posted 06-01-2008 09:35 AM            

r

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

posted 06-01-2008 12:01 PM            

I was hunting for that during the test. My cardio system does not generally leak. Pinch was on, and the cuff was set. I snugged all the connectors before the first chart and it cleared up. Graphic later.

r

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

posted 06-01-2008 04:06 PM         

Okay, I’ll jump in before asking a lot of context questions. I enjoy the challenge of blind stims. I am assuming the person must be lying to one and only one of the 5 RQs. I’m further assuming he knew the sequence in advance. Both of these assumptions are basic to my analysis.

I’m eliminating R5 & R2 from further consideration because of the relatively minor GSR amplitudes. I like R6 because it’s the biggest single response on the chart, it’s bigger than the preceding C7, and there’s a nice fall-off (my own unscientific term) on R5.

I also like R4 because of the anticipatory responses in GSR and cardio (both start just prior to the deeper-than-normal breath, so could not entirely be a physiological response to that). The GSR amplitude at R4 is smaller than I would expect, though. Parenthetically, it’s interesting to see a movement after R4 on the piezoelectric sensor that didn’t show up on the pneumatic sensor. Do you find that happening often?

I also like R3 because it has a significantly higher GSR amplitude than the preceding question, followed by a nice fall-off on R2. The amplitude is especially impressive coming so late in the sequence, though the C1 amplitude is surprisingly high.

If I had been conducting this stim, I would have continued on by repeating R5 (which I’d eliminated; repeating this absorbs the orienting response from its being the first to be repeated) followed by R3, R6, and R4, in that order, before making my decision.

But based solely on the chart presented, I’d go with R3 as my 1st choice, and R6 as 2nd choice.

I don’t see any decrease in lability in the chart, so if I were a betting man - which fortunately I’m not - I’d suspect he is more likely to be DI than NDI regarding the main issue.

After a healthy wait for others to offer their interpretations, I’ll look forward to seeing the ground truth, and hearing why you selected this chart for discussion.

posted 06-01-2008 05:05 PM            

qoute:
___________________________________________
I also like R4 because of the anticipatory responses in GSR and cardio (both start just prior to the deeper-than-normal breath, so could not entirely be a physiological response to that). The GSR amplitude at R4 is smaller than I would expect, though.
____________________________________________

posted 06-01-2008 05:07 PM            

It's an offender who has had 15 or more previous polygraphs.

R6 was just too easy. I like R3. However, R6 is bigger. R2 after R3, is more diminished than R5 after R6. R6 was just too good.

I didn't call it, and proceeded to the test. Afterward, stood up (from sitting on the number) and said "three."

r

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

posted 06-01-2008 08:06 PM            

The cost-benefit ratio is all out of whack with this type of test.

Not that big a benefit if you call it right but disaster if you call it wrong!

There good stims that are not blind and DO NOT involve trickery (another pet peeve of mine)

Chip

posted 06-01-2008 08:09 PM            

We've had this conversation. Are you trying to get stat going, or did you miss it?

posted 06-02-2008 09:09 PM         

First, Ray, what are your observations comparing the piezo-electric versus pneumatic movement sensors? Do they pick up different types of movements? Is one more sensitive than the other? Is one more useful than the other?

Second, regarding my blind numbers test article, keep in mind it was written decades ago during the era of the analog polygraph. I shortened the interstimulus interval to roughly 10 seconds. Nowadays I've had to shorten the procedure (make the decision from less data; more risky) because my Lafayette won't let me ask the next item in less than 25 seconds (If anybody knows how to do this, I would welcome the knowledge).

After a fully successful numbers test, I still like to comment how even a trivial lie shows up, because I think (hope) it helps focus the attention of the innocent subject on the comparison questions.

Peace.

posted 06-03-2008 12:46 AM            

In my view, the piezzo sensor is overly sensitive to high frequency activity and insufficiently sensitive to low frequency activity. If I turn the sensitivity (really a display zoom) on the pneumatic sensor way way up, you'll see some indication of what ever fired the piezzo. Muscle activity and behavior is low frequency activity. To me, those fast firing events don't model the activity correctly.

I've been using the same sensor for a few years now with no issues. It runs on residual air pressure - pull it out of the case. Connect it, and it works. Its not pressurized and as thin as the piezzo.

I've been running both for a while now. The piezzo seems noisier to me.

