posted 10-07-2004 06:03 AM            

When I explained it to another examiner he called it a $1500.00 stop-watch, a term I've now stolen because I think he's right.

The software seems to have glitches as it doesn't always display graphically how it arrived at a conclusion.

There is another problem though. I read of a study (I think in James Wygant's newsletter) saying it takes longer for a person to respond when he's lying than when he's telling the truth. If that's the case, then a truthful person will take longer to respond to the comparison questions than the relevants, which should lead to a CM decision. And that is consistent with what I've been seeing.

It does make a nice prop for the pre-test when I explain the instrument. After all, with that and the motion sensor no one should be able to get away with CMs.

(The system has two microphones on it: one for you and one for the examinee. They are supposed to pick up when you are talking and when the examinee is talking, mark the charts and do its computations. No matter how high I set it, I still don't always pick up the examinee's answer, so I constantly have to tell him to speak up, which is a pain.)

posted 10-07-2004 09:35 AM         

Thanks for the reply. I attended the APA 2000 seminar when Emmanuel Cohen and Chris Fausett gave a presentation on the system and design. They discussed the ‘overall' accuracy of the countermeasure detection system as being 90.7 %., although the tested population was small (N=37).

I've been using the system since August, and every test I've conducted (although a relatively small sample +/- 30) has been evaluated with CM's being utilized, which causes me concern. However, I predominately test sex offenders, and frankly they are a slippery bunch. All were numerically evaluated (and in agreement with Polyscore decision) as either DI or INC, except for two which numerically I would have called NDI (and also in agreement with Polyscore).

I was mainly curious to know whether or not any other examiners who have been using this system has conducted any exams with "No CM's" being detected by the system, and their experience with the system.

I too have noticed the ‘glitch' in the disappearing chart ‘bar graphs'- and I have also noticed that if the Charts for evaluation are selected "out of order" (for example in reverse order) a different result will occur. Considering this, I was wondering if the "Overall" result would be ‘better' obtained by evaluating each chart "independently," as opposed to the "combining" of the polygrams.

posted 11-20-2004 07:31 PM         

I hate to say it guys but I think buying into this is only going to add fuel to the CVSA fire. The polygraph as we know it today is based on the automomic nervous system response. If we start including voice delays and other such stuff, it will only take away from what we have acomplished so far. Unless a relationship between voice delays and the ANS can be established, I think we should leave this one alone.

posted 11-20-2004 10:18 PM         

I respectfully disagree with your post. I do not see ‘it taking away from our accomplishments', but rather adding to.

The voice countermeasure detection system doesn't have anything to do with the qual-quant numerical eval for NDI/ DI decisions (although an examiner may choose to render a No Opinion if CM's are detected- and is what we do now when we see ‘finger-tapping', ‘toe curling', controlled breathing, or a variety of ‘other' things).

It's just another method to detect the use of mental / physical CM's only. I think we would all agree if it takes an examinee three seconds to answer a question- then something is interfering with his thought processing. The system is merely measuring the time delay on a more ‘micro level.' Research on ‘multi-tasking' (also called ‘dual tasking') supports when someone is engaging in multi-tasking, their response time is slower- therefore easily defended from my perspective. Talking on the cell phone and driving your car at the same time, is ‘dual tasking',and I believe everyone would agree the driver response time to avoid a collision is slower. We are simply determining if the examinee is ‘talking on the phone' to mentally disassociate when he is taking our test.

The major problem with polygraph IS, as I see it, the inability to reliably detect counter measures. The use of Cm's was one of the major concerns expressed by Nat Acad of Sciences in their published report. We all say we can- but ‘proving that Cm's was or was not done' is a little more obscure and difficult(and I notice no one has taken George's challenge yet to make a chunk of change).

The profession however is rapidly approaching the ability of reliably detecting Cm's with the advent of piezo arm/ seat physical-movement sensors, and possibly very soon the ability to detect mental ones too (once the software 'flaw' is resolved that is).

Bob

posted 11-21-2004 08:03 PM            

The people who wrote the NAS report know nothing about polygraph testing. Not one person on that committee ever attended a polygraph school or ran a single polygraph test. Our profession is purely academic to them.

Whether countermeasures pose a serious threat to our accuracy is something that we should continue to discuss, but let's not waste time factoring in the opinions of people who are just plain ignorant about polygraph.

Lou

------------------
Louis Rovner, Ph.D.
Rovner & Associates
LouRovner@sbcglobal.net

posted 11-22-2004 12:08 PM            

I'm glad I you mentioned this.

