| [Back] | [Index] | [Next] |
Chapter 6
Factors Affecting Polygraph Examination Validity
INTRODUCTION
The analyses of both field and analog studiesreported in chapters 4 and 5 indicate that thereis considerable variability in accuracy rates ofpolygraph examinations. To interpret these variations,numerous factors, such as the restrictedrange of techniques and applications tested inthese studies, need to be considered. In addition,researchers have attempted to explain the variabilityin accuracy scores by proposing a numberof factors that theoretically may affect polygraphtest validity. These include characteristics of examiners,settings, and subjects. In addition, subjects have been known to use, or might be trainedto use, a number of countermeasures to beat"the polygraph. For many of these factors the researchevidence is contradictory. For others, therehas been little or no empirical testing. This chapterdescribes evidence from field and analog studies,as well as from laboratory investigations, on factorsthat may affect the accuracy of polygraphtests. The chapter also discusses possible prioritiesfor additional research on factors affecting polygraphvalidity.POLYGRAPH EXAMINER, SUBJECT, AND SETTING
The previously described analyses of field andanalog studies (see chs. 4 and 5) emphasize thecharacteristics of polygraph tests and their relationto accurate or inaccurate outcomes. In thepresent section, the focus shifts away from thetests themselves, to additional factors that mayaffect validity. These factors are sometimes referredto as dimensions of external validity andaid in the assessment of the generalizability ofresearch findings. Considerations of these factorswill enable evacuation of the conditions underwhich various levels of validity may be expectedfrom polygraph examinations. Differential validityin polygraph tests may be obtained with differentexaminers, subject populations, and withexaminations conducted in different settings.Examiner
It has long been recognized (cf. 108,122,135,154) that the examiners skill has an important effecton the validity of polygraph tests. Examinerexperience is an essential element reported by investigatorsand has often been used to explain differences in accuracy rates (137,138). There aresome data to indicate that experienced examinershave better accuracy rates. In recognition of thisoutcome, training has been accorded a high priorityboth within and outside Government agencieswhich conduct polygraph examinations and bypolygraph examiner groups (cf. 3). An extensivearray of training facilities now exists, offering asomewhat diverse set of orientations to polygraphtesting.
Experience
A number of studies have tested how examinerexperience relates to validity of polygraph examinations.Horvath and Reid (84), for example,had charts utilized in their validity study reexaminedby a group of 10 polygraph examiners.Seven of the examiners were experienced and threeof them were examiner-interns (each with less than6 months experience). According to Horvath andReid, experienced examiners made an average of91.4 percent correct judgments, while the averagefor inexperienced examiners was 77.5 percent.
Training
Experience in conducting polygraph examinationssuggests that there are a number of clinicalcomponents to detection of deception. To someextent, training programs capture these clinicalelements by extensive training in proper" examinerattitude and relationship with subjects. Increasingly,however, training programs emphasizestandardized techniques for constructing questionsand scoring examinations. In this respect, the U.S.Army Military Police School (USAMPS) is perhapsthe best example. The school serves as thecentral training site for almost all Governmentagencies which maintain polygraph examinerstaffs. USAMPS teaches several versions of thecontrol question technique (CQT) (including whatthey call the modified general question technique(MGQT) and the original Backsters zone of comparison(ZOC) method) and several specific protocolsfor selecting question sets and scoringpolygraph charts. Trainees receive both didacticclassroom training and supervised experience conductingpolygraph examinations. The current curriculumfor USAMPS uses Reid and Inbaus (139)text on polygraph testing, supplemented by materialsprepared especially for its trainees (179).USAMPS is one of a number of training programscertified by the American Polygraph Association(cf. 3).
