Utah Approach To Comparison Question Polygraph Testing

Handler & Nelsondisplayed in front of the test subject and thetest subject is instructed to deny picking anynumber between 1 and 7 while the polygraphrecords his or her physiological reactions. Theacquaintance test allows the examiner toensure the production of adequate qualityrecordings and to take corrective actions toremedy any lack thereof.the behavioral issue of concern providesinformation useful to the referring authority.Effectivelyformulatedrelevantquestions will directly assess the test subject'sbehavioral involvement in the issue ofconcern.Relevant questions should notintroduce confusion through the use oflanguageorconceptspertainingtopsychological motivation or intent, as theseare thought to introduce dimensions of excuseor rationalization on the part of test subjectsor skillful liars. Conversely, truthful testsubjects may produce spurious reactionsbecause of the ambiguity and lack ofconcreteness of such ucture are the best approach. Relevantquestions should be free of idiomatic and legaljargon that is unfamiliar to the test subject,and should not include issues of psychologicalassessment or inference. Relevant questionsare simple questions that can easily beanswered “Yes” or “No.” Reluctance, on thepart of the test subject to provide a simpleanswer to a simple question may be anindicator of a non-testable issue or a testsubject who is unable to disambiguate theissue.Discussion and resolution of thisshould be non-accusatory, but persistentenough to achieve a simple testable answer toa question that is behaviorally descriptive ofthe test subject's possible involvement in anissue of concern. The prevailing practicepreference for relevant questions is they areusually answered “No,” though certainexceptions have been suggested such asalleged victims of severe sexual assaults(Hardy & Murphy, 1996). The current authorsfound nothing to support that using “Yes”answered relevant questions would invalidatea test.Theexaminercanusetheacquaintance test during the question reviewto demonstrate to the test subject that he orshe is a suitable candidate for polygraph, andprovide assurances that successful completionof the examination can be obtained byanswering all of the test questions truthfully(in the PLC version of the examination) or thatclear indication was found when the testsubject was not answering truthfully (in theDLC version).Following the acquaintance test thetest questions are reviewed with the testsubject for clarity. Some agency or localtesting protocols may specify the testquestions be fully reviewed prior to attachingany components to the test subject. There isno theoretical rationale to suggest thisdifference would invalidate an examinationresult. Attaching the sensors earlier may allowthem to stabilize, especially the electrodes forelectrodermal recording.The examiner begins with the sacrificerelevant question followed by the relevantquestions. The sacrifice-relevant question isused to introduce the relevant issue underinvestigation during the testing and is notscored. In investigative polygraph testing,relevant question targets are dictated by thecircumstances of the investigation and arecommonly formulated around the most salientor intense aspects of the allegation. Inscreeningprograms,relevantquestionsshould describe the test subject's involvementin possible behavioral concerns to riskmanagers or adjudicators and should bedesigned to add incremental validity to theirparticular program.Next the examiner introduces thecomparison questions. PLC questions arepresented to the test subject as beingnecessary for further evaluating the testsubject’s character and the issue underinvestigation. PLC questions are based ontransgressions whose subject matter isgenerally or conceptually related to theallegations of the examination and whichvirtually all persons may have committed, butwhich are likely to be denied in the context ofthe examination. PLC questions are broad inscope and usually based on actionsPolygraph screening targets wouldideally be selected to investigate content areaspertinent to actuarial or empirically derivedprotocols for risk assessment and riskmanagement. The fundamental requirementfor relevant question target selection is that17Polygraph, 2009, 38(1)