Let me add this though: I have nothing but good things to say about Lafayette and their service. Aside from the CM system, which I knew nothing about when I bought it, they have been great.

Did he say how they planned to fix the software? I'm still concerned since a truthful person should be lying to the CQs, so it should take him longer to answer those questions anyhow. How will they adjust for that, if you know?

posted 11-22-2004 06:34 PM         

First, I agree with you in that I have nothing but good things to say about Lafayette, they have been great to work with.

Although the detection method became available in 2000 (I believe), keep in mind there has never been any "comprehensive and validated research studies" on the method, such as by DoDPi. Lafayette had been trying to get DoDPi to conduct research- but ‘money' issues relating to financing for the research gets in the way.

As far as I know Lafayette wasn't aware of any problem until sometime after September 20th. I talked with Chris on that date, and expressed several questions I had about the ‘inner workings' of the software, as well as expressing my concern about the accuracy. Chris commented to me that a Federal agency (DoDPi ?) finally expressed some interest in maybe conducting research on the program, but wanted someone from Layfayette to ‘physically measure' the response time intervals from a data set. During that meeting with Chris, I also requested information about how to contact Cohen, in an effort to learn more from Cohen directly about ‘how' the algorithm worked (minus any trade secrets of course). Chris provided the email address but gave a forwarning that getting return email from Cohen was unlikely as they had experienced difficulties too. He said he would have his software developer also speak with the Cohen's staff developer as well(because of better rapport), in an effort to address some questions. On Septemer 30th, I sent an email to Cohen (refer to an earlier post regarding the questions I had for Cohen) but never received a response.

Chris was at the Indiana Polygraph Association meeting on Nov 19th, and before I even had a chance to show Chris some stats that I had been keeping in relation to the voice Cm detection method, he told me there was an error in the software design. He explained the ‘error' resulted when Cohen's software developer did not ‘accurately' input the algorithm into the software as intended. Chris related they have discussed the ‘error' with Cohen's staff and it's Cohen's software developers who has to ‘fix the software.' Chris said they are hoping to have it resolved by sometime in January, but are dependant on the speed of Cohen's staff to get the software ‘fix' done.

Regarding your question, "since a truthful person should be lying to the CQs, so it should take him longer to answer those questions anyhow. How will they adjust for that, if you know?" ; I don't know either and was one of the questions (Q4) I had for Cohen in my email to him that never got answered.

Hope this info is helpful;

posted 05-03-2005 02:25 PM            

Have you heard any news on the software fix? Chris from Lafayette told me one would be out soon way back when we originally discussed it. I see Lafayette has new software out (9.9.1, or something like that), and it has a CM bug fix in it, but it doesn't look like it fixed the scoring algorithm yet.

Lafayette wants $20.00 for the update because it has the new Nero software that allows recording to CD/DVD, but I don't need it or want it. I doubt they can ship me the new software without it, but I've asked.

posted 05-05-2005 12:21 PM         

Sorry, I didn't catch your post when I went on and off Polygraphplace yesterday. I just recevied the latest version of Lafayette software(9.9.1) and have it installed- some nice changes I think by the way.. According to the 'Read Me' file, there is 'no mention' of a fix to the CM 'missing bar graphs' on the scoring report or to the scoring algorithm itself. So that I cannot answer for sure.

I called Lafayette and told them I didn't want to 'download' the new software because of my slow modem time (no dsl here), and they sent me the CD for installation (no $20 required). However, when you install the new version, it will ask for a Nero serial number. I dug out the one already on my system and everything seems to work Ok. BUT-I don't use Nero either at this time to 'burn' disks. Sometime in the future if I start to 'burn' the files to CD/DVD, I will probably have to go to the newer version of Nero.

posted 05-05-2005 12:41 PM            

FYI, Kari emailed me yesterday and said she'd send me the new software, but there is a glitch. She said they stopped production of the current fix (yours?) because they found another bug, but she didn't say what it was. The fixes fix should be out by the end of the week she thought.

posted 05-05-2005 08:48 PM         

The 'fix' was for new users of the LX4000- and those who were not using ver 9.8 of software. Apparently the 'glitch' in 9.9.1 had something to do with the software not communicating with the USB port very well during first time installation.

Used it today and it worked fine (so far). As a side note- I also used the 'on the chart' video screen. I placed a small computer camera on a shelf directly in front of the subject- zoomed in on the face- and great way to monitor the eyes and mouth movements.

posted 05-06-2005 11:55 AM         

FYI, the new software 9.9.1 did not 'fix' the occasional phantom 'missing bar graph' that occurs on the CM report.