On the basis of presently available data, it isnot possible to determine whether types of traininghave an effect on outcomes. A study by Raskin(133) indicates that examiners trained inschools that emphasize numerical scoring weresignificantly more accurate than examiners whoattended other schools (97.1 v. 86.9 percent). Itis difficult to determine, however, if training innumerical scoring is more efficient or if better examiners/schools select such techniques. The factthat examiners who were trained in numericaltechniques, but who did not use them, did morepoorly than examiners trained in numerical techniqueswho used them (88.5 v. 98.9 percent) suggeststhat numerical evaluation rather than examinerselection (or some other aspect of the training)provides an advantage.
Subjects
Much effort in recent years has been devotedto development of systematic training. Less attentionappears to have been paid to the characteristicsof subjects of polygraph testing. Frequently,research reports of polygraph examination donot report even the most easily available data onsubject characteristics (e.g., proportion of malesand females). There have, however, been a numberof studies of specific population groups (e.g.,psychopaths) hypothesized to be less detectable.In addition to subjects psychopathy, other diagnosticcategories and subject variables such asgender, intelligence, motivation, and responsivityto arousal may also affect validity.
Subject factors are often described in the literatureas personality or individual difference factors(136,194). They refer to traits associated withindividuals that may make them differentially detectablein a polygraph examination. Understandingthese effects should enable determination ofthe conditions under which polygraph testing willyield particular levels of validity. The mechanismby which subject variables affect polygraph examinationvalidity has to do with differentialautonomic arousal. Validity is affected when aninteraction results between arousal and polygraphtesting.
Psychopathy and Level of Socialization
One aspect of potential subject effects that hasreceived considerable attention is the effect of levelof socialization and psychopathy on detectability.In a series of studies by Waid and his colleagues(193,198,199) significant relationshipswere found in the laboratory between socializationand autonomic responsiveness. An initialfinding (193) was that college students who scoredlow on socialization (on a standard psychologicalinventory), gave smaller electrodermal responses(EDRs) to stimuli than did high scoring subjects.In a more directly relevant investigation (198), agroup of college students was asked to deceive ornot to deceive a professional polygraph examiner.Results indicated that subjects who were notdetectable were significantly less socialized thanthose who were detectable. Susceptibility to detectionseemed to be mediated by socialization;results indicated that low socialization subjectsshowed reduced EDRs. Highly socialized subjectswere more responsive electrodermally, and as aresult, several of them were misclassified asdeceptive.
Raskin (136) has criticized Waid, et al.s (198),research as not having practical significance forevaluations of polygraph validity. According toRaskin, simply demonstrating that there is a differencein responsivity on the first set of questionsdoes not mean that subjects would not be correctlydetected in an actual polygraph examination(which may involve three to four charts). Someof Raskins own studies (e.g., 21,137) suggest thatpsychopathic individuals are not less detectablethan nonpsychopathic individuals. In Raskin andHares study, convicted felons, half of whom werediagnosed as psychopathic, performed a mockcrime. These subjects were then administered apolygraph examination and offered a substantialmonetary bonus if they could produce a truthfuloutcome. In contrast to Waid, et al. s, findings,Raskin and Hare found that individuals diagnosedas psychopathic and/or low in socialization weremore reactive and easily detectable than those notpsychopathic and high in socialization. Earlierresearch by Raskin (21) supports this finding.Barland and Raskins (22) field study, on the otherhand, found that subjects who scored high on thepsychopathic deviate (Pal) scale of the MinnesotaMultiphasic Personality Inventory (MMPI) (ameasure of psychopathy) had smaller cardio (butnot respiration or skin conductance) scores thanlow Pd subjects.
In a previously described study, Balloun andHolmes (12) conducted an analog study of collegestudents using a cheating" situation. Theirresults indicated that subjects who scored high onthe Pd scale of the MMPI were just as easy todetect as were those individuals who scored lowon the scale. It is important to note, however, thatthe polygraph test was a concealed informationtype of test, not a CQT or relevant/irrelevant(R/1) test. A doctoral dissertation by Hammond(64a) also found no differences between normaland psychopaths.
