Podlesny - Forensic Science Communications - July 2003
July 2003 - Volume 5 - Number 3
Research and Technology
A Paucity of Operable Case Facts Restricts Applicability of the Guilty Knowledge Technique in FBI Criminal Polygraph Examinations
John A. Podlesny
Operations Research Analyst
Chemical Biological Sciences Unit
Federal Bureau of Investigation
To evaluate the applicability of the guilty knowledge technique, 758 polygraph examination cases were surveyed for operable case facts (keys). Polygraph examiners and Special Agents assigned to 25 FBI field divisions provided the case information. The 95% confidence interval for guilty knowledge technique applicability in criminal investigations was 5.0% to 8.6%. The guilty knowledge technique was inapplicable where contextual limitations precluded keys (85.0%), where potential keys had been revealed (4.4%), and where the number of potential keys was insufficient (one or two, 4.0%). Although its applicability in FBI criminal investigations is restricted, the guilty knowledge technique could be applied to a small proportion of examinations.
Lykken (1988b p. 303) asserted that “one of the most accurate and socially useful of psychological tests [the guilty knowledge test/technique (GKT) for deception detection] is being arbitrarily neglected.” Others agreed, presenting the guilty knowledge technique as a superior alternative to present methods (Ben-Shakhar 1991; Ben-Shakhar and Furedy 1990; Furedy 1993; Furedy and Heslegrave 1991; Iacono and Patrick 1987; Kleinmuntz and Szucko 1982; see also Lykken, 1981, 1991, 1992).
Indeed, the guilty knowledge technique is appealing. Described by Benjamin Burack (1955, p. 416) and adopted by Lykken (1959), the guilty knowledge technique is based on the plausible assumption that guilty, but not innocent, subjects will produce their largest physiological reactions to undisclosed case details (keys) presented among incorrect, but plausible, alternatives. The guilty knowledge technique has consistently demonstrated criterion validity in categorizing innocent and guilty roles of subjects in simulated crimes (e.g., Bradley and Rettinger 1992; Davidson 1968; Iacono et al. 1984; Iacono et al. 1992; Lykken 1959; Podlesny and Raskin 1978).
Lykken (1988b) attributed neglect of the guilty knowledge technique to the “belief that appropriate GKT items [keys] could not be designed in the field situation” (p. 303), an assumption that he considered “overly pessimistic in most specific-issue situations in which polygraphic interrogation might be contemplated” (p. 304). At the time, it was not known whether keys would be available in real-life contexts, since there was a lack of empirical information. Others claimed that there are practical limitations on the availability of keys (Abrams 1975; Forman and McCauley 1986; Kircher and Raskin 1992; Podlesny and Raskin 1978; Raskin 1988, 1989; Raskin and Kircher 1991; Rosenfeld 1995). Presumed impediments were the legitimate access of many suspects to case details; the release of case information through investigative interviews, the media, or legal discovery; and the possible failure of guilty suspects to notice or remember details of their crimes.
There have been few studies pertinent to the applicability of the guilty knowledge technique. For Israeli examinations in which guilty knowledge techniques had been used, Elaad (1990) and Elaad, Ginton, and Jungman (1992) reported that the number of keys per examination ranged from one to six and averaged 2.0 and 1.8, respectively. Those reports did not enumerate cases in which guilty knowledge techniques were not used because no keys were available. Podlesny (1993) reviewed 61 criminal examinations and found that four or more apparently usable keys appeared in the case records for eight (13.1%) of the examinations. In that study, 45 of the 61 examinees (73.8%) had legitimate access to the case details. However, the case information was obtained entirely from records, and the sample was small and selective.
More reliable facts are needed to determine the extent to which sufficient, operable (usable) keys are available to support the application of the guilty knowledge technique. The present study estimates the availability of keys in FBI criminal cases in which deception detection examinations are used. Case information related to 758 examinations was collected from FBI field offices throughout the United States. The results indicate that sufficient keys are likely to be available in about 5.0% to 8.6% of FBI criminal examinations and that most examinations are intractable to the guilty knowledge technique for explicit reasons.
Polygraph examiners from 25 FBI field offices collected case data for each criminal examination assigned to them from November 1, 1993, through August 31, 1994, a total of 758 examinations. The examiners collected the data while preparing for the examinations and provided the data after the examinations were complete (they did not conduct any guilty knowledge tests). The sample comprised 35.6% of the criminal examinations conducted by the FBI during that period (2131), (Table 1). The examiners received written and verbal instructions about the survey goals, methods for identifying keys, sources of information, and procedures for reporting data.
