- Welcome to AntiPolygraph.org Message Board.
A forum for free and open discussion of polygraph-related topics
Posted by Fair Chance
- Apr 24, 2003, 03:11 AMDear J.B.,
Your answers prove that at least people of different opinions on the use of polygraph can have civil discourses.
As I said before, it looks as if any practical use of alternatives is going to be lost in the noise over CQT prescreening polygraph use.
Nice having a discussion with you.
Regards.
Your answers prove that at least people of different opinions on the use of polygraph can have civil discourses.
As I said before, it looks as if any practical use of alternatives is going to be lost in the noise over CQT prescreening polygraph use.
Nice having a discussion with you.
Regards.
Posted by J.B. McCloughan
- Apr 24, 2003, 02:40 AMFair Chance,
The construction of the CQT and the GKT questions are polar in nature. The CQT poses a direct question to the incident under investigation and proscribes a direct answer.
The GKT asks a question with multiple alternative endings and those endings are repeated.
One advantage of using the GKT is that it can save time and resources, which can ultimately equate to a monetary savings.
Pre-employment screening, as is regularly described and discussed here, is not an eligible use of the GKT. The most recognizable reason is that there are no known keys to construct such a test.
The construction of the CQT and the GKT questions are polar in nature. The CQT poses a direct question to the incident under investigation and proscribes a direct answer.
The GKT asks a question with multiple alternative endings and those endings are repeated.
One advantage of using the GKT is that it can save time and resources, which can ultimately equate to a monetary savings.
Pre-employment screening, as is regularly described and discussed here, is not an eligible use of the GKT. The most recognizable reason is that there are no known keys to construct such a test.
Posted by J.B. McCloughan
- Apr 24, 2003, 02:13 AMGeorge,
Dr. Lykken writes more about statistical detection of countermeasures for the GKT. I personally find it difficult to believe that even a 'sophisticated' subject who possessed concealed knowledge could confound this examination, given all the processing need to be achieved in a very short window period.
I do not have a copy of that issue of Polygraph at this time but will read it again. It is understandable that college students taking a polygraph on a mock crime scenario may attempt spontaneous countermeasures, even those assigned to a truthful pool. I have seen this happen first hand. One reason an individual may use such countermeasures is that they want to see if they could beat the polygraph. They are no inherent penalties for them if they produced a negative result being in a truthful pool. In my opinion, the spontaneous countermeasure charts of a truthful individual on a CQT would not be indicative of those of a deceptive. I would like to expound on this issue, however I must guard some opinions for later testing through research.
Regardless of the above, a properly educated innocent suspect would not use countermeasures on a GKT in real world settings. It is up to the examiner to ensure that the test areas are foreign to the examinee, the examinee fully understands the procedure, and the examinee understands the ramifications of countermeasures on such a test. Personally, I have yet to see this phenomenon apply in field use of the GKT.
Dr. Lykken writes more about statistical detection of countermeasures for the GKT. I personally find it difficult to believe that even a 'sophisticated' subject who possessed concealed knowledge could confound this examination, given all the processing need to be achieved in a very short window period.
I do not have a copy of that issue of Polygraph at this time but will read it again. It is understandable that college students taking a polygraph on a mock crime scenario may attempt spontaneous countermeasures, even those assigned to a truthful pool. I have seen this happen first hand. One reason an individual may use such countermeasures is that they want to see if they could beat the polygraph. They are no inherent penalties for them if they produced a negative result being in a truthful pool. In my opinion, the spontaneous countermeasure charts of a truthful individual on a CQT would not be indicative of those of a deceptive. I would like to expound on this issue, however I must guard some opinions for later testing through research.
Regardless of the above, a properly educated innocent suspect would not use countermeasures on a GKT in real world settings. It is up to the examiner to ensure that the test areas are foreign to the examinee, the examinee fully understands the procedure, and the examinee understands the ramifications of countermeasures on such a test. Personally, I have yet to see this phenomenon apply in field use of the GKT.
Posted by Fair Chance
- Apr 23, 2003, 09:46 AMQuote from: Bill on Apr 23, 2003, 04:00 AMThanks guys, that actually helps a lot. However, how can we tell which test we are getting? The LBTLD gives an example of some GKT questions, but I'm just a little confused how they differ from the CQT. I'm understanding it almost like multiple choice answers to one question, but what are the response reactions compared to as in the CQT?