Other Psychopathology
Guilty psychopaths may escape detection becausethey are not concerned enough about a misdeedto create interpretable physiological responses.Individuals with other forms of psychopathologymay escape detection or be classifiedas false positives for other reasons (e.g., emotionalinstability, delusional thinking). The one studythat has investigated this possibility (74) found,in fact, that innocent neurotics and particularlypsychotics were likely to be identified as deceptive.There were no guilty subjects in this realcrime" analog study.
Gender
One of the most obvious subject differences isgender. Males and females may have different patternsof autonomic arousal, and such differencesmay affect polygraph testing validity (136,194).Unfortunately, few data exist to examine thishypothesis; most research only studies male subjects.The one study by Cutrow, et. al. (45), thatspecifically tested for sex differences did not findany. In another study (61), all female subjectswere tested in a mock-crime situation using theguilty knowledge test (GKT). GKT was found tobe highly accurate, but because males were notalso tested, it is impossible to determine if maleswould have been less detectable. The two Hontsand Hodes (76,77) analog studies described inchapter 5 included both females and males; theresearchers do not report any gender differencesin detectability.
Intelligence
Intelligence is an additional variable which potentiallymight affect detectability. The ability ofintelligent subjects to anticipate questions may affectpolygraph accuracy. One possibility is thatintelligent subjects are less detectable because, iftrained, they are able to anticipate questions andemploy countermeasures. Another possibility isthat because intelligent subjects better understandthe implications of a polygraph examination, theywill respond to relevant questions with heightenedarousal when they are attempting to deceive (2o).
There has been relatively little research on intelligenceand polygraph testing. In one of the fewexperiments which look at intelligence and detectability,Kugelmass (95) found no correlation betweenintelligence and responsivity on a peak oftension (POT) card test. On the other hand, researchby Gustafson and Orne (65) found that motivationto deceive increased the probability ofdetection. Barland and Raskin (20) feel this isevidence of the potential role of intelligence.Barland and Raskins study (22) which compareddetection rates among subjects of different educationlevels, found no difference. However, a separateanalysis of the sources of false positive errorsby Raskin (133) found that the majority offalse positives occurred among subjects who hadcollege degrees. Level of education, of course, isnot a perfect indicator of intelligence, and thereis a need to better understand these relationships.
Ethnic and Group Differences
Another category of subject differences thatmay affect polygraph validity has to do withethnic and group differences in physiological response.Research conducted cross culturally (e.g.,97,104,158), indicates that there are ethnic differencesin response to stress. Such differencesmay, in turn, affect detection of deception. Asnoted earlier, these effects may interact with theethnic identification of the examiner. However,effects of ethnic differences have not been directlytested with respect to polygraph examinations.
Autonomic Lability
A final individual difference is what Waid andOrne (194) have referred to as autonomic lability.Regardless of other differences among subjects,there may be consistent individual differences connectedwith their level of autonomic arousal.
Although there is considerable variance for anindividual in autonomic responses to most physiologicalmeasures of autonomic nervous system(ANS) arousal, electrodermal lability maybe different.Given the importance of the EDR for polygraphexaminations, it maybe essential to understandmore about this factor. Unfortunately, mostof this research (e. g., 200) has been conductedwith concealed information tests and not withCQT or R/I tests.
Setting
One theory underlying lie detection using thepolygraph is that the threat of punishment leadsan individual to manifest a physiological reaction(48). This suggests, then, that settings in whichan individual is more certain of being detected andin which the consequences are greatest, will permithigher levels of detection. Furthermore, inorder to be certain of being detected, a subjectmust believe in the efficacy of the polygraph proceduresin order for it to function. According tosome (e.g., 194), the polygraph is often usedsomewhat like a stage prop," and its presenceis meant to enhance the subjects concern."Stimulation tests, used in almost all fieldpolygraph examinations, serve the same function,albeit more directly. There is considerable discussion(e. g., 202) in the literature about how frequentlywithin a polygraph examination suchstimulation tests should be utilized in order to increasethe validity of the examination.
