posted 08-16-2005 11:31 AM            

I have no research but I have a little experience. I have tested many child molester suspects and when doing so use adult/sex lie controls rather than general lie or general sex controls. The reason is that (I heard somewhere, which made sense) that general sex controls on a child molester are (possibly) too overpowering and will possibly make the CQ's relevant since most are usually repeat offenders by the time they get caught the first time. Besides that, they had to start their deviant behavior somewhere, probably with adults, even so...

Also, that if the suspect is innocent the adult/sex lie controls will be significant enough to get them through the emotionality of the false allegation (especially, if a parent of the victim).

Just my $.02

posted 08-16-2005 12:55 PM            

Sex comparison questions are prevalently used for sexual crimes, for they fit the ‘profile’ of the discussion at hand and test issue. Lie comparison questions are used sometimes when the introduction of such material may not be age appropriate. A sex comparison question can be worded so that it is not as salient to the guilty person but is quite so to the innocent.

As for the test differences, there are other factors that may be involved in why polar opinions or results occurred. The test format, relevant questions wording, and the chart interpretation may be the difference, along with many other variables.

posted 08-16-2005 05:09 PM            

When I run an instant offense sex case I use sex comparisons but if it is a child crime I qualify it by saying "with an adult". For instance I might ask "Have you ever done anything sexually abnormal with an adult?". I also like using a masturbation question; Between the ages of 13 and 18, did you masturbate more than normal? Of course I tell them normal is once or twice a month.

I believe DODpi says to use sex comparisons with sex issues not lie comparisons.

Jack

posted 08-16-2005 05:16 PM         

Sex controls should be OK as long as they are not too close to the relevant issue being tested. In other words, if the victim is a child, sex controls should avoid having a child as part of the question.A question such as "Have you ever masturbated while looking at a picture of a naked woman" might be appropriate. A question like "Have you ever masturbated while looking at a picture of a naked underaged girl" would not be appropriate.

Any other thoughts??

posted 08-16-2005 08:33 PM            

Personally, I don't think lies are strong enough, so I run a Utah test and use a lie, hurt / harm and something else fitting (perhaps a "not connected with this issue (sex)..." dishonest or illegal type CQ, etc.). I think "abnormal" sex with an adult would be appropriate too.

The fear with sex CQs - and it's not proven yet - is that a sex CQ can tap into fantasy land in the subject's mind and lead to arousal, making the CQs more salient. Since nobody knows - and I suppose it's theoretically possible - many avoid it.

J.B.,

posted 08-16-2005 10:56 PM            

One study I cited on a previous thread, are time bars necessary. It was conducted by one of the participants here, John J. Palmatier PhD. I will find any other studies I can find for you.

Here is a recap of that previous post:

quote:

---

It is my understanding from training and in reading research that time bars are not necessary. In fact one research study done on the use of exclusive (time bar) and nonexclusive (no time bar) comparisons found that the nonexclusive had significantly better results in correct decisions (+32%), false positives (-15%), and inconclusive (-18%) but had a slight increase in false negatives (+1%). This study is on criminal specific single issue testing though.

---

If still following the board, it is better that John answer the questions regarding his study and issues on the topic of exclusive vs nonexclusive.

A couple examples of nonexclusive sex controls:

quote:

---

"Did you ever masturbate as discussed?"

---

(The discussion involves frequently masturbating to pornographic material or other fetishes, as only sexually deviant people would do such a thing.)

quote:

---

"Did you ever participate in any odd sexual activities not disclosed?"

---

(or for the IQ challenged)

quote:

---

"Did you ever do any odd sex not told(or disclosed)?"

---

posted 08-17-2005 04:00 PM         

I too went to polygraph school over 15 years ago, but no longer use the controls that I was originally taught.

posted 08-18-2005 09:20 AM            

just my $.015

posted 08-18-2005 10:59 AM            

Absent solid research, I don't care for the idea of using a general lie control against a sex issue due to the sensitivity of the issue. I don't believe a general lie is strong enough against a sex issue. The issue alone could evoke a response. I believe balance is achieved with sex controls, similar to theft controls for theft issues, etc. The examinee is probed during pre-test for the most suitable and and appropriate sex control.

