The following document is a follow-up memorandum to Dr. Donald M. Kerr, the FBI Assistant Director for the Laboratory Division, from FBI Supervisory Special Agent Dr. Drew C. Richardson, the FBI's leading scientific expert on polygraphy. In September 1997, Dr. Richardson testified before the U.S. Senate that polygraph screening is "completely without any theoretical foundation and has absolutely no validity."
David Tenenbaum had unsuccessfully sought release of this memorandum in a lawsuit stemming from unsubstantiated government accusations that he was an Israeli spy: polygraph abuse figures prominently in his case. While the FBI inexplicably withheld this relevant document from Mr. Tenenbaum's counsel, the James Madison Project has obtained it under the Freedom of Information Act (FOIA).
With Dr. Richardson's written warnings to FBI officials now a matter of public record, senior FBI officials cannot claim that they were unaware of the problems of polygraph screening.
In this memorandum, Dr. Richardson provides a harsh assessment of polygraph validity and utility. After pointing out that there is virtual unanimity in the relevant scientific community that polygraph screening has no diagnostic value, Dr. Richardson poignantly observes that
FEDERAL BUREAU OF INVESTIGATION
Precedence: ROUTINE Date: 10/25/1999
To: Laboratory Attn: Dr. Kerr
From: Laboratory
Administrative Section
Contact: Drew C. Richardson, Ext. 2375
Approved By: Richardson Drew C:dcr
Case ID #: 66F-HQ-A1265133 (Pending)
Title: POLYGRAPH PROGRAM
LABORATORY DIVISION
Synopsis: To provide requested and other material for the
Laboratory Division's (LD) Executive Management as it relates to
the FBI's polygraph program.
Reference: Memo from Dr. Richardson to Dr. Kerr, dated 9/13/99
and EC from Mr. Bogner to Dr. Kerr, dated 9/22/99
Enclosures: 1) Memo from Dr. Richardson to Dr. Kerr, dated
9/13/99 (not previously serialized and provided for benefit of
those receiving this EC), (2) Journal of Applied Psychology
article entitled "The Validity of the Lie Detector: Two Surveys
of Scientific Opinion," (3) Journal of Applied Psychology article
entitled "Validity of the Control Question Polygraph Test: The
Problem of Sampling Bias (4) Forensic Reports article entitled
"The Emperors's [sic] New Clothes: Application of Polygraph Tests in
the American Workplace,",(5) Report entitled "Report of Peer
Review of Johns Hopkins University/Applied Physics Laboratory,"
(6) Communications from Dr. Alan P. Zelicoff, Senior Scientist,
Sandia National Laboratories to Dr. Andrew Ryan, Chief of
Research, DOD Polygraph Institute dated 10/9/99.
Details:
Preliminary Issues
Per referenced memo prepared by Mr. Bogner, I was asked
to provide additional materials related to my background, the
sources of the opinions I expressed in earlier referenced memo,
and for an elaboration on some issues I raised.
My terminal academic degree is a Ph.D. degree in
physiology obtained through study at the George Washington
Medical Center. My dissertation emphasis was in cardiovascular
To: Laboratory From: Laboratory
Re: , 10/25/1999
physiology and centered around various non-invasive technologies
used to measure Autonomic Nervous System (ANS) changes associated
with a "lie detection" task. I also completed the course work
requirements for a Ph.D. degree in pharmacology and was awarded
for having maintained the highest academic average of any
graduate student while in said program. During the approximately
two year period a decade ago that I was formally associated with
the Bureau's polygraph research program, I prepared oral and
written communications related to various cardiovascular indices
used in "lie detection" tasks, pharmacological countermeasures,
neurophysiological indices that might be used in a concealed
information task and methods (and mathematical models) of
polygraphic data transformation. During that time I was a
featured speaker at both Federal Polygraph Symposiums that
occurred and a speaker at an invited panel presentation at the
annual international meeting of the Society for
Psychophysiological Research (SPR). Furthermore I was given a
letter of accommodation [sic] by the Director for my commentary on the
particulars of pharmacological countermeasures related to a
specific criminal case which ultimately led to a conviction in a
major espionage case. I am also a graduate of the basic examiner
course taught at the Department of Defense Polygraph Institute.