Utah Approach to CQTcategorically similar to that of the issue underinvestigation. That is, relevant questions ontheft would normally be associated withcomparison questions about theft or generalhonesty. Relevant questions about violent actsare typically associated with comparisonquestions about causing harm. Standardcomparison question construction, as taughtin polygraph schools accredited by theAmericanPolygraphAssociationandAmerican Association of Police Polygraphists,is recommended for ensuring saliency. Thereis no reason, however, to prohibit the use ofstandard “lie” comparison questions in nearlyany testing context.examiner choose to employ non-exclusionarytype comparison questions.As in other CQT techniques, the testsubject is strongly, but indirectly, discouragedfrom making admissions to PLC questions. Ifthe test subject makes an admission to a PLCquestion, the examiner notes that admissionwith some dismay, “Really, you did somethingthat would make me think you are a thief,” andeither minimizes the admission, “No, I am onlyconcerned about serious things,” or modifiesthe comparison question. An example of thelatter is: “Other than what you told me about,before this year did you ever lie to anyone whotrusted you?”Comparison questions in the UtahCQT are traditionally “exclusive” in that theyare separated from the relevant issue by time,place or category. Comparison questions notseparated from the relevant issue aresometimes referred to as non-exclusionarytype. Three studies (Horvath, 1988; Amsel,1999; Palmatier, 1991) failed to establish anyclearandconsistentadvantageofexclusionary comparison questions over nonexclusionary questions (Krapohl, Stern &Ryan, 2003).Podlesny & Raskin, (1978)showed some superiority for e (SCR) half-recovery time, SCRrecovery half time width and Skin PotentialResponse (SPR) amplitude were significantlymore effective with exclusive comparisonquestions.Note the italicized modifier precedingthe comparison question. The ultimate goal isto discourage admissions to PLC questions toensure that the test subject perceives them asambiguous and broad in nature. It is alsoimportant the examiner imply to the testsubject that lying to any of the relevant or PLCquestions will result in a failure of thepolygraph test and the conclusion of deceptionto the relevant issue under investigation.The examiner then introduces andreviews the neutral questions which providetime to return to a baseline when there isdistortion or a physiological reaction to aspecific question.Kircher, Kristjansson,Gardner, & Webb (2005) suggest ar response should be increasedto a minimum of 35 seconds to allow recovery,or a neutral question inserted. In general, thepreferred approach is to wait to allow a returnto, or at least toward, baseline levels. Theneutral questions should be non-emotional innature and are generally answered “Yes” toensure the test subject is paying attention tothe test questions. There is nothing tosuggest, however, that an exam in which anyneutral question is answered “No” would beinvalid. The examiner may review additionalneutral questions in case they are neededduring testing to re-establish a baselinetracing.Podlesny & Raskin (1978) alsoreported both types of comparison questionsproduced significant identification of innocenttest subjects, but only exclusive comparisonquestions produced significant identificationof guilty test subjects using numerical scores.Collectively these reports suggest thatexclusionary comparison questions may ionswhendataareevaluated using reaction criteria typicallyemployed in field testing (for descriptions ofthose reaction criteria, see: Bell et al., 1999;Handler, 2006; Raskin & Honts 2002; Kircheret al., 2005).The use of exclusionarycomparison questions may avoid possiblecriticism that the PLC questions are alsorelevant and may cause a false negativeresult. The current authors found nothing tosuggest a test would be invalid should anPolygraph, 2009, 38(1)The examiner next reviews theintroductory question that is similarly wordedto one of the “symptomatic” questions used inother CQT formats. The introductory questionattempts to assure the test subject that no18

Handler & Nelsonun-reviewed questions will be asked duringthe examination and may allow an orientingresponse at the beginning of an examination.Research by Honts, Amato & Gordon, (2004)has failed to demonstrate the symptomaticquestion functions as described and mayactually produce poorer accuracy, especiallyfor innocent test subjects.Example of a Utah PLT 3-question wordingFor an event-specific, single-issue testsurrounding a bank robbery occurring lastThursday, one might ask the followingquestions:Introductory 1: Do you understand I will onlyask you the questions we discussed?The consistent trend illustrated bythese investigators and others suggests thatthe invention and addition of new types ofquestions should not be encouraged in an ageof modern scientific polygraph testing unlessresearch shows the efficacy of a new approach(Hilliard, 1979).Sacrifice Relevant 2: Regarding whether ornot you robbed that bank do you intend toanswer all of these questions truthfully?Neutral 1: Are the lights turned on inside ofthis room right now?While it is wise for field examiners toadhere to the general principles andprocedures taught in basic training, there isequal or greater wisdom in adapting fieldpractices to conform to modern approacheswith proven validity. We do not believe thatminor departures from the above questionsequences would cause a test to be invalid,and the varying formulations of the Utah-CQTsince its emergence suggest that the scientistswho developed the Utah-CQT method did notseek validity through simplistic adherence to a“paint-by-numbers approach”, but soughtdemonstrablevaliditythroughtheconstruction of CQT methods according tosound testing principles.Comparison 1: (Before turning X), Did youever do anything that was dishonest orillegal?Three-Question FormatNeutral 3:Do you sometimes listen tomusic while riding in a car?Relevant 1:Did you rob that bank locatedat in Austin?Neutral 2:Are you now physically locatedwithin the State of Texas?Comparison 2: (Between the ages of X andY), Did you ever take anything that did notbelong to you?Relevant 2:Did you rob that bank locatedat in Austin last Thursday?The Utah-CQT has two versions, athree-question version and a four-questionversion (Raskin & Honts, 2002).Comparison 3: Did you ever take anythingfrom a place where you worked, (beforeage X)?The three-question version was thefirst designed and was primarily used forsingle-issue testing but can also be used formultiple-facet testing of a single knownallegation. The three-question version of theUtah-CQT allows a great degree of flexibility inrelevant question format.Relevant 3:?Did you rob that bank at onThe examples above are shown withthe exclusionary clause of the comparisonquestion in brackets.For a multiple-facet examination, theexaminer has a choice of asking two rewordedrelevant questions with the same meaning andanother relevant question that is directlyrelated to the issue under investigation. Thisthird relevant question can be an evidenceconnecting, guilty knowledge or secondaryinvolvement question.The following describes an example ofquestion numbering and type of question usedin the three-question version of the Utah-CQT.For a single-issue examination, there will bethree relevant questions, each slightlyreworded.19Polygraph, 2009, 38(1)