During pre-test I've had success with the examinee agreeing that performing or involvement in certain sex acts (could) be considered deviant. No one wants to be considered deviant! Adding, (others) might believe continuing the (identified) deviant act could progress into involvement in the activity/crime now under investigation.

In establishing controls I like to use questions asking if the examinee performed certain acts or thought of performing certain acts. In answering, I look for the manner in which they talk about the (sex) question, i.e., "I never -," knowing there is a probability the examinee could have done something I asked about.

No doubt you'll be reviewing question construction before the charts.

posted 08-18-2005 12:47 PM            

Krapohl, D.J., Stern, B.A., & Ryan, A.H. (2003). Exclusionary or nonexclusionary: A review of the evidence. Polygraph, 32(4), 245-250.

In summary:

quote:

---

Though it is a scientific cliché to suggest more research on an issue at the end of a paper, it does not seem to be warranted in this case. With a 0-4 record, the notional superiority of the exclusionary PLC is probably a settled matter. This does not discount its usefulness, of course. Polygraphers and agencies can rely instead on the nonscientific considerations cited earlier to continue a preference for this type of comparison question.

---

The research is cited within this article and, if you do not already have a copy, a back issue may be obtained from the APA.

posted 08-18-2005 12:52 PM            

I do like the set up you mention: "Have you ever done or even thought of doing...?"

posted 08-18-2005 06:47 PM            

"How do you know there is a probability the examinee could have engaged in the sex act(s) you might use in your CQs?"

Relative to establishing a sex control, similar to any other control, I (don't) know if an examinee did in fact engage in the act addressed in the control question. I have found masturbation to be a good control topic with a male adult as it is an act that is frequently denied. Once denied, I go a step further with the examinee and state "good, I'm glad you're not that type of person (inferring deviance previously discussed)." I then establish the wording and set the time bar I believe to be appropriate/proper.

I've provided a synopsis of how I've established at least one sex control I had success with in the past. I believe a solid pre-test is important, no matter how long it takes.

END.....

posted 08-22-2005 01:28 PM            

I would agree with Barry in that the ‘Relevant Question’ wording is most likely the culprit in this case, if a false negative. The test questions should, at a minimum, be provided for discovery.

The CQT is relatively more prejudice toward the innocent and, inasmuch as this is the case, false positives are more prevalent. There are ways to change the errors from false positive to false negative but there are consequences related with the changing of the decision rules as well (increase in false negatives). One could technically significantly reduce both errors and significantly increase accuracy. However, this method would result in a significant increase in inconclusive results as well, which is what latent print examiners do.

posted 08-22-2005 03:03 PM            

I wouldn't review any charts without reviewing technique/question construction. It sounds to me as though the examiner is not confident in their work.

posted 08-22-2005 06:12 PM            

Since I don't know if the sex CQs are bad crowd are on to something, I stay away from them as their point is a possibility (you tap into fantasy, which certainly could evoke a response - but not for the reason you want - or don't want - one. Instead I use, guilt / shame, hurt / harm, (non-sex) illegal, dishonest, etc. There are plenty of options out there.

posted 08-23-2005 12:33 AM            

I use the same type of CQ as Taylor. I add the "in front of a mirror" portion to the masturbation question if I get an admission.

My intent in adding this is not to lead them to think they are being "disbelieved" but for them to be concerned about being thought of as a deviant. I know the CQ and RQ presents this possibility, but this is why I choose to use a sex control. IMO A lie question may not be strong enough to be used against a sex RQ that is always a "hot" issue & sensitive. I believe this is especially true regarding molestation issues.

After covering deviant acts in pre-test and the examinee agreeing involvement in front of a mirror would likely be considered an escalation of deviancy, I'm looking for them to deny involvement in the deviant act. I may add a statement - "it's been proven that someone who has progressed to this level of deviancy is capable of committing this (crime being investigated).