A potential key was defined as a specific case fact that would likely be known to guilty person(s); that, if not revealed, would likely not be known to innocent persons; and for which it was believed at least a few plausible alternatives could be listed (Burack 1955; Lykken 1959, 1974). An operable key was a potential key that had not been revealed outside the investigation and was, therefore, assumed to be usable in a guilty knowledge test. To identify potential keys, the examiners queried all available sources of information including case officials, case records, and media reports. For each examination, the examiner provided a case summary, the purpose of the examination, and a list of potential keys. Where there were no potential keys, the examiner provided an explanation. Where potential keys had been revealed, the examiner provided a list of those keys with an explanation concerning their release.
The reports were reviewed for completeness and intelligibility. In those instances where further information was needed, an assistant interviewed the examiner. The revised information was then reviewed and, as required, the examiners and case Agents were interviewed by the author. The case Agents were the investigators who requested the examinations. Data collection continued in each instance until all apparent keys were identified or a cogent explanation was obtained for a lack of keys. This process produced a database containing the revised data from the original reports and the interviews.
In the case information related to the 758 examinations, there were 587 potential keys. Of these, 232 had been revealed, either through the news media or during interviews of the examinees. Thus, there were 355 operable keys, an average of 0.47 operable keys per examination request.
|Distribution of the Examination Cases by the Number
of Potential and Operable Keys Found
Table 2 shows the distribution of the examination cases by the number of potential and operable keys found. It was necessary to select a criterion in order to decide whether the guilty knowledge technique would have been applicable to any given examination. Lykken (1988b) recommended that six or more keys should be used, claiming that this would be possible in “many criminal investigations...with a little effort and imagination” (p. 304). Six or more keys would provide strong protection against chance false-positive errors. By Lykken’s criterion, the best estimate for applicability was 2.1% (16 of 758 examinations). A more lenient criterion of at least three operable keys may be considered minimally sufficient (Forman and McCauley 1986). Assuming three operable keys and four incorrect alternatives per key, the binomial probability that the largest physiological reactions would occur by chance in response to two or three of the keys is .10. In the context of an investigation, this might be sufficient to provide direction to investigators. By this criterion, there were 51 (6.7%) examinations to which the guilty knowledge technique could have been applied. A resampling procedure (Diaconis and Efron 1983; Efron and Tibshirani 1991; Simon and Bruce 1991) produced a 95% confidence interval of 5.0% to 8.6% for guilty knowledge technique applicability to FBI criminal examinations for the latter (lenient) criterion. This estimate, produced as follows, assumes that the present sample is representative of FBI criminal investigations in which examinations are conducted. Fifteen thousand samples each of n = 758 were drawn at random with replacement from the integers 1 through 758 (total examinations). A count of the occurrences of 51 integers (1 through 51, the number of guilty knowledge technique-applicable examinations in the sample) was obtained for each sample. The 95% confidence interval of 5.0% to 8.6% was obtained from the distribution of counts for the 15,000 samples.
The guilty knowledge technique was inapplicable to 707 of the examination cases (93.3%). There were 33 examinations (4.4%) with three or more potential keys, but less than three operable keys because some keys had been revealed. There were 30 examination cases (4.0%) with only one or two potential keys. There were 644 examination cases (85.0%) with no potential keys. The case circumstances in which keys were lacking in those cases is as follows:
|1.||The examinee had legitimate reason for access to the case information (281, 37.1%). Examples were cases in which bank tellers, armored-car personnel, or other employees, who had direct access to business activities and conversations with coworkers, were suspected of internal thefts.
|2.||The examinee was the source of information requiring verification (152, 20.1%). Examples were cases in which there was a need to verify the statements of sources or allegers in the absence of factual information.
|3.||A violation was suspected only on the basis of allegations (90, 11.9%). Examples were cases in which individuals were suspected of child molestation or bribery, based on the allegations of other individual(s) in the absence of factual information.
|4.||The examinee admitted involvement, but denied any intention to commit a violation (29, 3.8%). Examples were cases in which individuals admitted fraudulent activities, but claimed they had not known that those activities were illegal.
|5.||It was unknown whether a violation had actually occurred, and the examination was based on the suspicions of investigators (19, 2.5%). Examples were cases in which individuals pleaded guilty to drug violations and were suspected of having engaged in other unspecified illegal activities or of withholding information about recoverable assets.
|6.||There was a lack of factual case information for various reasons (8, 1.1%). An example was a case in which a former bank teller was suspected in the theft of an indefinite amount of money that had disappeared mysteriously.
|7.||Various combinations of the above (65, 8.6%).|
Over half of the 51 guilty knowledge technique-applicable examinations were in cases of bank robbery (19) or kidnapping (10). The proportions of the sample totals for those case types were 25.0% and 17.2%, respectively. Post hoc-resampling procedures (similar to the above) produced 95% confidence intervals of 15.8% to 35.5% for bank robberies and 8.6% to 27.6% for kidnappings. Thus, guilty knowledge techniques were more applicable in bank robberies and kidnappings than in the general sample (5.0% to 8.6%).
