said, the needle won’t jump. The tests … were invalid because the examiners were too friendly.²⁹

False positives. In 1983, the OTA concluded "that the mathematical chance of incorrect identification of innocent persons as deceptive (false positives) is highest when the polygraph is used for screening purposes [emphasis added]. The reason is that, in screening situations, there is usually only a very small percentage of the group being screened that might be guilty."³⁰

The fact that false positives are widely known to be a problem is illustrated by part of the proposed legislation: "The Secretary shall prescribe any regulations necessary to carry out this section. Such regulations shall include procedures, to be developed in consultation with the Director of the Federal Bureau of Investigation, for identifying and addressing ‘false positive’ results of polygraph examinations."³¹

The following decision tree illustrates the problems with false results. It starts with 5000 employees being tested where 1% (50 persons) are assumed to be subversives (S) and the remaining 99% (4950 persons) are assumed to be not subversive (S*). This tree shows that even when a very generous accuracy of 90% is assumed for this screening application, 91.7% of those charged as guilty by the "lie detector" are, in fact, innocent. This represents a bias against the innocent of more than 10 to 1.

Because it is impossible to prove a negative, using such techniques to determine an employee’s suitability puts the employee at a great disadvantage. Raising doubts about a person’s loyalty or security performance can adversely impact that person’s career. With reported polygraph screening accuracy rates, 10% to 50% of national laboratory employees interrogated might be labeled security risks. Furthermore, relying heavily upon such a technique would result in a false sense of security. As discussed in Section 2.4, subversives can learn countermeasures to evade detection.

The accuracy of polygraph tests for screening is poor even with examiners who were probably unbiased. The large-scale implementation of polygraph screening at the weapons laboratories will require hiring many more examiners. Yet to prove that examiners are fair, DOE will have to construct tests to winnow the list of examiners (including current ones). Given that people who discriminate tend to believe in their actions, what kind of tests should be used? The ability to come up with a list of qualified examiners who can also create impartial fear and intimidation is a daunting task.

Examples. Examiners may influence polygraph tests in a number of ways. According to Norman Ansley, the difference between direct and blind polygraph tests can affect accuracy (See Section 2.3):

A summary of more than 2000 specific-incident cases in the 1980s shows an accuracy of 98% for direct examiner decisions. In more than 900 specific-incident cases during the same time period, the accuracy was 90% for evaluators performing blind tests.³²

When experts were asked if they would they submit to a ‘friendly’ polygraph (e.g., one administered by their lawyer), if they were guilty of a crime, 73% responded in the affirmative. However, only 35% would agree to take an ‘adversarial’ polygraph (say one administered by a prosecutor), if they were innocent.³³

Perhaps the greatest danger is that a clever and convincing psychopath can talk a polygraph examiner into believing him even though the polygraph charts indicate deception.³⁴

This issue subsumes the issue of examiner certification. Certification is necessary but may not be sufficient. Ames’ examiner was certified, the examiners involved in CIA sex discrimination cases ³⁵ were certified, and it seems reasonable to assume that the examiners involved in the accuracy studies given in the first example were certified. Yet, in all these cases, examiner influence is clear.

Who will guard the guardians? We recognize that DOE will use controls to reduce examiner influence. However, we believe that additional actions may be necessary. DOE needs to ensure that examiners do not place any individual at a disadvantage for extrinsic reasons. This can happen during the pre-interview, the test, or re-examinations. DOE needs to determine whether the procedure is more threatening to particular ethnic groups, age groups, or genders. We believe that statistics should be kept and made available to the public regarding all non-negative results (deception indicated, no opinion, refusal to be tested, and test termination). Also, demographic and other pertinent information on all examiners should be a matter of public record.

Other agencies are being sued because of alleged abuse and discrimination. DOE should minimize potential diversion of national security funds to litigation and should demonstrate a commitment to diversity. Although the above measures may help, we believe that the best way for DOE to do this is to refrain from polygraph screening tests.

DOE plans to reduce the number of positives to reduce the issues resulting from false positives. The "Catch 22" is the minimization of real positives and the increased risk that subversives will not be detected. Reducing the target number of positives to an arbitrarily low number (2% is the security czar’s suggested number) will almost ensure that some guilty people will pass. Further, by giving individuals accused of wrongdoing the opportunity to exonerate themselves by taking a polygraph and by speeding up clearance processing by offering applicants a polygraph (now permitted), ill-intentioned people may more easily remain or become workers in the weapons complex.