Depending on the education level and sophistication of the examinee I may try to add support to the definition. I state "studies of persons convicted of this crime have verified many had been involved in this type of activity." I am cautious here as I don't want to say too much. I don't want to comment on something that will be challenged and which I can't substantiate. Either way, I can identify a few ficticious "recent studies" that sound as though they support my statement.

In this manner I believe I can establish a good CQ that will help structure a test that is fair to the examinee.

END.....

posted 08-23-2005 07:49 AM            

I do the essentially the same thing as you with any CQ, making them "profile" questions in addition to the other reasons I introduce. (I heard Ed Gelb did a great APA presentation some time ago on setting CQs as profile questions in such a way as to cause anybody great concern, which I wish I'd seen.) I know an examiner who introduces them something like this: "It has been determined by various studies that there is a 98% chance that people who answer yes to the folling questions are pedophiles or have those tendencies." Then he asks the CQs, and plays them up from there.

Someday I'd like to just watch videos of people setting CQs and see what I could glean. There are so many clever little nuances that likely make a big difference with some examinees.

J.B.,

I'd argue the CQT is not prejudice against the innocent, but certain examiners seem to be. If you show different examiners the same tests, many will ID the truthful at near chance levels. Yet, some will ID them (and the DI) at levels well above chance, which means the CQT is not the problem - it's the scorer. The Utah studies ID'd the DI and NDI at about the same rate when the charts were scored using their validated scoring method, which would show the necessary data is there if you know what to look for.

posted 08-23-2005 09:50 AM            

I ensure as best I can, the examinee agrees (some, if not all people) would consider certain types of acts as sexually deviant. I will continue to change/increase the sensitivity of this question with words that I believe are appropriate based on information I learn about the examinee prior to and during pre-test.

"COULD have done - not necessarily did" pertaining to question construction is not very different. The bottom line is the negative response to either can be reasonably believed (by the examiner) to be a probable lie.

What I may consider an appropriate question for one examinee I might not even consider for another. I agree being able to observe videos of examinations can be very helpful. A great learning tool for interview & interrogation techniques related to question construction for different violations (burglary, robbery, homicide, sex, etc).

Audio tapes would also be beneficial but being able to observe the examinee during the interview is best.

END.....

posted 08-23-2005 11:52 AM            

My CQ's are probable lies but sometimes the probability factor doesn't exist. ie, In Utah, there are a lot of Mormons...Part of their belief is they are not to masturbate (now don't quote me on this because I don't practice this religion - I just have some knowlege on the Mormons). Now we all know at some point in a persons life, they may have masturbated - some more than others. However, there are a few that have not done this act. If my pretest isn't good, and they are one of the few, if my CQ is relating masturbation then they will not have a response and of course the RQ is hot and they would respond to that Q. However, if I pump it up enough, I don't have to know if they did the CQ or didn't, because in the pretest I would have them labled as deviant if they did or even considered doing the CQ act - I then tell them if they considered it they progress in the level of deviancy that is needed to complete the RQ act. I then will add a lie questions similar to "In the past ten years, have you even considered telling a lie to a church official?" I get them every time on that lie CQ. So, I am mixing Sex and Lie CQ's.

posted 08-23-2005 04:12 PM            

I have to suspect no matter what a person believes the masturbation question could work. Again, whether or not I choose to use this type of question(s) will be determined by what I learn about the examinee right up to the time I begin question construction. We always run the risk of talking to a person that has not "touched themself inappropriately" or been involved in the activity we're trying to establish as a control. If, in my opinion, I might be talking to a person not involved in the act I'm trying to establish a CQ around, I move to another subject, i.e., sex act.

If I, for any reason, believe the examinee DID NOT get involved in the act I am referring to I've lost the critical "probable lie" element of a CQ.

Although religion could be used as a good lie CQ, as I've indicated before I prefer to stay with the same subject for the CQ and RQ. I will definitely try and use the examinee's religious belief during a post-test interrogation if I ran an exam that was "DI."

posted 08-23-2005 08:53 PM            

I personally don't like the masturbation PLCQ as there's the chance a person never has. You can pre-test a guilt / shame question to include that and a hundred other things which increase you odds drastically of hitting something good with the question.