The aforementioned having been said, I maintain that
aside from the general maturity associated with biomedical
graduate study, the most serious analytical and relevant
evaluative skills that I possess come from my twenty year
involvement in the serious sciences of chemistry, toxicology, and
pharmacology. During that time, I served as a research scientist
in the Organic Chemistry Department of Burroughs Wellcome
Company, a pharmaceutical company located in Research Triangle
Park, North Carolina, and as an examiner in the Chemistry-
Toxicology unit of the FBI Laboratory (in excess of 2000 cases
worked resulting in approximately 100 expert testimonies). It is
largely through these experiences that I developed an
appreciation for serious science and the ability to recognize
what is science and what is not. It is the concepts of
experimental design, validity, scientific control, and methods of
statistical analysis, etc. demanded by the aforementioned
scientific disciplines which allows me to evaluate that put forth
within the realm of "polygraph research."
I have presented the aforementioned background material
in response to a request for same as contained in referenced EC,
but I would maintain that any particular emphasis being put on
this material is to miss the point and to trivialize the issue.
Of considerably more value than my opinions or supporting
background for such opinions is where does truth lie with a given
issue and the relevant collective opinion(s) of qualified experts
as expressed in the peer reviewed scientific literature. Because
2
To: Laboratory From: Laboratory
Re: , 10/25/1999
of the paramount importance of this, I will return to this theme
after addressing a few more preliminary issues.
I was asked if the opinions and assertions that I have
expressed regarding polygraph practice are derived from the
scientific literature or from the accounts of polygraph
examinees. The scientific research and opinion survey literature
would indicate that there are serious shortcomings arising from
the use of polygraph as a diagnostic tool. The manifestation of
the error suggested by scientific theory can only be adequately
explored through an analysis which includes the accounts of
examinee experiences. My answer, therefore, is that both are
sources of potentially meaningful information, and that neither
should be ignored. Other issues such as allegations of bias,
falsely-claimed confessions/admissions or unethical conduct on
the part of polygraph examiners can only be evaluated by
interviewing the participating parties to include the
complainants (polygraph examinees).
I was asked about the validity of other aspects of the
applicant process, i.e., background, panel interview,
psychological testing, etc. I have no particular knowledge or
expertise concerning these phases of an applicant investigation.
My interest, concern, and expertise lies in the use of the
polygraph as a diagnostic tool. I would suggest that polygraphic
validity is independent of the relative strengths and weaknesses
of the other phases of the applicant process and that as
presently constituted none of the other phases will overcome the
serious consequences of a false positive polygraph result in any
given applicant investigation. The consequences of such error
contained in polygraph results, however, might be ameliorated by
background investigation (not presently done).
Scientific and other Relevant Community Opinion
The two academic bodies which probably most represent
physical and mental health in the minds of the lay public are the
American Medical Association and the American Psychological
Association. Both groups have adopted formal positions strictly
opposing the use of polygraph examinations for screening
purposes. Perhaps of more concern and import is a recently
published survey representing the opinion of the relevant
scientific community. The article (enclosure 2) entitled "The
Validity of the Lie Detector: Two surveys of Scientific Opinion"
is a survey of the members of the Society for Psychophysiological
Research (SPR) and the Fellows of the American Psychological
Association's Division I (General Psychology). If polygraphy is
to have a parent academic discipline, it would be
psychophysiology, and it would be represented at the highest
3
To: Laboratory From: Laboratory
Re: , 10/25/1999
level in these two bodies. By a large majority these scientists
held the following positions: (1) The usefulness of the control
question test (CQT, polygraphic mainstay in Bureau examinations)
is questionable, (2) that it (CQT) is not based on principles
which are "scientifically sound," (3) that it should not be
admitted as evidence in court, (4) that proponent estimates of
polygraph accuracy are grossly overstated, and (5) that mock-
crime derived estimates of polygraph accuracy are unreasonable
estimates of real-life polygraph accuracy, etc.etc.