Instrument
Some research, reported by Orne and his colleagues,addresses the question of the situationalfeatures necessary for a polygraph examination.In one component of a study reported by Orne,et al. (123), subjects were led to believe that thepolygraph recording equipment was not operative.There was some indication that the pretestcondition in which subjects were led to believethat the polygraph instrument was inoperativeproduced a lower detectability; however, resultswere not statistically significant. In an earlierstudy (161), detectability was not affected by subjectsbelief in whether the machine was recording.Both of these studies involved use of concealedinformation tests.
A more recent study by Ornes group (198)tested a similar hypothesis using a different procedure.In this study, subjects saw the polygraphmachine turned off, although the experimentersactually ran the leads to a second polygraph deviceand were able to record responses during apretest review of questions. The results indicatedthat subjects who were aware of being recordedhad significantly higher responses to relevantquestions and not significantly different responsesto control questions.
Bogus Pipeline
An interesting and potentially important aspectof how the polygraph achieves valid results isbased on what social psychologists such as Jonesand Sigall call the bogus pipeline" (87). Thebogus pipeline is a procedure used to elicit truthfulattitudes in situations where social desirability effects(i.e., subjects desire to express socially acceptableopinions) may mask actual attitudes. Theprocedure involves attaching subjects (via skinelectrodes) to an ostensible physiological recordingdevice called the electromyograph" (EMG)and providing subjects with a steering wheel"device to record their attitudes. In a typical study(87), subjects were told that the EMG measuredimplicit muscle potentials and that it was an improvedpolygraph or lie detector." The recordingdevice is actually electrical junk" (87), and thepurpose of the procedure is simply to convincesubjects that their actual attitudes are detectable.
Results from a number of investigations whichhave used the bogus pipeline procedure (e.g.,131,150) support Jones and Sigalls premise. Severalstudies indicate that when subjects believethat their attitudes are detectable by a physiologicalrecording device, they more readily expresstheir actual attitudes. Although it is difficult toknow what actual" attitudes are, higher truthfulnessis assumed with the bogus pipeline becausethe procedure yields more socially undesirable responsesthan when it is not used. For example,in Sigall and Pages (150) initial experiment, theyfound that subjects in the bogus pipeline conditionwould admit to negative attitudes aboutNegroes." Similar subjects in nonbogus pipelineconditions using paper-and-pencil tests would notreveal such attitudes. Later research has shownthat this findings holds for attitudes toward handicappedindividuals and for confessing" to havingprior knowledge about a psychological experiment.
Although the bogus pipeline research suggeststhat the conditions of testing (in particular, theperceived complexity and accuracy of equipment)may have important effects on polygraph subjects,it is not clear how or to what extent theseeffects influence the validity of the test itself. Ina substantial number of criminal investigationssubjects voluntarily confess after having the polygraphprocedure explained or being shown the resultsof the examination. In personnel screening,subjects often admit to errors in their job applicationor past indiscretions (24,165). Most availablefield and analog research does not permit determinationof the extent to which the polygraphis used in this way.
Specific Settings
Polygraph examinations take place in a numberof settings, ranging from facilities specifically designedfor this purpose to motel rooms. Specificallydesigned facilities usually include one-way mirrorsfor observation and audio recording capabilities,and are located so as to prevent interruptionsduring the examination. It is reasonable toassume that the setting may interact both withsubject and examiner characteristics to affect thevalidity of polygraph tests. No research, however,directly tests the impact of different settings onpolygraph testing validity.
COUNTERMEASURES
Countermeasures are deliberate techniques used by deceptive subjects to avoid detection during a polygraph examination (23,108,139,194,195).Countermeasures can range from simple physical techniques, to so-called mental countermeasures,to the use of drugs and biofeedback techniques. There is a potentially large list of such countermeasures,and there are a number of plausible, butnot yet validated, techniques to avoid deception. The research on polygraph countermeasuresis summarized below by type of countermeasure.Physical
Physical measures taken by a subject during apolygraph examination are, perhaps, the most frequently discussed countermeasures used by subjects (20,108). Any physical activity which couldaffect physiological response is a potential problemfor interpretation of a polygraph test record.There is no question that physical measures, fromtensing muscles to biting the tongue, to squeezingtoes, to shifting ones position can affectphysiological response.