I am still crazy enough to believe polygraph is going to make it into evidence here someday (I've got a DA willing to try it if the right case comes along), and I don't want to have to explain to a judge or jury how I know - which I couldn't - that the person in front of me probably masturbated. I can cite all kinds of studies showing people lie all the time. I don't think I've ever had more than four admissiions (total) to three CQs, so I know (based on scientific data) there are 10,000 +/- lies most 25 year olds have failed to disclose. I think judges and juries would agree, intuitively, to a good hurt / harm CQ and the like.

posted 08-23-2005 09:18 PM            

The numeric scoring system for the CQT (ZCT) that is used by most is NDI= + in every spot and a +6 overall vs DI = -3 in any spot or a -6 overall. The MGQT is even more difficult to reach the threshold of NDI. It is hard to argue that those systems are not prejudice. The aforementioned scoring system is designed for a reason, investigative polygraphs. It is opined that an error of a false positive in an investigative polygraph will cause the innocent to be subjected to further investigative measures that they would have been subjected to absent the polygraph and that a false negative will allow a guilty person back into society, to potentially commit other crimes. Insomuch as this is the ideology, the acceptable error is to set the cutting scores for false positive errors.

You stated,

quote:

---

the CQT is not the problem - it's the scorer.

---

That is of my opinion as well but I think it is much more than just scoring. However, to know your error rate for blind scoring you would need to establish it through some method (i.e. the Marin Protocol).

You say,

quote:

---

It's true shifting scoring criteria will only increase errors in one direction or the other, but for some that is necessary to balance them out.

---

I am not completely sure what it is that you mean by this. Are suggesting shifting the criteria/cutting scores on a case by case basis (i.e. if truthful than this and if deceptive than this) to balance things out?

posted 08-24-2005 12:52 AM            

We face the identical problem in an explanation to a judge/jury about how we know the examinee "considered" or "thought about" (masturbating). How can we explain we know an examinee "considered" or "thought about" a subject at any time?

I agree, people lie all the time. The CQ by definition is a "probable lie" question. Hopefully, we could know data about an examinee prior to the exam, and; based on their answer during question review we'd know for a fact they are being deceptive. The best possible CQ. Without something like this we have to depend on our knowledge, training & experience to help us structure the best question.

I'm not following the scoring system concern addressed here. All questions are evaluated on a scale of +-2 or more. Using a Zone of Comparison technique(ZOC)I use the +-6 or more (total) to render an opinion. Anything below 6 is considered INC. Relative to the "spot" totals I also follow the +-2 scoring in two or more of three tests for an opinion.

With the MGQT I use the same +-2 or more @3,5,8 & 9 in two or more of three tests.

While scoring a "3" is possible, especially in the GSR ratio, I don't recall scoring this high. Very conservative scoring, giving all benefit of doubt to an examinee when scoring should leave a defensible & solid opinion.

I disagree this scoring system is prejudicial or that it is only designed for "investigative polygraphs." I think it can be fairly/accurately stated that all polygraph examinations are "investigative." The polygraph, and obviously the scoring is subjective, not prejudicial. Two examiners can look at the same charts and may score higher or lower at each question. The end result (opinion) should be the same.

An examinee is presented to an examiner because they are suspected of involvement in an incident. The exam may give an investigator direction, but it should only be facts/evidence in an investigation that are used to "clear" any person suspected of involvement, not examination results.

What is the "acceptable error rate" that is being referred to? IMO there is not acceptable error rate relative to polygraph.

posted 08-25-2005 03:40 PM            

According to Don Kraphol in 2004, (and sir if I make an error, please correct me) wrote that there were five guiding principles for comparison questions:

#1 The examinee must believe that there are consequences to being deceptive to the CQ that approach, but do not exceed, the consequences for deception to the RQ's.
#2 The examinee must believe that he/she is not "entirely" truthful to the CQ's
#3 The structure of the CQ must not be evocative in and of itself" (cannot evoke emotion/feelings by the mere nature of the question itself)
#4 The CQ must NOT be explicitly relevant whereas the RQ must be.
#5 "For face validity to the examinee, the CQ must have some tangential relationship to the topic covered in the RQ.

