posted 08-06-2008 02:33 PM            

My question from examiners is; What type of control and relevant questions are most effective for this type of test? Do I ask relevant questions that are answered with a yes, or have her write a statement and ask her if she lied about anything in the statement regarding the sexual acts. Can anyone give me good examples of questions for the exam?

posted 08-06-2008 02:50 PM            

Ray has discussed statement tests here before, explaining why he doesn't like them.

posted 08-06-2008 03:14 PM            

I agree with Barry. Tough test, BUT! It can be done.

Since we're not here to argue the merits and validity of the "statement test", I would suggest you use just that.

Have her prepare a short statement including the elements of the crime, sign, date, make it "official", etc. Review it with her, ie make her read it aloud to you, or visa versus, then conduct a Bi-zone "yes" (answered) CQ test with something like, "Is the handwritten statement you provide here today absolutely truthful?" with"...to the best of your knowledge/recollection?" etc...

CQ's: BTY, did you ever lie to get another person in trouble, to get even, etc. This has worked very well for me in the past.

My $.02

Jim

posted 08-06-2008 03:46 PM            

It talks about having the relevants answered with a “yes” and the comparisons answered with a “no.” Use lie comparisons.

I’ve tested victims of sexual assault (don’t like to, but if the conditions warrant it, I’ll do it if certain criteria is met).

Actual relevants I have used are:

Did _______physically hold you against that wall?
Is________the person who unzipped your coveralls?
Did you try to stop _________from having sexual intercourse with you?

posted 08-06-2008 03:55 PM            

A polygraph result is a probability statement. If you have a DI, you've got to have made sure you've eliminated everything other than deception (insofar as that's possible) as reasons for reactions to the RQs. Remember, an innocent person has no memory of the event tested in the RQs. A victim does, which is why we don't test them by asking if they are victims, which the above questions do.

posted 08-06-2008 04:19 PM            

Legitimate victims of sexual assault may lie or conceal information in their statements to police. The two biggies are lying about how much they resisted the rape in order to convince their significant other that they didn't just give up and concealing certain acts by their attacker that they find too disgusting to talk about.

Lying about either one of these issues could cause them to fail and exam even though their allegation of rape is absolutely legitimate.

A short statement confirmatory exam like Sackett describes removes those two problems.

------------------
Ex scientia veritas

posted 08-06-2008 05:08 PM            

I like to make the statement a big deal. A form that has legal sounding language stating lying on the form is a crime is what I like to see. That way, you can test on a crime. "Did you lie on that form?"

posted 08-06-2008 06:23 PM            

quote:

---

Use it for a tool, and not the deciding factor.

---

Let's follow this through. If you only want to test people there is reason to suspect are lying, then what do you do with the results of your "tool"? Put the two together and the case is essentially over if you get a DI.

posted 08-06-2008 11:40 PM         

Your memory is indeed correct. I don't have time to reply tonight in the depth the issue requires. I'll post details tomorrow.

posted 08-07-2008 12:09 AM            

On those occasions, it has been my anecdotal experience that the statement test works the best for these cases (I have done it both ways).

As Barry stated, the instant action questions for a true victim of this type of crime has too much potential to carry great salience.

I usually have the subject write a statement and then push the statement as a sworn statement or letter to the prosecutor/DA or judge. If there has been a court hearing in which they have testified, I use there testimony pressing the potential of perjury.

The comparison questions I use for the aforementioned are lie, falsify, and breaking of other laws.

Did you ever lie to stay out of trouble not discussed?

Did you ever lie to get someone else out of trouble not discussed?

Did you ever lie about anything else to get someone else in trouble not discussed?

Did you ever falsify anything else for personal gain not discussed?

Did you ever break any other laws not discussed?

Just my opinion.

I am not aware of any research on this issue.

posted 08-07-2008 06:22 AM            

Has this given you any ideas? I think you can extract the information here and run a good test. There is more than one way to skin a cat and the end result is generally the same. Let us know how it turned out.

I've only done five in the last ten years. 3 NDI, 1 DI and 1 No Op. The low numbers alone tell you I try to avoid them unless it is really necessary.

Gordon,

I'd like to know what the false positive rate is on testing of victims of violent crimes if you have that information.

posted 08-07-2008 10:38 AM            

quick note. Re: your posting at 5:08pm yesterday. The problem with getting an official statement via the detectives or investigatiors is that you will end up with an elongated opinion filled statement containing a LOT of extra crap you do not want (regardless of your instructions) which will in deed cause false positives, due to the reasons already mentioned.