Although scientific debate continues on the accuracy of polygraph techniques for ascertaining past criminal activities, the validity of using polygraphy for screening employees to predict future behavior is very questionable.

DOE must act to minimize the undesirable side effects of polygraph screening. The most immediate side effect is that of low morale and possible inconclusive and false positive responses. The announced policy of transferring people from a "cleared" job to an "uncleared" one is not enough—there are issues of records and career progression and development within the laboratories. In the longer term, DOE will have to refine its internal security systems to detect individuals who can deceive the polygraph. DOE will also have to address the issue of polygraphs in recruiting. How can negative recruiting effects be mitigated when a potential recruit is told that such testing may be required for employment?

Will those who are already employed and cleared have their access withdrawn until their evaluations are complete? Who will make career-impacting decisions and on the basis of what additional information? Will a standard background reinvestigation suffice or will a more thorough one be initiated? Will DOE focus its finite resources on individuals who probably pose no threat to national security instead of on effective systems to eliminate subversives? During the polygraph process, examinees will provide a great deal of information from both control and security-related questions. How will this information be used? Should individuals have the right to receive a copy of their polygraph results? What will the DOE’s policy be for passes, fails, and inconclusives?

DOE should establish a much clearer process regarding polygraphs. The process should include:

• a clarification of employees eligible for polygraph testing (Positions that DOE has determined have "access to the most sensitive categories of classified information and materials, as well as applicants for such positions" is too broad.),

• a clear indication of the types of behavior that are being searched for in the testing,

• a clear indication of the process to be followed once a positive or inconclusive indication is found or if the use of deliberate or inadvertent countermeasures is suspected or detected,

• the right of appeal and what constitutes an acceptable defense against an accusation,

• an indication of the documentation that will accompany an accusation,

• some proof that the follow-up will not bear any resemblance to anecdotal accounts of past practices for reviewing security suitability of employees accused in other venues (Note that in Figure 2, the far right hand side— referring loyalty concerns to the FBI—raises the specter of McCarthyism.), and

• a definition of the exceptions and an explanation about why should they be allowed (In Section 709.25 the DOE reserves the right under "a limited national defense and security exception" to rely on the results of a polygraph as the sole basis for taking action.).

What happens to national laboratory employees who refuse to be polygraphed? Supposedly, they will be moved to positions of equal responsibility and opportunity that do not require a access to sensitive information. However, finding an equivalent position may be impossible because of the specialized nature of work at nuclear weapons laboratories. Thus, it is possible that refusing to be polygraphed will result in career impacting consequences.

Polygraphy testing will impact recruiting and retention. Some persons may "fail" the test and others may refuse on principle to take the test because of the polygraph’s demonstrated lack of validity. In the long term, this will erode the caliber of the laboratory’s technical staff, with obvious impacts on research and development.

In the short term, employee commitment and morale may be lowered because polygraph screening tests create an atmosphere of distrust between employer and employee, are demonstrably unreliable, and indicate that DOE is unwilling to base security concerns on evidence. Daniel Jeffreys quotes some cogent words of warning on this subject: "‘The polygraph test is undermining morale throughout the [CIA],’ says Michael Kelly, a former intelligence officer who is now an attorney specializing in employee lawsuits against the CIA."³⁷

A recent survey of Sandia National Laboratories employees has indicated similar concerns regarding morale, recruiting, and retention:

With respect to external recruiting … an estimated 27% of the technical staff would not have applied to Sandia if a polygraph examination had been required.

With respect to retention, a total of 32% would (9%) or might (23%) transfer out of a position that required a polygraph and 15% would (2%) or might (13%) resign from Sandia if a polygraph was required.

[T]he effect on morale is another concern with respect to staff quality and productivity. Overwhelmingly … the respondents anticipate a negative effect … . About one-half anticipate a somewhat negative effect and another one-third anticipate a very negative effect, in contrast to the 3% that anticipate a positive effect.³⁸

We need to recreate the DOE culture of security consciousness. Due to environment, safety, and health concerns, former DOE Secretary Watkins opened operations in the complex to such a level that an agent could more easily piece together operations at the plants. Former DOE Secretary O’Leary subsequently ordered the declassification of thousands of documents and the use of uniformly colored badges for all employees, cleared or not.