Utah Approach to CQTA third alternative is to ask threeseparate relevant questions relating to thesame specific issue under investigation.Readers are reminded that research hasshown that accuracy rates are higher for testsin which the test subject is either completelytruthful or deceptive to all of the testquestions as opposed to just some of them(Honts, Kircher, & Raskin, 1988; Raskin,Kircher, Honts, and Horowitz, 1988; Barland,Honts and Barger, 1989).Relevant 3:Did you participate in any wayin the robbery of that bank?Note that this example is provided withnon-exclusionary comparison questions.Four-Question FormatThe four-question format is similar indesign to a version of the Air Force ModifiedGeneral Question Technique (DoDPI 2006)using pairs of relevant questions that arebracketed by comparison questions. Thisallows the examiner greater flexibility coveringmore than one aspect of the relevant issuesand in scoring by using the surroundingcomparison questions. The relevant questionscan range from one to four distinct behavioralaspects or facets of a single crime orallegation. The question construction rules arethe same as those described above for themultiple-facet version of the three-questionversion.The current authors would stronglyrecommend examiners, if possible, attempt tolimit the examination to one in which the testsubject is truthful or deceptive to all of therelevant questions.If one were to construct a multiple-facetpolygraph examination surrounding a singlecrime event involving a bank robbery,examples of alternative relevant questionsmay be:Introductory 1: Do you understand I will onlyask you the questions we discussed?The following describes an example ofquestion numbering and type of question usedin the four-question version.Sacrifice Relevant 2: Regarding whether ornot you robbed that bank do you intend toanswer all of the questions truthfully?I1IntroductorySR2 Sacrifice RelevantN1 NeutralC1 ComparisonR1 RelevantR2 RelevantC2 ComparisonN3 Neutral (optional) This neutralquestion may be inserted at the option ofthe examiner to allow some decrease oftension and recovery to baseline. Ifinserted, the examiner will skip over thisneutral question during scoring.R3 RelevantR4 RelevantC3 ComparisonN2 NeutralNeutral 1:Are the lights turned on insideof this room right now?Comparison 1: Did you ever steal anythingfrom someone who trusted you?Relevant 1:?Did you rob that bank at onNeutral 2:Are you now physically locatedwithin the State of Texas?Comparison 2: Did you ever steal anythingfrom a friend or family member?Relevant 2:Did you plan or arrange withanyone to rob that bank at ?In Test OperationThe test subject is instructed to sit stillandanswer each question truthfully.However, the approach is to avoid doing thisin a heavy-handed manner. For example thefollowing admonition would be typical for thisapproach:Neutral 3:Do you sometimes listen tomusic while riding in a car?Comparison 3: Did you ever steal anythingfrom a place you worked?Polygraph, 2009, 38(1)20

Handler & Nelson“I need you to sit still during theasking of the questions. Movement willcreate distortion and artifacts in therecordings that will require me to repeatthe questions and that will make thetest longer.”After the third chart, the charts arenumerically scored. However, the test subjectis only told the examiner always stops at thispoint to carefully check the quality of therecordings before collecting more charts. Ifthe scores meet the threshold of the decisioncriteria, the data collection phase is complete.If the test result is inconclusive following thefirst three charts, two additional charts areconducted following the same rotationalpatterns described above. Following the fifthchart, all scores are totaled to make adetermination of veracity. The DefenseAcademy for Credibility Assessment (DACA),the Federal Training facility, (2006) permitsthe examiner to conduct just a fourth chart,and if necessary a fifth test chart. We areaware of no theoretical rationale or evidence tosuggest this procedural difference wouldinvalidate a test result.The examiner rotates the neutral,comparison,andrelevant(ifdesired)questions during the next and subsequentpresentations.The examiner may preferleaving the relevant questions always in thesame position, and rotating only thecomparison and neutral questions, making iteasier to score the charts by having a fixedorder of relevant questions.Moving thequestions helps to prevent pattern recognitionand anticipation of a specific order ofquestions during the examination.The following are examples of serialpositioning in the question strings showingone example of question rotation.The questions are presented to the testsubject at least three times across threecharts, with a brief discussion between chartsto clarify and resolve any perceived problemsraised by the test subject and to reinforce afocus on both the relevant and comparisonquestions (Raskin & Honts, 2002). After eachpresentation of the test questions, theexaminer should ask the test subject if he orshe has any concerns with the test questions.Honts (1999) reviewed data from 19 studiesthat involved 1092 polygraph tests and foundbetween chart discussion (even when limitedto only the comparison questions) reduced therisk of error. In 9 of the11 studies whichincluded examinations in which the questionswere reviewed between charts both therelevant and comparison questions werereviewed. In 2 of those studies, only thecomparison questions were reviewed. Hontsshowed that between-chart stimulation andquestion review reduced the false negative rate(54%), had a modest reduction of false positiverates (2.9%) and a substantial decrease ininconclusive outcomes for truthful testsubjects (42%).Three Question VersionFirst ChartI1,SR2,N1,C1,R1,N2,C2,R2,N3,C3,R3Second ChartI1,SR2,N2,C3,R2,N3,C1,R3,N1,C2,R1Third ChartI1,SR2,N3,C2,R3,N1,C3,R1,N2,C1,R2Four Question VersionFirst ChartI1,SR2,N1,C1,R1,R2,C2,N3(N3 is optional),R3,R4,C3,N2Second 1Third 2As can be seen above, each relevantquestion has an opportunity to be comparedto each comparison question across the threechart series. As discussed above, if the resultsare inconclusive after three charts, twoadditional charts are run. The examiner maysimply use the first and second serialpositioning question strings for the fourth andfifth chart.The following is typical of the type ofexchange that might take place betweencharts.Note how the examiner places equalemphasis on each group of questions duringthe stimulation and review.21Polygraph, 2009, 38(1)