Anyway, avoidance of these test, as suggested by RGuillford is in fact best. But, if you have to... ya gotta...

posted 08-07-2008 11:25 AM            

This is a case in which a female alleged victim is volunteering for a polygraph to prove her credibility regarding an alleged sexual victimization. The suspect refused to take a polygraph. This attorney, before he takes on the case, wants to have some sort of proof of her credibility. The police were unable to gain enough evidence to pursue the matter in criminal court, and this female is now wanting to pursue the hospital and the suspect in civil court. The attorney stated to me that he does not intend to use the polygraph in court, he just wants to use it to assure himself that this is a case worth taking.

posted 08-07-2008 03:21 PM            

quote:

---

The attorney stated to me that he does not intend to use the polygraph in court, he just wants to use it to assure himself that this is a case worth taking.

---

read: this case will bite you in the arse

Something tells me she would have a stronger civil case if there were a criminal filing.

I've met a handful of nurses, doctors, and other health care professionals who have either assaulted or engaged in some form of sexual relationship with a patient. I think we all know this type of thing can happen. We also know about deep-pocket lawsuits, and settlements.

Attorneys know how to evaluate the merits of a case independent of the polygraph. I would bet your attorney already has an idea about the value of this type of polygraph exam as evidence. If he thought the case had merit he wouldn't need the polygraph.

Seems more likely it is hoped to be a leveraging point in some form of shut-up-and-go-away settlement.

Try this: double your fee, and inform him you will write in the report that the examination is for investigation purposes only and not intended for evidentiary use in any negotiated settlement or court decision (or something like that. Or, double your fee and tell him you won't give him any results in written form. OK, don't double your fee. Your attorney will want the result in writing.

my .02

r

------------------
"Gentlemen, you can't fight in here. This is the war room."
--(Stanley Kubrick/Peter Sellers - Dr. Strangelove, 1964)

posted 08-07-2008 05:27 PM            

Settlements aren't made in courtrooms. They are sometimes announced in courtrooms. Sometimes not.

Polygraph tests get discussed often in courtroom hallways.

Even if its not a courtroom hallway - could be a telephone conference - my point is the same. Anytime some attorney says to me that some polygraph doesn't actually matter or won't actually be used for anything, I expect a subpoena and an conversation with the opposing counsel or their expert.

Try this. Because this such an informal test, offer to cut your fee in half, and provide a verbal report only. Your attorney will want a written report.

The correct thing to do is to approach every test as if it matters, and as if it is intended to serve a decision-support role in a contested court matter.

r

------------------
"Gentlemen, you can't fight in here. This is the war room."
--(Stanley Kubrick/Peter Sellers - Dr. Strangelove, 1964)

posted 08-07-2008 11:50 PM            

Horvath, Frank

Journal of Applied Psychology. 1977 Apr Vol 62(2) 127-136

10 field-trained polygraph examiners (evaluators) made blind judgments of a stratified sample of the polygraph records of 112 criminal suspects drawn from a pool maintained by a large police agency. Correct calls averaged 63.1% (p < .001), 64.1% on records where ground truth was established by confessions (verified), and 62.1% on records where ground truth was not known (unverified) but where the criterion measure was the original testing examiners' judgment. Evaluators' hit rates in both situations were quite similar, averaging about 77% true positives and about 51% true negatives; the variable having the greatest effect on evaluators' errors was the type of investigation from which records were drawn. Reliability coefficients showed high interevaluator agreement on both verified and unverified records, .89 and .85, respectively. (16 ref) (PsycINFO Database Record (c) 2007 APA, all rights reserved)

Digital Object Identifier: 10.1037/0021-9010.62.2.127

It should be noted that all of the original examiner opinions in this study were 100% correct (as noted here http://truth.boisestate.edu/amicus/brief.html).

posted 08-08-2008 05:02 PM         

By way of background, Frank’s study, along with my own dissertation, was seized upon by those critical of the polygraph, because in our studies involving criminal cases, we both had a high false positive rate (about 50%) on the blind analysis of the charts from criterion innocent subjects.

In my field study, the primary criterion for ground truth was a majority decision of 5 panelists who independently examined the investigative information on the criminal suspects I had examined. The dossiers had been sanitized to remove all reference to the polygraph outcome. For example, when a suspect confessed following a DI test, the dossier stated “The suspect told an investigator that….” If 3 out of the 5 panelists agreed on the guilt or innocence of a suspect, that was considered “ground truth.” Of the 11 criterion innocent suspects, 6 were DI on the polygraph. The critics hailed this as proof that the polygraph couldn’t clear innocent subjects any better than the flip of a coin. I consider the criterion an approximation of ground truth, but not the actual ground truth.