Security clearances are the first line of defense against the insider threat. However, the rigor and quality of the security clearance process has degraded through the years, for both bureaucratic and budgetary reasons. Under the Atomic Energy Commission, all employees and contractors were subject to a Q-level background investigation performed by the FBI. Today, uncleared investigators do background checks and L-cleared administrators manage the database of clearances. Clearly, the present system needs greatly increased rigor. We need more Q clearances in the laboratories. We also need more Q clearances outside the laboratories (e.g., for background investigators and DOE database administrators).

In addition to the requirement for a security clearance, the laboratories operate under the DOE policy of an employee’s "need to know." This security principle requires that access to classified matter be limited to persons who possess appropriate access authorization and who require such access (need to know) in the performance of official duties. The Seniors believe that the need-to-know processes must be improved by increased use of Sigma levels, compartmentalized information, and code words for specific categories of information.

If the polygraph screening proposed by DOE is implemented, it must be integrated with the existing system of assessing the reliability of people who do weapons work. That system includes recruiting and hiring selectively, having a clearance process, doing periodic clearance updates, and asking managers to be vigilant for deviant behavior. Although the existing system is not perfect and its reliability is difficult to quantify, we doubt that polygraph screening will improve this system.

New security system requirements. Cyber security in particular needs to be improved throughout the national defense complex. Recent news regarding Moonlight Maze (where the Russians are suspected of computer hacking "sensitive military secrets, including weapons guidance systems and naval intelligence codes ..."⁴⁰) highlights the need for improvement.

We believe that the entire national laboratory security system should be improved using a systems approach in which the cost and benefits of changes can be measured. Valid indications of security levels and continuous improvement would result. A full systems evaluation is necessary because computer technology has fundamentally changed threats to national security.

Preventing compromise of information by individuals having custody is extremely difficult. Individuals can, if necessary, memorize documents and transcribe them at home. Therefore, we must ensure that a single insider (the most common subversive profile) cannot steal "the whole store" or some large subset of it. Techniques for preventing such extended compromise include strengthened need-to-know processes and cyber and physical security techniques to minimize the possibility that an individual with limited access to data can expand that access.

Paul Robinson provides an apt summary of the situation:

In my estimation, the counterintelligence program addressing laboratory espionage must become much more sophisticated if it is to be effective. An insider spy at the laboratories is likely to be a Ph.D. in a technical discipline and possess advanced knowledge of computer systems and their vulnerabilities. It will be important for DOE to work in close partnership with the FBI and other law-enforcement and intelligence agencies on methodologies for detecting and apprehending such spies.⁴¹

Countermeasures and false negatives. Most psychology experts believe that skilled subversives can use countermeasures to defeat polygraph tests. Countermeasures are a serious concern because false negatives give adversaries easier access to information. The potential for false negatives may also give the laboratories an unwarranted sense of security. Because of countermeasures, we don’t think that polygraph examinations will accomplish DOE’s intent—to deter or detect subversive individuals.

Accuracy. Reasonable accuracy can be expected for adequately controlled, specific-incident tests. However, it is unwarranted to assume these accuracy rates for screening applications, where accuracies have not been proven to be much better than chance.

False positives. The mathematical chance of incorrect identification of innocent persons as deceptive (false positives) is high in screening applications because only a very small percentage of the group being screened might be guilty. Many innocent individuals will have careers damaged by testing and the relationship between this cost and benefit is not evident. "Tuning" polygraph tests to decrease positive results increases the probability of false negative results, and further reduces its effectiveness in identifying subversives. No technical evidence supports the contention that false positive rates can be as low as 2%. Furthermore if rates are this low, it is doubtful that any subversives will be caught or deterred.

Security system. The entire security system should be improved using a systems approach in which the cost and benefit of changes can be measured. The system should be able to be prototyped, have mechanisms to measure its effectiveness, and be amenable to improvements. A real "service in the national interest" would be to define such a security system that improves national security in both the short and long term.

2. Scientific Validity of Polygraph Testing: A Research Review and Evaluation, Office of Technology Assessment (Henceforth called the OTA Report), November 1983, p. 8. (available at http://www.wws.princeton.edu/~ota/disk3/1983/8320.html).

3. References are listed in Appendix II.

4. Polygraph Examination Regulation, Federal Register, v. 64, 45062 (1999) (to be codified at 10 C.F.R. pts. 709, 710, and 711) (proposed Aug. 18, 1999).

5. Pete Domenici, "Domenici Concerned Over Polygraph ‘False Positives,’" Press Release, www.senate.gov/~domenici/press, August 6, 1999.

6. Jeff Bingaman, "Proposed Department of Energy Polygraph Examination Regulation," Memo to Secretary Bill Richardson, September 16, 1999.