The results are consistent with previous studies indicating a paucity of operable keys in practice (Elaad 1990; Elaad et al. 1992; Podlesny 1993). The present survey provides more reliable estimates than those studies because the sample is large, because it contains information from all available case resources, including examiners, investigators, records, and news media, and because examinations were not selected on the basis that guilty knowledge techniques had been used, as in the studies of Elaad and coworkers (1990, 1992). Although many favorable comments on the applicability of the guilty knowledge technique have appeared (cited above), these have not yet been supported by facts.
Lykken (1974, p. 728) suggested that “...the guilty knowledge method simply cannot be used in many situations in which the lie detector is now used...” The present empirical estimate indicates that the guilty knowledge technique cannot be used in the large majority of FBI criminal investigations in which deception detection is used. Furthermore, where there were no potential keys (85.0% of the sample), this was attributable to contextual limitations of the cases rather than to insufficient preparation or “effort and imagination” (Lykken 1988b p. 304). It does not appear that the guilty knowledge technique has been “arbitrarily neglected” (Lykken 1988b p. 303); rather, it is likely that the GKT has not been used because it is rarely applicable.
Implications for Practice
In consideration of restricted applicability, the guilty knowledge technique is not a general alternative to direct-questioning methods, which are the standard of practice (as suggested by Ben-Shakhar 1991; Ben-Shakhar and Furedy 1990; Furedy and Heslegrave 1991; Iacono and Patrick 1987; Kleinmuntz and Szucko 1982; Lykken 1981, 1988a, 1991). Direct-questioning methods, such as control-question tests, do not require operable keys.
|Prior to the introduction
of guilty knowledge techniques,
there is a need to establish
for selecting operable keys,
selecting alternative stimuli,
structuring and presenting tests,
and interpreting tests.
Because sufficient operable keys are available relative to a small proportion of examinations, there is unrealized potential for applying guilty knowledge techniques in some instances. Where applicable, guilty knowledge techniques might be introduced as a supplement to other methods, possibly adding protection against false positives (Iacono and Patrick 1987) or as a primary means for demonstrating probable connections between guilty individuals and violations. It may be possible to increase the proportion of guilty knowledge technique-applicable examinations by a few percent through stronger controls on the unnecessary release of case information. However, in practice, it is often necessary to release information in order to apprehend perpetrators and locate victims (e.g., clothing worn by bank robbers and kidnapping victims). Prior to the introduction of guilty knowledge techniques, there is a need to establish scientific standards for selecting operable keys, selecting alternative stimuli, structuring and presenting tests, and interpreting tests.
The examiners who provided much of the case information used in this study normally practiced direct questioning methods. Thus, it may be suggested that the examiners, knowing the purpose of the study, would have favored the finding that keys were lacking. However, the examiners were not unfamiliar with the use of case facts in examinations, since they were trained with the “peak of tension test,” a technique that similarly depends on operable case facts and differs in test format from the guilty knowledge technique. Further, the study procedure of collecting a summary of each case, a statement of the purpose of each examination requested, and a cogent explanation in each case where keys were reportedly lacking tends to counteract bias. It was not sufficient for an examiner to report only that there were no keys. This procedure made it possible to provide a circumstantial categorization and enumeration of cases in which keys were lacking (above). The reader may wish to consider how it could be possible to formulate operable keys in each of the seven categories of case circumstances described in the Results. Finally, exhaustive efforts were made to identify potential and operable keys wherever possible. In instances where it seemed that keys might have been overlooked, the author interviewed the case Agents; the case Agents were not examiners and had no apparent commitment to any polygraph technique.
Because the sample included only cases in which polygraph examinations were conducted in FBI criminal investigations, the results may not generalize to other contexts. Guilty knowledge techniques could be either more or less applicable to investigations in other jurisdictions with differently distributed caseloads. For example, burglaries, usually investigated locally, may be productive of keys based on the methods of entry and specific items taken. The results also do not necessarily generalize to FBI criminal investigations where examinations are not requested. Judging by the reasons for the lack of keys (above), the FBI frequently employs polygraph examinations to provide direction to ambiguous investigations. It is possible that guilty knowledge techniques might be more applicable in cases where better evidence generates potential keys; however, in those cases there may be less need for examinations.