There are frequent references to the use of suchmeasures (see e.g., 40,108). But little systematicresearch has been conducted to establish the impactof the use of such measures on polygraphdecisions. Kubis (93) found that when subjectspress their toes towards the floor they were ableto reduce the probability of detection from 75 to10 percent. A replication of this experiment, however,by More (119) found that there was no decreasein detectability caused by toe movements.In two more recent studies discussed in chapter5, by Honts and Hodes (76,77), the efficacy oftwo physical countermeasures was tested in analogsituations. Both studies found that countermeasuresallowed subjects to beat" the polygraph.Furthermore, there were a large percentageof inconclusive. In addition, both studiesfound that experienced examiners were not ableto detect use of the countermeasures. A recentstudy by Honts, Raskin, and Kircher (78) alsofound that the use of physical countermeasuresdecreased detectability; the false negative rate forcountermeasure subjects was 78 percent. However,examiners using a separate EMG analysiswere abic to detect the use of countermeasures 80percent of the time.
Thus, the evidence, while limited, is that deceptivesubjects who use physical countermeasuresand who can distinguish nonrelevant from relevantquestions (in a CQT or R/I test) can increasetheir chances of avoiding detection.
Drugs
In contrast to physical measures, which potentiallymay be detected by an observant polygraphexaminer by running multiple charts or by carefulcomparison of particular physiological measures,the use of various pharmacological agents is probablymore difficult to detect. Not only may drugsbe difficult to detect by observation, but they mayalso not be detected by multiple polygraph tests.Some theorists have suggested that such pharmacological agents have the potential to produce incorrector uninterpretable polygraph records.
Research on drugs factors is only beginning tobe conducted. Recent research by Waid (197) indicatesthat the tranquilizer, meprobamate ("Miltown"),permits subjects who are being deceptiveto increase their ability to avoid detection in apolygraph examination. One feature of tranquilizerssuch as meprobamate is that they suppressautonomic activity which may not be accompaniedby any observable psychom*otor differences.In Waid, et al.s, study a GKT was used in a polygraphtest. Subjects were all male and divided intothree groups: 1) a tranquilizer group; 2) a placebogroup; and 3) a control group. Only 3 of 11 guiltysubjects who had taken meprobamate werescored as deceptive.
It should be noted that because Waid, et al.s,investigation involved GKT, the ability togeneralize from the results is limited. Accordingto Raskin (136), a different problem would be encounteredby attempts to utilize tranquilizers todefeat an examination employing CQT. The useof such drugs in a CQT polygraph examinationwould be more likely to yield inconclusive findings,rather than errors, because the drugs wouldlikely result in no difference between the responsesto control and relevant questions. This interpretationis supported by the recent analog study ofGatchel, et al. (59), which found that the use ofpropranolol, a beta-blocking drug, resulted in a32.2-percent inconclusive rate, although the overallerror rate was low. An additional finding wasthat examiners could not tell which subjects hadused the drug. Conclusions drawn from this studymust be limited by the fact that subjects were students.Other studies using college students (e.g.,76,77) have also resulted in large numbers of inconclusive.
A recent study by Iacono, et al. (86), found thatingestion of neither 10 milligrams of diazepam("Valium") nor 20 milligrams of methylophenidate(Ritalin) affected the accuracy of detection.Results in both active drug conditions were moreaccurate than when subjects ingested a placebo(a capsule containing lactose).
Research on other psychoactive drugs has notbeen reported in the literature, although suchresearch is now being planned under the auspicesof the National Security Agency and the ArmyIntelligence and Security Command. There arealso no data as to the use of common drugs byactual polygraph examination subjects. Althoughexaminers normally ask subjects to report use ofmedications or other drugs, blood samples orother detection means are typically not employed.It is thus difficult to assess the magnitude of druguse by subjects in previous research on the validityof polygraph testing.