7. Eli Lehrer, "Lies, Damned Lies and Polygraph Tests," Insight on the News, v. 14, n. 28, August 3, 1998, p. 44.

8. From page 98 of the OTA report: "A principal use of the polygraph test is as part of an investigation (usually conducted by law enforcement or private security officers) of a specific situation in which a criminal act has been alleged to have, or in fact has, taken place. This type of case is characterized by a prior investigation that both narrows the suspect list down to a very small number, and that develops significant information about the crime itself. When the polygraph is used in this context, the application is known as a specific-issue or specific-incident criminal investigation."

9. In a blind polygraph test, the evaluator of the test uses only the information recorded during the test, has absolutely no interaction with the person being tested, and is assumed not to have any other information (such as demographic data) about the person tested.

10. Dawson et al, "The Electrodermal Response," Principles of Psychophysiology—physical, social and inferential elements, J. T. Cacioppo and L. G. Tassinary, Eds., 1990, p. 312.

11. W. G. Iacono and D. T. Lykken, J. App. Psych., v. 82, 1997, pp. 426-433.

12. Norman Ansley, "The Validity and Reliability of Polygraph Decisions in Real Cases," Polygraph, v.19, 1990, pp. 169-181.

13. OTA Report, p. 97.

14. N. Ansley and M. Garwood, The Accuracy and Utility of Polygraph Testing, US Department of Defense Report, Washington, DC, 1984, p.61.

15. OTA Report, pp. 7-8.

16. OTA Report, p. 8.

17. D. T. Lykken, A Tremor in the Blood, Plenum Press, NY, 1998, p. 161.

18. Comparison of Psychophysiological Detection of Deception Accuracy Rates Obtained Using the Counterintelligence Scope Polygraph and the Test for Espionage and Sabotage Question Formats, Department of Defense, Fort McClellan, AL, Polygraph Inst. Report No.: DODPI93-P-0044; DODPI-R-0008, June, 95, Abstract.

19. OTA Report, p. 4.

20. OTA Report, p. 4.

21. OTA Report, p. 5.

22. OTA Report, p. 91.

23. C. R. Honts, D. C. Raskin, and J. C. Kircher, "Mental and Physical Countermeasures Reduce the Accuracy of Polygraph Tests," J. Appl. Psych., v. 79, n. 2, 1994, pp. 252-259.

24. W. G. Iacono and D. T. Lykken, J. App. Psych., v. 82, 1997, pp. 426-433.

25. Robert L. Park, What’s New, Washington, DC, Jun. 25, 1999.

26. Polygraph Examination Regulation, Federal Register, v. 64, 45062 (1999) (to be codified at 10 C.F.R. pts. 709, 710, and 711) (proposed Aug. 18, 1999).

27. James Matte, Forensic Psychophysiology Using the Polygraph, J.A.M. Publications, 1996, p. 296.

28. The KGB told Ames, "Get a real good night’s sleep. Be fresh and rested. Be cooperative. Develop rapport with examiner. … And try to remain as calm and easy as you can." (See David Wise, Nightmover, Harper Collins, 1995, p. 146.)

29. David Wise, Nightmover, Harper Collins, 1995, p. 211.

30. OTA Report, pp. 5-6.

31. National Defense Authorization Act for Fiscal Year 2000 (Printed w/ House Amend.), S. 1059, 106th Cong. § 3187(d) (1999).

32. Norman Ansley, "The Validity and Reliability of Polygraph Decisions in Real Cases," Polygraph, v.19, 1990, pp. 169-181.

33. W. G. Iacono and D. T. Lykken, J. App. Psych., v. 82, 1997, pp. 426-433.

34. James Matte, Forensic Psychophysiology Using the Polygraph, J.A.M. Publications, 1996, p. 296.

35. Daniel Jeffreys, "Getting Down on ‘The Farm.’ (CIA’s humiliating polygraph tests are making it difficult to hire and keep operatives: reprinted from The Independent, Nov. 27, 1996)," World Press Review, v. 44, n. 3, March, 1997, p. 30.

36. This description was derived based on a draft version of DOE N 472.2 "Use of Polygraph Examinations," a memo from Vic Reis to Rose Gottemoeller (Subject: Issuance of Notice on Use of Polygraph Examinations) with attached comments dated March 11, 1999, conversations with Richard Brown DOE/Defense Programs, and a DOE draft policy from the Office of Counterintelligence, 10 CFR part 709, Polygraph Examination Regulations.