Dear Bill,
The questions asked could be very similar in both cases. Both CQT and GKT tests can be used after an investigation and gathering of known facts and thus the creation of test questions could be similar.
The GKT is normally only used with an investigation that results in key facts that are hidden from the examinee. The CQT is not always used in this manner. The time and cost of a GKT test is quite a bit more than a CQT test if developed properly.
George is correct that someone attempting countermeasures who is not well learned on the test or vice-a-versa could skew the test .
J.B. is correct that if the examinee is not using countermeasures that it could be a more accurate exam than the CQT. In any case, rarely would a GKT be constructed for employment prescreening.
Regards.
Posted by George W. Maschke
- Apr 23, 2003, 04:12 AMQuote from: J.B. McCloughan on Apr 23, 2003, 01:49 AMFair Chance,
The way countermeasures could be detected in the field is by utilizing statistical probabilities of an unassuming subject's ability to augment their response to any or all but the key without the knowledge of the key. Expected rank scores may be check as well. Considering the key is randomly moved to different position, their ability diminishes even more with each test and each control added. As I stated to Bill in my previous post, there are no reasons for a false suspect to employ countermeasures to this type of test.
J.B.,
A sophisticated subject might also produce a reaction to one or more keys in an attempt to confound any such statistical analysis.
I agree with you and Professor Lykken that there are no (good) reasons for an innocent suspect to employ countermeasures to a GKT. But that does not necessarily entail that innocent suspects won't attempt countermeasures.
In a laboratory study recently published in Polygraph (Honts, C.R., Amato, S.L., Gordon, A.K., "Effects of Spontaneous Countermeasures Used Against the Comparison Question Test," Polygraph, 2001, 30, 1-9), 45.8% of innocent subjects reported that they attempted "spontaneous" (untrained) countermeasures. It's possible that innocent suspects might attempt such countermeasures during a GKT.
Posted by bill
- Apr 23, 2003, 04:00 AMThanks guys, that actually helps a lot. However, how can we tell which test we are getting? The LBTLD gives an example of some GKT questions, but I'm just a little confused how they differ from the CQT. I'm understanding it almost like multiple choice answers to one question, but what are the response reactions compared to as in the CQT?
Posted by George W. Maschke
- Apr 23, 2003, 03:58 AMQuote from: Fair Chance on Apr 22, 2003, 12:43 PMGeorge and/or J.B.,,
I have read much about the GKT test. In this illustration that you quoted, were the tests blind (i.e. were there five hundred test subjects and 20 unknown subjects who tried to use countermeasures)?
My question leads to the obvious. If I am offering a reward for beating the GKT with countermeasures but I know this ahead of time and I know that all test takers are going to try and use countermeasures, how do I objectively transfer this study to the real world where I am not going to be sure if a test subject is trying to use countermeasures?
Regards.
I haven't read the study to which Lykken refers (D. Lykken, The validity of the guilty knowledge technique: The effects of faking, Journal of Applied Psychology, 1960, 44, 258-262). It's possible that the polygraph charts were blindly scored. Note that Lykken is not saying that countermeasures were detected in this study. Rather, he is saying that they were not successful against the GKT.
Posted by Fair Chance
- Apr 23, 2003, 02:47 AMDear J.B.,
I understand your "key" in using the statistical probabilities during GKT usage. I believe that it deserves more research and serious study. This is coming from a victim of polygraph prescreening. Like Marty, I agree that CQT polygraph cannot differentiate between truth and deception but attempts to document physiological repeatable events during countermeasures might be feasible. That is another discussion altogether.
My previous question goes more into the weak studies that most previous statistical results are based upon. I do not doubt that the GKT is a different animal then the CQT test. I only wish that any studies done by proponents of this system would seriously construct a valid blind test which could withstand normal scientific scrutiny. "Paying" known subjects to try and "beat" the test is certainly not scientific and does not represent the way to persuade authorities such as the NAS panel or people like myself who are willing to be "constructively skeptical" (and willing to accept new technology if research is acceptable).