Utah Approach to CQTExaminer: OK Roy, did you have anyproblems with any of thosequestions on the test?Roy:anything from a business” and label thatquestion C1a. The examiner should thenreview all test questions with the test subject.The examiner conducts the next two chartsand again starts by instructing the testsubject to sit still and answer all of thequestions truthfully.No.Examiner: Anything come to mind when Iasked you those questions?Roy:Test Data Analysis and DecisionCriteriaNo.Examiner: How about those questions aboutthe drug transaction? Is it clearwhat I am asking you? Do youunderstand them?Roy:The Utah Scoring System (Bell, Raskin,Honts & Kircher, 1999) is a simplified versionofthenumericalscoringtechniquesintroduced by Backster in 1963 and modifiedby the US Army around 1970 (Weaver 1980;Swinford 1999).Yep.Examiner: How about those questions aboutlying? Any problem with any ofthose?Roy:The Utah scoring system is a simpleand elegant scoring system designed toimproveaccuracy,reduceinconclusiveresults, and improve interrater reliability. Ithas fewer rules to follow and fewer criteria toscore than the other scoring systems currentlyin use. The Utah Numerical EvaluationScoring System was designed, refined andtested by Raskin and his colleagues.Nope.Therehasbeencontroversysurrounding the review of question betweentest charts. Abrams (1999) and Matte (2000)argued that review of comparison questionsbetween tests is incorrect and Offe & Offe(2007) found no contribution to improved ordegraded decision outcomes as a result ofbetween test review of the test questions. Inconsideration of these findings, we feel itprudent to recommend a review of thequestions between each chart, but find noreason to support an argument that theinclusion or exclusion of this review wouldcause a test result to become invalid orerroneous. Honts (1999) did not speculate asto the psychological cause of these findings.He correctly stated “The essence of science isempiricism. That is, scientific knowledge isbuilt on data, not speculation nor authority.”Honts chose to accept the data for what itstated on its’ own merit.The Utah scoring System is based onphysiological response data that has beenproven to be a valid and reliable indicator ofsympatheticarousal.Theinter-scorercorrelations of results produced using theUtah Scoring system are typically around 0.90(Bell et al., 1999). The accuracy of the UtahScoring system from several analog studieswas 90%, as reported by Bell et al., whenaveraged for programmed innocent and guiltytest subjects. The results of field studies usingthe Utah scoring system are consistent withanalog study results (Bell et al.1999)Numerical evaluation of the test data isaccomplished by comparing the relativestrengths of responses to comparison andrelevant questions. The Utah system uses a 7position numerical scoring approach. Therelative strengths of physiological reactions foreach sensor are compared and a score isassigned. The possible scores range from -3 to 3. The reaction of each relevant question iscompared to the reaction to the precedingcomparison question in the 3-question CQTformat or the stronger of the two surroundingcomparison questions in the 4-questionMGQT format. If the relative strength of theShould a test subject make additionaladmissions to comparison questions or needto modify a relevant question, the examinershould do so and re-label the question. Forexample, if during a Utah PLT the test subjectmakes an admission to question C1 “Beforethis year did you ever steal anything from abus

develop material for comparison questions to be used during the testing phase of the examination, although the nature of the issues to be resolved usually dictates the general content of the comparison questions. The examiner does not, however, lecture the test subject regarding past transgressions during this comparison question material review.