Informed decisions concerning introduction of the guilty knowledge technique require information on validity as well as applicability. The results do not provide information on the validity of guilty knowledge techniques or any other deception detection method(s).
This study provides an estimate of guilty knowledge technique applicability where deception detection examinations are requested in FBI criminal investigations. There is a lack of research regarding the applicability of guilty knowledge techniques in other law enforcement contexts, such as state or local investigations or those of other federal agencies. Research might also address applicability to investigations in which examinations are not presently requested. The results indicate that attempts to identify applicable cases may, at least occasionally, be successful; however, anecdotal case reports (Lykken 1981, pp. 257-296) are not sufficient to provide estimates of applicability. Appropriate sampling schemes and full reporting of both applicable and inapplicable instances are required for such estimates.
Laboratory studies, contrived to have adequate numbers of operable keys, have consistently demonstrated the validity of guilty knowledge techniques (citations above); but the field validity studies produced large proportions of false negatives (Elaad 1990; Elaad et al. 1992). Those authors suggested that the small average number of keys (2.0 and 1.8, respectively) may have contributed to the false-negative rates. It appears that the guilty knowledge tests had frequently been used in cases where they were inapplicable. Field-validity research on cases with sufficient operable keys would provide more reliable estimates of field validity. Elaad, Ginton, and Jungman (1992, p. 765) suggested that “...some guilty examinees do not notice or remember all the facts of the crime available to the investigator.” The likelihood a guilty person will recognize case details may be affected by the passage of time, the examinee’s characteristic interests, active versus passive acquisition, advance preparation versus incidental contact, other similar violations remembered or planned by the examinee, and the distinctness of keys and alternatives (Elaad 1990). Research on such factors could clarify the requirements for valid operable keys and alternative stimuli.
The findings evidence both restrictions on and opportunities for application of the guilty knowledge technique. Applicability to FBI investigations is restricted; as a consequence, direct-questioning methods are advantageous where applicability is concerned. Nevertheless, the guilty knowledge technique is applicable to some examinations, and practitioners may wish to consider the utility of guilty knowledge techniques in cases where sufficient operable keys do exist.
The author is grateful for the assistance of the FBI polygraph examiners who provided the case reports for this research: Claude T. Adams; Ronald L. Clenney; Clifford Cormany, Jr.; Frank A. Cryan; Lester L. Davis, Jr.; Thomas F. Donlan, III; Robert J. Drdak; Salvador R. Escobedo; Edward T. Hamara; Michael Hanna; William C. Holloman; Ronald R. Homer; Mark R. Johnson; Walton G. Jones; James E. Judd; Stephen T. Kives; W. Ronald Lilly; H. Hadley McCann, Jr.; Michael A. McCrery; Blaine D. McIlwaine; Robert H. Meredith; Leonard J. Michaud; Michael R. Napier; Donald W. Ramsey; Robert E. Reich, Jr.; Arthur G. Richards, Jr.; William K. Teigen; John R. Trimarco; and Donald E. Wright. Debrah L. Hauger assisted with data collection. Colleen Wade and Tonya L. Catterton assisted with the preparation of the manuscript. Francis T. Morgovnik provided administrative assistance.
Abrams, S. A response to Lykken on the polygraph, American Psychologist (1975) 30(6):709-711.
Ben-Shakhar, G. Future prospects of psychophysiological detection: Replacing the CQT by the GKT. In: Advances in Psychophysiology, Volume 4. J. R. Jennings, P. K. Ackles, and M. G. H. Coles, eds. Jessica Kingsley, London, 1991, pp. 193-199.
Ben-Shakhar, G. and Furedy, J. J. Theories and Applications in the Detection of Deception: A Psychophysiological and International Perspective. Springer-Verlag, New York, 1990.
Bradley, M. T. and Rettinger, J. Awareness of crime-relevant information and the guilty knowledge test, Journal of Applied Psychology (1992) 77(1):55-59.
Burack, B. A critical analysis of the theory, method, and limitations of the “lie detector,” Journal of Criminal Law, Criminology and Police Science (1955) 46:414-426.
Davidson, P. O. Validity of the guilty-knowledge technique: The effects of motivation, Journal of Applied Psychology (1968) 52(1):62-65.
Diaconis, P. and Efron, B. Computer-intensive methods in statistics, Scientific American (1983) 248(5):116-130.
Efron, B. and Tibshirani, R. Statistical data analysis in the computer age, Science (1991) 253: 390-395.