As stated by many on this website, the baby is getting thrown out with the bath water because of the insistance that CQT pre-screening polygraph is a valid and accurate test in applicant and employment screening. CQT has certain utility for eliciting confessions but certainly little or no ability to detect a trained spy.
If the CQT proponents would have admitted that maybe the "science" of its use was weak upon the publication of the NAS report than the door would have been more open to review other forms such as GKT.
Despite what many polygraph proponents think, I am open to science but it must be independent and scientifically repeatable by organizations with no vested interest in the outcome of the studies. This would remind me of the days of tobacco companies doing lung cancer research as their profits depend upon the sales and use of cigarettes. The APA cannot be the fox watching the hen house.
As it is, your well thought out intelligent presentations are getting lost in the CQT "polygraph noise."
Regards.
I understand your "key" in using the statistical probabilities during GKT usage. I believe that it deserves more research and serious study. This is coming from a victim of polygraph prescreening. Like Marty, I agree that CQT polygraph cannot differentiate between truth and deception but attempts to document physiological repeatable events during countermeasures might be feasible. That is another discussion altogether.
My previous question goes more into the weak studies that most previous statistical results are based upon. I do not doubt that the GKT is a different animal then the CQT test. I only wish that any studies done by proponents of this system would seriously construct a valid blind test which could withstand normal scientific scrutiny. "Paying" known subjects to try and "beat" the test is certainly not scientific and does not represent the way to persuade authorities such as the NAS panel or people like myself who are willing to be "constructively skeptical" (and willing to accept new technology if research is acceptable).
As stated by many on this website, the baby is getting thrown out with the bath water because of the insistance that CQT pre-screening polygraph is a valid and accurate test in applicant and employment screening. CQT has certain utility for eliciting confessions but certainly little or no ability to detect a trained spy.
If the CQT proponents would have admitted that maybe the "science" of its use was weak upon the publication of the NAS report than the door would have been more open to review other forms such as GKT.
Despite what many polygraph proponents think, I am open to science but it must be independent and scientifically repeatable by organizations with no vested interest in the outcome of the studies. This would remind me of the days of tobacco companies doing lung cancer research as their profits depend upon the sales and use of cigarettes. The APA cannot be the fox watching the hen house.
As it is, your well thought out intelligent presentations are getting lost in the CQT "polygraph noise."
Regards.
Posted by J.B. McCloughan
- Apr 23, 2003, 01:49 AMFair Chance,
The way countermeasures could be detected in the field is by utilizing statistical probabilities of an unassuming subject's ability to augment their response to any or all but the key without the knowledge of the key. Expected rank scores may be check as well. Considering the key is randomly moved to different position, their ability diminishes even more with each test and each control added. As I stated to Bill in my previous post, there are no reasons for a false suspect to employ countermeasures to this type of test.
The way countermeasures could be detected in the field is by utilizing statistical probabilities of an unassuming subject's ability to augment their response to any or all but the key without the knowledge of the key. Expected rank scores may be check as well. Considering the key is randomly moved to different position, their ability diminishes even more with each test and each control added. As I stated to Bill in my previous post, there are no reasons for a false suspect to employ countermeasures to this type of test.
Posted by J.B. McCloughan
- Apr 23, 2003, 01:36 AMBill,
You are correct that in a Guilty Knowledge Test the controls are indiscernible to a person not involved in a crime or its scene. This is considering that the correct answers (known as keys) have been properly concealed.
You are incorrect to assume that false positives will be much higher. Inasmuch as the key and controls would appear equally plausible as the key to the 'innocent' suspect, the person not involved in the crime or its scene will have a low probability of giving their strongest response to the key. If one test is given and there are 6 alternative answers to a question but only one correct, an unassuming subject has theoretically a 1 in 5 chance of giving their strongest response to the key. If there are five of these tests on separate areas of the crime and its scene, the probability of the subject giving their strongest response to the key diminishes with each test and as well with the increase in the number of controls within a test. Furthermore, an 'innocent' suspect would be unwise to use countermeasures on this type of examination because they may well augment their response to the key and the examiner can advise them prior to the examination of the statistical probability of error for the test that is to be administered.
I hope this answered your questions.