37. Daniel Jeffreys, "Getting Down on ‘The Farm.’ (CIA’s humiliating polygraph tests are making it difficult to hire and keep operatives: reprinted from The Independent, Nov. 27, 1996)," World Press Review, v. 44, n. 3, March, 1997, p. 30.

38. Robert G. Easterling, "Commentary on DOE Proposed Polygraph Examination Regulation, 10 CFR, Parts 709, 710, 711," September 16, 1999.

39. National Defense Authorization Act for Fiscal Year 2000 (Printed w/ House Amend.), S. 1059, 106th Cong. § 3168(d) (1999).

40. Ron Edmonds, "Russian hackers steal US weapons secrets," Times Newspapers Ltd., July 25, 1999.

41. C. Paul Robinson, Sandia National Laboratories, "Testimony before the Senate Select Committee on Intelligence," July 14, 1999.

ANS	Autonomic Nervous System
CIA	Central Intelligence Agency
CQT	Control Question Technique
DOE	Department of Energy
EPPA	Employee Polygraph Protection Act
FBI	Federal Bureau of Investigation
OTA	Office of Technology Assessment

N. Ansley and M. Garwood, The Accuracy and Utility of Polygraph Testing, US Department of Defense Report, Washington, DC, 1984.

Jeff Bingaman, "Proposed Department of Energy Polygraph Examination Regulation," Memo to Secretary Bill Richardson, September 16, 1999.

Comparison of Psychophysiological Detection of Deception Accuracy Rates Obtained Using the Counterintelligence Scope Polygraph and the Test for Espionage and Sabotage Question Formats, Department of Defense, Fort McClellan, AL, Polygraph Inst. Report No.: DODPI93-P-0044; DODPI-R-0008, June, 95.

Dawson et al., "The Electrodermal Response," Principles of Psychophysiology—physical, social and inferential elements, J. T. Cacioppo and L. G. Tassinary, Eds., 1990.

DOE Cover Letter for "Issuance of Notice of Proposed Rulemaking on Polygraph Examination Guidelines," Mar. 11, 1999 (DOE Notice 472.2 extended to include contractor employees).

Pete Domenici, "Domenici Concerned Over Polygraph ‘False Positives,’" Press Release, www.senate.gov/~domenici/press, August 6, 1999.

Robert G. Easterling, "Commentary on DOE Proposed Polygraph Examination Regulation, 10 CFR, Parts 709, 710, 711," September 16, 1999.

Ron Edmonds, "Russian hackers steal US weapons secrets," Times Newspapers Ltd., July 25, 1999.

C. R. Honts, D. C. Raskin, and J. C. Kircher, "Mental and Physical Countermeasures Reduce the Accuracy of Polygraph Tests," J. Appl. Psych., v. 79, n. 2, 1994, pp. 252-259.

W. G. Iacono and D. T. Lykken, J. App. Psych., v. 82, 1997, pp. 426-433.

Daniel Jeffreys, "Getting Down on ‘The Farm.’ (CIA’s humiliating polygraph tests are making it difficult to hire and keep operatives: reprinted from The Independent, Nov. 27, 1996)," World Press Review, v. 44, n. 3, March, 1997.

Eli Lehrer, "Lies, Damned Lies and Polygraph Tests," Insight on the News, v. 14, n. 28, August 3, 1998.

D. T. Lykken, A Tremor in the Blood, Plenum Press, NY, 1998.

James Matte, Forensic Psychophysiology Using the Polygraph, J.A.M. Publications, 1996.

National Defense Authorization Act for Fiscal Year 2000 (Printed w/ House Amend.), S. 1059, 106th Cong. § 3168(d) and § 3187(d) (1999).

Robert L. Park, What’s New, Washington, DC, Jun. 25, 1999.

Polygraph Examination Regulation, Federal Register, v. 64, 45062 (1999) (to be codified at 10 C.F.R. pts. 709, 710, and 711) (proposed Aug. 18, 1999).

C. Paul Robinson, Sandia National Laboratories, "Testimony before the Senate Select Committee on Intelligence," July 14, 1999.

Scientific Validity of Polygraph Testing: A Research Review and Evaluation, Office of Technology Assessment (Henceforth called the OTA Report), November 1983. (available at http://www.wws.princeton.edu/~ota/disk3/1983/8320.html).

David Wise, Nightmover, Harper Collins, 1995.

FAS | Secrecy | Polygraph || Index | Search | Join FAS