Elaad, E. Detection of guilty knowledge in real-life criminal investigations, Journal of Applied Psychology (1990) 75(5):521-529.
Elaad, E., Ginton, A., and Jungman, N. Detection measures in real-life criminal guilty knowledge tests, Journal of Applied Psychology (1992) 77(5):757-767.
Forman, R. F. and McCauley, C. Validity of the positive control polygraph test using the field practice model, Journal of Applied Psychology (1986) 71(4):691-698.
Furedy, J. J. The “control” question “test” (CQT) polygrapher’s dilemma: Logico-ethical considerations for psychophysiological practitioners and researchers, International Journal of Psychophysiology (1993) 15:263-267.
Furedy, J. J. and Heslegrave, R. J. The forensic use of the polygraph: A psychophysiological analysis of current trends and future prospects. In: Advances in Psychophysiology, Volume 4. J. R. Jennings, P. K. Ackles, and M. G. H. Coles, eds. Jessica Kingsley, London, 1991, pp. 157-189.
Iacono, W. G., Boisvenu, G. A., and Fleming, J. A. Effects of diazepam and methylphenidate on the electrodermal detection of guilty knowledge, Journal of Applied Psychology (1984) 69(2):289-299.
Iacono, W. G., Cerri, A. M., Patrick, C. J., and Fleming, J. A. E. Use of antianxiety drugs as countermeasures in the detection of guilty knowledge, Journal of Applied Psychology (1992) 77(1):60-64.
Iacono, W. G. and Patrick, C. J. What psychologists should know about lie detection. In: Handbook of Forensic Psychology. I. B. Weiner and A. K. Hess, eds. Wiley, New York, 1987, pp 460-489.
Kircher, J. C. and Raskin, D. C. Polygraph techniques: History, controversies, and prospects. In: Psychology and Social Policy. P. Suedfeld and P. E. Tetlock, eds. Hemisphere, New York, 1992, pp. 295-307.
Kleinmuntz, B. and Szucko, J. J. On the fallibility of lie detection, Law and Society Review (1982) 17(1):85-104.
Lykken, D. T. The GSR in the detection of guilt, Journal of Applied Psychology (1959) 43(6):385-388.
Lykken, D. T. Psychology and the lie detector industry, American Psychologist (1974) 29(10):725-739.
Lykken, D. T. A Tremor in the Blood: Uses and Abuses of the Lie Detector. McGraw-Hill, New York, 1981.
Lykken, D. T. The case against polygraph testing. In: The Polygraph Test: Lies, Truth and Science. A. Gale, ed. Sage, London, 1988a pp. 111-125.
Lykken, D. T. Detection of guilty knowledge: A comment on Forman and McCauley, Journal of Applied Psychology (1988b) 73(2):303-304.
Lykken, D. T. The lie detector controversy: An alternative solution. In: Advances in Psychophysiology, Volume 4. J. R. Jennings, P. K. Ackles, and M. G. H. Coles, eds. Jessica Kingsley, London, 1991, PP 209-214.
Lykken, D. T. Controversy: The fight-or-flight response in homo scientificus. In: Psychology and Social Policy. P. Suedfeld and P. E. Tetlock, eds. Hemisphere, New York, 1992, pp. 309-325.
Podlesny, J. A. Is the guilty knowledge polygraph technique applicable in criminal investigations? A review of FBI case records, Crime Laboratory Digest (1993) 20(3):57-61.
Podlesny, J. A. and Raskin, D. C. Effectiveness of techniques and physiological measures in the detection of deception, Psychophysiology (1978) 15(4):344-359.
Raskin, D. C. Does science support polygraph testing? In: The Polygraph Test: Lies, Truth and Science. A. Gale, ed. Sage, London, 1988, pp. 96-110.
Raskin, D. C. Polygraph techniques for the detection of deception. In: Psychological Methods in Criminal Investigation and Evidence. D. C. Raskin, ed. Springer, New York, 1989, pp. 247-296.
Raskin, D. C. and Kircher, J. C. Comments on Furedy and Heslegrave: Misconceptions, misdescriptions, and misdirections. In: Advances in Psychophysiology, Volume 4. J. R. Jennings, P. K. Ackles, and M. G. H. Coles, eds. Jessica Kingsley, London, 1991, pp. 215-223.
Rosenfeld, J. P. Alternative views of Bashore and Rapp’s (1993) alternatives to traditional polygraphy: A critique, Psychological Bulletin (1995) 117(1):159-166.
Simon, J. L. and Bruce, P. Resampling: A tool for everyday statistical work, Chance (1991) 4(1): 22-32.