This scientific community opinion data has not been
lost on the legal system. In the last several years (based on
oral reports from well known expert witnesses), it is estimated
that proponents have tried to introduce polygraph evidence into
U.S. Federal District Court testimony approximately fifty (50)
times. Exact figures will be provided when obtained. These
proponents were unsuccessful all but one time, this result
corresponding with the overwhelming opposition to such
introduction by the relevant scientific community.
Issues Relating to Polygraph Validity and Scientific Control
Although an elaborate treatment of the subject is
beyond the scope of this EC, polygraph validity determinations
are characterized by (1) environment (mock crimes vs. analysis of
field charts) and criteria for establishing ground truth, i.e.,
determined a priori in laboratory studies and confessions with
field studies; (2) the polygraphic format under consideration,
e.g., CQT with probable lie controls, CQT with directed lie
controls, etc., and (3) polygraph application, e.g., criminal
specific tests, applicant screening exams, etc. With regard to
the latter characteristic, "application," although often confused
and misused by polygraph operational personnel, a polygraph
validity study done in a specific criminal setting has no bearing
on the validity of polygraph screening. I will return shortly to
differences between the two.
Both laboratory "mock-crime" studies and confession-
based field studies suffer from certain inherent problems. The
laboratory or analog studies are characterized by a lack of
external validity, i.e, volunteer subjects are not representative
of criminal suspects, have different motivation and concerns, and
generally take the exams under different circumstances than their
real life counterparts. Field studies which use confessions
following a polygraph exam as a criteria for determining ground
truth must of necessity overestimate the accuracy of a CQT exam.
By [a] priori ground rules the results of an innocent subject who is
found to be deceptive and who does not confess will be excluded
from such a study. The default result is that the accuracy of
the polygraph exam for determining the lack of deception by
innocent examinees will be overstated. Although a plethora of so
4
To: Laboratory From: Laboratory
Re: , 10/25/1999
called polygraph accuracy studies have been reported in polygraph
trade journals and other non peer reviewed literature, and
although a substantial number with the aforementioned flaws have
been reported in the peer reviewed literature, probably the only
one validity study concerning criminal-specific studies which
would meet the muster of external validity and a reasonable basis
for determining ground truth is one (enclosure 3) whose results
were published in the Journal of Applied Psychology entitled
"Validity of the Control Question Polygraph Test: The Problem of
Sampling Bias". In this field study, confessions were not based
on someone having failed a polygraph exam, allowing the
physiological charts of polygraph exams to be blindly rescored.
The results of this rescoring were that 45% of the innocent
suspects were erroneously classified as deceptive by the CQT.
This outcome would suggest a high false positive rate may be the
outcome of a CQT-based polygraph examination.
With regard to the accuracy of polygraph screening,
only one study has been performed which uses the expected base
rate for the criminal activity being screened for. In this case,
the screening related to counterintelligence issues, the exam
format was the CSP exam used by the Department of Defense, and
the results of this study (enclosure 4) were published in
Forensic Reports and entitled "The Emperor's New Clothes:
Application of Polygraph Tests in the American Workplace." The
study found that the CSP exam correctly classified only two
percent of the guilty subjects and was characterized as "being a
nearly worthless screening device." Furthermore the author
concluded that "Given that polygraph tests used for screening are
likely to be inaccurate with guilty subjects to begin with, the
existence of effective countermeasures virtually assures that a
well-prepared and determined opponent could achieve nearly a 100%
penetration of the national security polygraph screen.["]
It should be pointed out that the author of the last
mentioned article is one of the few serious academic supporters
(to include his mentor Dr. David Raskin and a handful of Dr.
Raskin's students from the University of Utah) of criminal
specific testing. Even this group joins the rest of academia in
almost universally characterizing polygraph screening as
worthless as a diagnostic tool. If this be the case, then one is
left wondering if the test has no diagnostic benefit, why are we
willing to accept the possible diagnostic cost of false positive
results and damage to national security due to a complete lack of
any ability to detect espionage and related activities.