I like to see the normal respiratory oscillation, because that means the senor is working properly. No-body ever sits perfectly still.

In rooms with dirty power, I can sometimes fire the piezzo by taping my foot on the some cheap old nylon carpeting.

The piezzo, as far as I understand, is a ceramic crystal that produces an electrical current when moved/bent, or will vibrate/bend when a current is applies. Because they produce a frequency of their own, the piezzo will require filtering. My concerns could be addressed through more effective adjustment of that filtering.

The Lafayette will show more high frequency activity than the Limestone, and does a slightly better job showing the low frequency respiration data. However the high-frequency activity is still a problem for me.

This is all just my .02, and I am aware that all of the sensors work well. We are better off with them than without them.

Imagine using an f-test of variance, in a step-wise manner, to evaluate changes in the rate of activity. The high-frequency activity begins to define the normal rate of activity. If the filter is insufficiently sensitive to low-frequency data, and if the data when the sensor fire is extreme, then actual movements will sometimes produce standardized values in excess of 10 standard deviations. In most mathematical models, these would be regarded as outlier values, but in this case we supposed to regard those extreme values as data. In visual analysis it makes no difference, but if we want to develop and train an algorithm to detect an artifact then its better if the data behave more typically.

The pneumatic sensor is ideal for low-frequency activity (e.g., muscle activity, respiration, etc.), and works without any filtering.

----------------------------

I'll put up today's number test in the morning.

The most interesting thing about the present numbers test was, for me, the large and misleading reaction to #6. I would have chosen #3, because of the fall-off response to #2 compared to that at #5. #6 was just too good and too easy, and that' my unquantifiable gut feeling.

Also, I frequently observe large reactions to the last item. I suspect this is a novelty or dishabituation effect, because the subject knows it's the last item. I haven't completed it yet, but I'm trying to collect enough data for a chi-square or canonical correlation of the first, middle and last positions, with the notion that they produce significant position effects of their own. I generally add a neutral to the end of an AFMGQT in PCSOT exams and offenders (who are non-naive subjects) routinely react to it. Part of the novelty and dishabituation might be the 'yes' answer, but in the CIT all answers are 'no,' so that suggests it is the position (last), not just the answer that contributes to the effect.

Barry is correct that the C1 and C7 tags are funny. There is nothing else odd. The really should be Ns. I started out with Cs because stat suggested come of the algorithms wanted at least on CQ. The Miritallo Rank and Response Magnitude algorithms are set to score CQs or Ns, but that can be changed. I'll fix the templates.

The offender is a smart guy, who works as a driller on a gas rig. He was a union meat-cutter before probation. He's had 15 to 20 or more polygraphs, since 2003, along with a handful of POs and a bunch of therapists. One of his former therapists suicided. He'll tell you that he feels a lot of his treatment to be arbitrary. He'll also tell you he does not want to go to prison. Some offenders are smart enough to simply do whatever they are told. (If we told them to drink battery acid every morning, they'd probably think about it.) There is sometimes no way to get completely beyond the superficial things with these guys. I assume he's had enough polygraphs that somewhere along the way he has more than likely experienced a test error.

r

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

posted 06-03-2008 09:21 PM            

I like it. Ray's right to suggest that in theory, if a blind stim won't work on a given examinee, than it isn't far fetched to ponder that the examinee might be unfit or a likely error subject in the waiting.

On the positive, and I direct this last comment to the wonderful antipolygraph blogger Jim Sackett----that TLBTLD strictly prohibits "messing with" a stim test. That is, the lastest incarnation/edition of it anyway.

Thanks Ray for posting your experiments. Please also give commentary as to if and when you begin PRETEST gentle challenges/confrontation regarding your stim results to the examinee---and the what I've experienced anyway---WINDFALL of additional 11th hour pretest admissions as a result of the amazing "expectations-crushing polygraph effect" of the blind stim on sex offenders.

As Dr. Barland stated so well, the blind stim in theory is a powerful aid to the innocent person as well as a heart stopping red light to guilty individuals.In my most sincere advice, I wouldn't have used it for 4 years straight on first-time examinees (1st time with me anyway) if I wasn't convinced it was a very effective tool.But, as Ray will soon see--I might have been very wrong. I doubt it though.

posted 06-03-2008 10:03 PM            

r

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