He wrote further that none of these principles require exclusionary CQ's. Exclusionary CQ's would work, but so would carefully crafted non-exclusionary CQ's.
Varying CQ's (exclusionary/non-exclusionary may off some degree of protection against countermeasures from those looking for time bars.

Currently and in general, the type of CQ one chooses really boils down to the philosophy of format, not science.

The rest of this is my opinion...........
If you accept Krapohl's comments, and I think they make sense, then lie CQ's whether time-barred or not should work as well as sex CQ's providing you establish a tangential relationship that convinces the examinee that the consequences of lying to a CQ approaches but does not exceed the consequences of a lie to an RQ.

If you choose lie CQ'S, in order to establish that relationship, the Pre-test interview one would have to stress that the significant issue is not the commission of an alleged act but instead the examinees truthfulness regarding questions concerning the act.

Let's assume you are investigating an accusation of frottage occurring in 2005.
You construct and carefully pre-test the following CQ: "Prior to 2004 do you remember engaging in an act of sexual deviance?" Unknown to you the subject had committed a brutal lewd molestation and murder 5 years ago. The reaction to that CQ stands a very good chance of not only overpowering the RQ's, but might very well wash out the symptomatics as well.

Considering the risk that a Sex CQ that excites an examinee’s fantasy or touches on a previous more serious undiscovered criminal act might trigger a significant reaction of the level that the relevant issue reaction suffers due to the anticlimax-dampening concept, lie CQ's seem to make sense. Especially if you regard that, in this scenario, it would only be the guilty subject that would benefit from an inappropriate CQ response.
In that same scenario, using a lie CQ, I would argue that you still might not get much reaction to the RQ, but your symptomatic questions have a better chance of doing their job.

posted 08-25-2005 03:57 PM            

Every scientific testing procedure has an error rate. In polygraph scoring that error rate can be shift away from or towards zero. This shift can reduce or increase the probability of error but increase or reduce the number of inconclusive results.

Example. ZCT 3 position scale has a scoring possibility of
+27.............0..............-27

by moving the cutting scores outward from zero, error probability is reduced because you increase the numbers required for a conclusive result. Inconclusive results are not error and a broader inconclusive range benifits the accuracy of the results when they are definitive.
Acceptable error is an agreed balance between the acknowledgement that errors occur and the need to obtain conclusive results.

posted 08-25-2005 08:15 PM            

The science is in on symptomatics: they don't work. If anything, they lower the scores of the truthful, which leads us to the other post that sprung from this.

posted 08-26-2005 06:29 AM            

Also where can I find an authoritative article establishing symptomatics don't work or lower the scores of the truthful?

Can you agree that a strong reaction to a symptomatic is indicitative of an outside issue problem even though lack of reaction does not mean that there isn't an outside issue problem?

As for "What is optimal?" Theoretically it makes for good discussion but realistically, I think if we are strictly adhering to the rules governing a validated testing format we are more or less obligated to use the cutting scores that accompany the format. Research may alter cutting scores over time; however, I think if someone alters the cutting scores for a DODPI ZCT or a Backster You-Phase they then become a Billy-Bob ZCT or a Poindexter You- Phase and whatever validity they previously enjoyed disappears.

TO: Billy-Bob Poindexter, if you are reading this, no offense intended.

posted 10-14-2005 07:53 AM            

I use lie comparisons in PCSOT sex tests whenever possible, but I "strengthen" them by using the kind of lie qustions that no PCSOT people like to admit ("Have you been lying to your probation officer" or "do you ever lie in treatment" or "Are you the type of person who would lie to the parole board?"...these work well.

For polys on new sex offense allegations, I use a mixture of lie and sex controls.

I just use some kind of bar when separates the relevant issue from the comparisons (adult vs. juvenile behavior or time frame bars).

I dropped symptomatic questions after talking with the folks at DoDPI.