In addition to drugs, there have also been reportsof the use of various chemicals to confusephysiological recording (see 20). Placing antiperspirantpowder, clear nail polish, or other agentson the balls of ones fingers may make EDRs lessreliable. Such measures, however, should be detectableby a trained examiner.
Hypnosis/Biofeedback
There is a substantial literature in psychologyabout the use of hypnosis and biofeedback to alterand condition physiological responses. There issome evidence (see 146) that hypnosis, for example,induces declines in skin conductance levels.A number of investigations have attempted toshow that hypnotically suggested amnesia is aneffective countermeasure. Such research seems toindicate that hypnosis is not effective (see 20).
Recent research by Corcoran, Lewis, and Garver(42) has examined the effects of biofeedbacktraining on suppressing EDR. They found thatboth hypnosis and biofeedback groups were ableto reduce detectability after training as comparedto a control group. In another study, Rovner,Raskin, and Kircher (143) reported that subjectswho received extensive information about the natureof lie detection and practiced using countermeasureswere detected significantly less than subjectswithout such training. It seems clear that ifhypnosis or biofeedback operate as countermeasures,especially with commonly used tests suchas CQT, that extensive training would have toaccompany their use.
Mental
Another category of countermeasures involvesthose that get the subject to think differently aboutthe test. As noted earlier, most polygraph examinationsrely on the subjects motivation to avoiddetection rather than on any response directlyconnected with telling a lie." Simple cognitivecountermeasures include patterns of thinking thatsuppress responses to control or irrelevant questions.More complex cognitive countermeasuresare based on knowledge of the results of the examination and lack of belief in one's detectability.
Controlling Thoughts
Any individual who understands the basicstructure of a particular polygraph examinationshould be able to differentiate irrelevant and controlfrom relevant questions (when an R/I or CQTexamination is conducted). Given that when aCQT is employed, the questions are reviewedprior to their presentation, a subject intent on deceptioncould try to alter cognitively responsesto various questions (although, since the order ofpresentation is varied, this is made difficult duringan actual test), The procedure would be totry to dissociate oneself from the relevant questionsand heighten response to control questions.Various means of such mental dissociation havebeen described (cf. 20,108). They range from concentratingon an irrelevant object or by convincingoneself that the question means something differentthan intended.
There is little research evidence available to ascertainthe potential of these techniques. Only onestudy on the use of such cognitive countermeasuresappears to have been conducted, and it doesnot provide much information. In this study,Dawson (49) used as subjects actors trained in themethod" school of acting. Subjects were instructedto try to defeat the polygraph in a CQTexamination. The results indicated that withoutexception the actors were detected and were notable to control their affective responses. It shouldbe noted, however, that the actors in Dawsonsstudy were merely told to appear innocent" andwere not told what would make them appear non-deceptiveon a polygraph test.
Knowledge of Results
Another set of countermeasures is based on subjectshaving knowledge of the results of their polygraph examination. In criminal situations, particularlywhen an inconclusive outcome is obtainedor when the subject disagrees with the outcomeof an initial polygraph test, subjects are retested.In noncriminal situations, individuals are oftentested at the beginning of their employment andat a number of subsequent points. In addition,subjects who know that they will receive a polygraphexamination may seek training in methodsto avoid detection. For all of these reasons, it isimportant to understand how feedback aboutpolygraph examinations affects validity andwhether prior experiences represent a potentialcountermeasure.
In an early laboratory study (67), subjects weregiven a stimulation test and feedback concerningits outcome. Feedback was manipulated so thatsome subjects thought they had successfullyavoided detection and others thought that theyhad not. Subjects were motivated on a subsequenttrial to avoid detection (they were told, onlymature and stable individuals are able to fool thelie detector"). The results indicated that subjectswho believed that they had avoided detectionwere much less detectable on the second trial (13out of 16 were not detected, while only 1 of 16in a control group were able to avoid detection).It should be noted that a stimulation test is a formof a concealed information test and the result maybe due to lowered overall arousal. Beating" aCQT represents a somewhat different problem.