Issues Relating to Polygraph Utility and Deterrence
Even if polygraph screening has no diagnostic value,
what about issues of deterrence and utility. With regard to
5
To: Laboratory From: Laboratory Re: , 10/25/1999 deterrence, I would suggest that (1) a technique which has no diagnostic value would require such a universal bluff and disinformation campaign as to be impractical, if not comical, to continue over a period of time, (2) examples of major espionage figures, i.e., Aldrich Ames, etc. who committed their acts in the midst of ongoing polygraph programs does not speak highly of polygraph deterrence, and (3) If one were to examine CIA penetrations vs. FBI penetrations during the twenty year period preceding the Bureau's present applicant screening program (a time in which the CIA did polygraph screening and the FBI did not), one would find a far greater number of agency penetrations than Bureau penetrations. This result is of course not consistent with any great deterrent effect stemming from a polygraph program. With regard to utility, I was asked about the high number of deceptive polygraph results which were followed by admissions or confessions of some sort. Because a confession is sought from an examinee following a deceptive polygraph examination, it is not surprising that there would be a higher correlation between confessions and this phase vs other phases of the applicant investigation where no confession is elicited. I believe the following issues need to be addressed regarding notions of polygraph utility: (1) If one is willing to believe that confessions validate the preceding polygraph results, is one willing to accept that those polygraph results which are not followed by a confession are erroneous, i.e., false positive results? (2) If one accepts these confessions to be accurate and true, is there any evidence that a skilled interviewer/interrogator in the absence of a polygraph exam would not have obtained the same results? and (3) Are the confessions necessarily true? A small number of examinees have come to me alleging that in addition to erroneous polygraph results, polygraph examiners have falsely claimed admissions/confessions to have been made by these individuals. These examinees claim to have discovered this circumstance through the serendipity of discovery process (FOIPA, Personnel Division response letters, etc) not initially focused on what they discovered to their horror. Obviously, I do not know where truth lies with this issue, but I do believe if in fact it did occur with these individuals, it may have occurred with many others, i.e., most examinees would have no reason to expect such a thing had occurred and would not initiate the lengthy discover[y] process which would reveal it. At one time (perhaps still), Bureau examiners were evaluated (critical elemant of annual performance evaluation) with regard to the number of confessions/admissions they obtained following deceptive polygraph results. Although I would hope and expect that the vast majority of examiners are people of integrity who would not be tempted by such motivation, it is obvious that a less than scrupulous examiner could easily 6
To: Laboratory From: Laboratory
Re: , 10/25/1999
fabricate or stretch the truth about an admission with the high
probability of never being exposed. I don't know what percentage
of claimed admissions/confessions are supported by signed sworn
statements, but obviously this would have bearing in determining
whether such a problem could exist.
For all of the aforementioned reasons, I find it
questionable at best as to whether polygraph utility justifies
the almost certain lack of any validity associated with polygraph
screening.
Other Issues
Although not directly related to issues regarding
polygraph screening, because they seem to be foundational to the
present polygraph program, I have provided materials (enclosures
5 and 6) which relate to the research capabilities and insight of
the Department of Defense Polygraph Institute and to the merits
of the hardware and software scoring algorithm for the
computerized polygraph system currently used by Bureau examiners.
The articles are authored by preeminent scientists and are highly
critical of both aforementioned topics. These materials are
self-explanatory and are provided for the information of those
receiving this EC.
7
To: Laboratory From: Laboratory
Re: , 10/25/1999
LEAD (s):
LABORATORY
AT WASHINGTON, DC
SSA RICHARDSON will continue to provide materials(s)
upon request or as he becomes aware of additional relevant
information.
** 1 - Dr. Kerr
1 - Mr. Bogner
1 - Mr. Devincentis
1 - Dr. Richardson
8
Transcription and HTML by AntiPolygraph.org