You are correct that in a Guilty Knowledge Test the controls are indiscernible to a person not involved in a crime or its scene. This is considering that the correct answers (known as keys) have been properly concealed.
You are incorrect to assume that false positives will be much higher. Inasmuch as the key and controls would appear equally plausible as the key to the 'innocent' suspect, the person not involved in the crime or its scene will have a low probability of giving their strongest response to the key. If one test is given and there are 6 alternative answers to a question but only one correct, an unassuming subject has theoretically a 1 in 5 chance of giving their strongest response to the key. If there are five of these tests on separate areas of the crime and its scene, the probability of the subject giving their strongest response to the key diminishes with each test and as well with the increase in the number of controls within a test. Furthermore, an 'innocent' suspect would be unwise to use countermeasures on this type of examination because they may well augment their response to the key and the examiner can advise them prior to the examination of the statistical probability of error for the test that is to be administered.
I hope this answered your questions.
Posted by bill
- Apr 22, 2003, 09:11 PMI'm misunderstanding. Does this mean that the GKT has a much higher percentage of accuracy than the others? And if you're innocent the control questions are unrecognizable? I guess that would mean there is a higher risk of false positives, since control question reactions couldn't be augmented. Does that make sense?
Posted by Fair Chance
- Apr 22, 2003, 12:43 PMGeorge and/or J.B.,,
I have read much about the GKT test. In this illustration that you quoted, were the tests blind (i.e. were there five hundred test subjects and 20 unknown subjects who tried to use countermeasures)?
My question leads to the obvious. If I am offering a reward for beating the GKT with countermeasures but I know this ahead of time and I know that all test takers are going to try and use countermeasures, how do I objectively transfer this study to the real world where I am not going to be sure if a test subject is trying to use countermeasures?
Regards.
I have read much about the GKT test. In this illustration that you quoted, were the tests blind (i.e. were there five hundred test subjects and 20 unknown subjects who tried to use countermeasures)?
My question leads to the obvious. If I am offering a reward for beating the GKT with countermeasures but I know this ahead of time and I know that all test takers are going to try and use countermeasures, how do I objectively transfer this study to the real world where I am not going to be sure if a test subject is trying to use countermeasures?
Regards.
Posted by George W. Maschke
- Apr 22, 2003, 07:04 AMThe same kinds of techniques could be used to produce reactions to the control questions in the Guilty Knowledge Test (GKT). David T. Lykken discusses countermeasures and the GKT at p. 303 of the 2nd edition of A Tremor in the Blood: Uses and Abuses of the Lie Detector:
QuoteCountermeasures
The GKT will be susceptible to distortion by a guilty suspect who is skillful enough to covertly augment his reactions to the controls. As is true also in lie detection, the best defense against such countermeasures is an observant examiner. Since the GKT is to be used only in criminal investigations, as a guide to the police, it does not much matter that a suspect invalidates the test in this way as long as his efforts are noticed. In lie detection, an innocent suspect might augment his control responses because of a justified fear of failing the test even though innocent. Only a guilty suspect would have reason to try to beat the GKT, the cardinal virtue of which is a vanishingly small likelihood of false-positive errors. More importantly, an innocent suspect could not systematically self-stimulate on the controls, since he would not know which alternatives are controls and which are relevant. Therefore, the police investigator learns just as much from the fact that a suspect tries to defeat the test as he would learn from a score in the guilty range. In both instances, he will focus his investigation on this suspect, searching for admissible evidence against him.
Under special circumstances, it may be worthwhile to attempt to get a valid GKT score in spite of obvious attempts by the respondent to defeat the test. This will require perhaps five presentations of a ten-item set in order that the method of expected ranks, described earlier, can be employed. When this technique was tested experimentally, 20 subjects were allowed to practice self-stimulation to produce misleading responses, and they were offered a money prize if they could "beat" the test; none were successful. Whether the method will work as well in real-life applications remains to be seen.[/font][/size]
Posted by bill
- Apr 22, 2003, 06:09 AMDoes the countermeasures discussed in chapter 4 apply the same to the guilty knowledge test the same as the others? There wasn't as much info on this type of test, and I was getting pretty confused when comparing to the CQTs