A recent study, by Rovner, et al. (143), testeda similar hypothesis in a CQT examination. Severalgroups of subjects were placed in a mockcrime situation. One group was given informationabout the nature of a CQT examination andinformation on what physiological reactions theyshould try to simulate. Another group was giveninformation plus two practice tests involving actualphysiological recordings after which theywere told whether or not they had beat the polygraph.A third group served as a control and wasgiven a typical polygraph examination. The resultsindicated that the information only and controlgroup were not able to avoid detection; however, 25 percent of the guilty subjects in the informationplus practice group were able to avoiddetection. Raskin (136) maintains that this 25percenterror rate should be considered the upperlimit" because, in actual field situations, motivationwould be much higher. Although Raskin is,perhaps, correct, it is also possible that in actualsituations (where motivation is high), subjectsmight engage in more practice.
Belief in Machine"
A final countermeasure is based on researchabout the bogus pipeline (87) and the role of thesetting in inducing valid outcomes. If the validityof polygraph testing is dependent on the beliefby subjects in the efficacy of the procedure, thena possible countermeasure would involve trainingsubjects to believe that the polygraph does notwork. This might be done, for example, by providingsubjects with false feedback on a polygraphexamination. Unfortunately there is little researchin this area, and the two studies that have beenconducted come to different conclusions about theeffect of belief in the techniques effectiveness. Inone study, Bradley and Janisse (35) tested thehypothesis by rigging a stimulation test at variouslevels of effective detection. Depending on thecondition, subjects were detected" on one, two,or three trials to create the impression that thedetection measures were ineffective, sometimes effective,or perfectly effective. For the EDR measure,the more effective the apparatus appeared tobe, the more the innocent subjects scored as nondeceptiveand the more the guilty subjects scoredas deceptive. In an earlier study, however, Timm(162) found that feedback about the techniqueseffectiveness had no effect on whether subjectsdeceptiveness or nondeceptiveness could be detected.The theoretical support provided by researchon the bogus pipeline indicates that subjectsbelief in the technique may be important,and that additional research is needed to assessthe effects of belief in the machine on actual polygraphtests.
RESEARCH IMPLICATIONS OF FACTORS AFFECTING VALIDITY
If further research on polygraph testing is carriedout, a number of research priorities can beidentified on the basis of the present analysis.These priorities include research on the theory ofpolygraph testing, polygraph techniques, countermeasures,comparison with other techniques,and field-based studies.Theory
Polygraph testing is premised on the belief thatlying produces reliable physiological reactions.Testing the efficacy of this assumption is an importantresearch need, Basic research could examinethe physiological reactions to differenttypes of lies and under different conditions ofarousal.
Scoring
Research is currently being conducted by theU.S. Army on development of computer scoringsystems and more reliable measures of physiologicalarousal. There is some evidence (e. g., 92) thatthe validity of polygraph examination decisionsis improved if the clinical judgment of examinersis removed (see also, 27) and related evidence thatnumerical scoring is more accurate (91,133) thannonnumerical scoring, Research should proceedon developing analogs to digital scoring systems.Such research, however, would not address theimpact of examiner-examinee interaction, and thisarea also needs further study.
Question Techniques
Another research priority is to develop additionalprotocols for question development.Perhaps the most important research need in thisregard is to develop and field-test the concealedinformation test. Basic research and theory (see,e.g., 27,108, 136) suggests that such examinationshave the highest likelihood of detecting deception,although no field research has yet been conductedto examine their use. Such research should bothestablish means of constructing GKTs and theirvalidity in actual use.
An additional priority is to develop and testquestion techniques that may be employed inscreening situations. Some examiners for exampleclaim to use a version of CQT for screeningexaminations (see ch. 2). This application of CQThas not been subjected to scientific tests of validity.In addition, efforts should be devoted to testingthe use of CQT with different subject groupsand in national security investigations.