There were several indicators of bias towards innocence on the part of the panel. The first was the nature of the material they were provided. Because the study was done on a shoestring budget, we hired an undergrad to gather the investigative material. She was not a professional investigator. Consequently the investigative information was usually inadequate, often short of what would be required for an indictment, much less a conviction in court. The packet didn’t always contain all of the evidence against the suspect, but did always include the suspect’s alibi or other statement of innocence which he told me during the pretest interview. Particularly in cases where the case information was weak, the presence of the exculpatory statement(s) might have had a greater than normal impact on the panelists’ judgments.

The second source of possible bias was the nature of the panel itself. They were all legally trained and oriented. There were 2 defense attorneys, 2 prosecuting attorneys, and a judge. I had instructed them to make common-sense decisions, not legal ones. When I debriefed them following the study, several told me that they could not always set their legal training aside. They wanted to help me by avoiding a vote of inconclusive, but in cases in which the evidence was weak, they tended to vote “probably innocent.” For my analyses, a vote of three “probably innocent” and two “inconclusive” would be tabulated as innocent.

A third source of bias for the panel as a whole was the bias exhibited by one of its members, a defense attorney with no prosecutorial experience. Statistically, his decisions were more frequently “innocent” than any of the other four panelists, none of whose decisions were statistically different from each other. In one case I remember in which the suspect confessed, that panelist voted “innocent.” When I called to see if he had unintentionally checked the wrong box, he said that he hadn’t. The fact that he was an ex-con and had confessed led him to believe that he was “playing games,” perhaps to protect a friend.

One other factor that should be considered when deciding how much weight to put upon that finding is that the total number of criterion innocent subjects was only eleven. Whenever a study has a cell size that small, it is difficult to assign much weight to the result.

Now let’s turn our attention to the Horvath study. Because his study, using a more rigorous criterion for approximating ground truth, independently concluded that blind analysis of the charts yielded a 50% false positive rate, the critics were ecstatic. I was perplexed. I noted that none of the examiners who had blind scored the charts had scored them numerically. I asked to evaluate the charts, myself. I made two predictions about how I would do. One was that by scoring them numerically, I would have a higher inconclusive rate. The second was that when I made a decision, it was very likely to be correct.

Neither prediction was correct. I had very few inconclusives, and a whole lot of false positive errors, just like the previous evaluators. However, I noticed that in many cases, the relevant questions had been answered “yes.” When I received the question lists, I saw that these were confirmatory tests on statements made by victims or witnesses. Frank’s article described the sample population as criminal suspects. When I asked him about this, he explained that at the time of the tests, there was genuine doubt as to whether the statements were correct or not, otherwise they would never have been asked to take a polygraph test. Thus the alleged victims & witnesses were suspected of possibly submitting false statements. My thought was that people claiming to be witnesses or victims are identifiable as a distinct group prior to the examination, and therefore it might be appropriate to treat the results separately if there is reason to believe their results differ from the “regular” criminal suspects.

What is interesting is that the original examiners correctly cleared 100% of the verified innocent subjects. How was that possible? One possibility is that close analysis of behavioral clues might (a) subconsciously influence how the examiner evaluates the charts or (b) make the examiner consciously decide to overrule the chart reactions, correcting what would otherwise be a false positive error. There are other possibilities, too, such as the influence of case facts.

I once had occasion to talk to one of the examiners in Frank’s study. When I asked him how he cleared the victims, he explained that in one of the cases he was obtaining charts that were clearly DI. He took a break, wandered down the corridor to observe the exam being conducted on the suspect in that case. The other examiner was taking the suspect’s confession. To use the examiner’s words, this “made it very easy” for him to clear the victim in the case. (Remember: this was in the early 1970s and the charts were not numerically scored).

Because I never published the results of my review, I no longer remember the precise numbers of hits and misses. However, as a ballpark figure, it was something like twelve false positives, ten of which were on victims/witnesses and two on criminal suspects. Because there were more criminal suspects than there were victims/witnesses, a very high percentage – perhaps 80 percent – of the victims & witnesses scored DI. The error rate on innocent criminal suspects was much lower. I must emphasize that I don’t have the actual figures available, and the results I’m indicating are from a memory more than a quarter century old. It’s possible that Frank may have retained my results in his files.

Peace.