Countermeasures
If polygraph testing is to be more widely employedin national security investigations, thereis an urgent need for research on countermeasures.Particular priorities would be research on drugs,biofeedback training, and subject gullibility, andmotivation. Such research needs to be carried outboth in field situations and in the laboratory.
There are a number of drugs that are suspectedof lowering ANS arousal and that theoreticallymay be able to invalidate the results of a polygraphexamination or compel an inconclusive"finding. A first priority is to extend Waid, et al.s(197), research on meprobamate (which reduceddetectability) to other psychoactive drugs.
Biofeedback training, as well as other forms oftraining have not been investigated, yet their effectson polygraph examinations may be substantial.Subjects beliefs about the accuracy of thepolygraph may also be critical. As suggested bythe research on the bogus pipeline," individualswho believe their underlying thoughts are detectableare more likely to provide truthful responses.The reverse phenomenon seems feasible and itwould seem possible to train individuals to believethat the polygraph is ineffective. Such trainingmight be accomplished by providing individualswith false feedback on the polygraph as well asby specific instructions during simulated polygraphexaminations. Similarly, subjects who canbe easily trained to beat the polygraph may bemore desirable as intelligence agents.
Comparison With Other Techniques
Only one study in the available literature (207)systematically compares the polygraph with otherinvestigatory tools. There is a need to examinewhether the polygraph provides independent orcorroborative evidence and whether the judgmentsmade by polygraph examiners are merelya function of their clinical judgment as investigators,or whether it is a function of the polygraphexamination itself.
A complication with this research is that thepolygraph functions, in many situations, as athreat. Individuals fear of taking the examination,in many instances, may lead them to confess orprovide incriminating evidence. The threat potential,however, is in part a function of theirs andothers knowledge of research results. If, for example,it became widely known that the polygraphwas beatable, it is likely that this threatwould be reduced and, hence, the validity (andutility) of the polygraph would be reduced.
Field Studies
As described in chapters 3, 4, and 5, there arenumerous problems with the available field andanalog evidence. Field studies are problematic because they can only only be conducted where anindependent criterion of guilt or innocence isavailable. As such, these studies may representa select sample of cases (e. g., where guilt is overwhelming)and a select set of examiners. Analogstudies have a different set of problems and havenot adequately motivated subjects or may nothave the appropriate number of cases. What isneeded is research which deals with the problemsof the available field and analog studies.
One theoretical" solution to the problem ofconducting systematic field studies is to conductABSCAM"-like investigations using bogus unauthorizeddisclosures (instead of bribes) thatwould enable investigators to set up situationswhere they have knowledge of who is guilty orinnocent. The polygraph could be used to selectguilty from innocent with a known base rate andground truth. Such methods, of course, raise anumber of ethical, legal, and pragmatic questions,and it is not clear whether they could provide definitiveanswers. They could not be used frequentlynor with a wide range of techniques/situations.Conducting polygraph research presents seriousconceptual and methodological problems; in theabsence of such research, however, it will not bepossible to develop fully an assessment of the validityof polygraph examinations.
CONCLUSIONS
The description in this chapter of factors affectingvalidity and potential countermeasures suggeststhat there is a great deal more to understandabout polygraph tests before one can be assuredof their validity. Despite our lack of full understanding,however, several factors that affectvalidity are known. In part, the history of polygraphdevelopment over the past 15 to 20 yearshas been to systematize and improve polygraphtesting procedures based on these factors. Onecentral problem, not adequately addressed byeither the literature on improvements in validityor countermeasures, is the extent to which thesefactors affect false negative and positive error ratesor affect numbers of inconclusive. For policy purposes,clearly such distinctions and a sense of themagnitude of false decisions is needed. Substantialresearch, beyond what is currently available,would have to be conducted in order to answersuch questions.| [Back] | [Index] | [Next] |
