posted 02-16-2006 01:39 PM            

I have a test that I would like some review on.

Nothing formal is necessary, but I'm interested in what others think about this test.

Here is a link to the charts.
http://www.raymondnelson.us/c/021606.html

This is another interesting case (actually they are all very interesting). I'll provide more information later.

you can login using:

user: polyguest
password: torquemada

only users of this forum have access to this login.

People should beware that last time I allowed open access to the secured area I got login attempts from the Netherlands.

posted 02-17-2006 04:25 AM            

are you with me?

Dario Karmel

posted 02-17-2006 06:53 AM            

This test is, unfortunately, concerning a former Sheriff's Department employee, now facing felony charges.

This test looks not to bad to me, except for the fact that he has a heart mumur or prolaps condition that causes occasional extrasystole events in the data.

posted 02-17-2006 03:57 PM            

I currious about what features signify the mental imagery?

posted 02-17-2006 09:28 PM            

C1-Pg2 (R5-C8) 5-8?
P 0
G 0
C -1

C1-Pg3 (Q?)
Nothing to compare against; some GSR activity just before the end of the chart

C2-Pg2 (R5-C8)
P -1
G 0, Cut off start of GSR for Q5
but not affected
C +2

C2-Pg3 (?)
End

C3-Pg2 (R5-C4)
P 0 Change in vol @ C4, not enough
cycles from answer to evaluate
G +2
C 0 Extra systoles in both, not
enough chart after C4 to evaluate

C3-Pg3
P Holding & blocking just before and
after C6

END.....

[This message has been edited by Poly761 (edited 02-17-2006).]

[This message has been edited by Poly761 (edited 02-18-2006).]

posted 02-18-2006 05:14 AM            

my scoring:

chart 1:
R5: P (0) , G (1) , C (1)
R7: P (0) , G (1) , C (0)

chart 2:
R5: P (-1) , G (1) , C (1)
R7: P (0) , G (1) , C (1)

chart 3:
R5: P (0) , G (1) , C (-1)
R7: P (1) , G (-1) , C (1)

total scoring: R5 (3) R7 (4)

if there no was a CM (check the video) i think that is a NDI

posted 02-18-2006 06:07 AM            

That is a NDI score. (You need a +4 total for NDI.)

Poly761,

I don't understand what you're scoring. Why do you compare the SR to anything? It's not scored. Also, for each relevant (R5 and R7) did you compare to the largest adjacent CQ component for each question? That's how you score a Bi-Zone.

posted 02-17-2006 09:28 PM
--------------------------------------------------------------------------------
This is my scoring and comments after reviewing your charts.
C1-Pg1 (SR3-C4)
P 0 Controlled breathing (C/B) apparent
G -1 Response prior to identified C/B
C 0 Affected by C/B

Let's just look at that one. (Let's pretend the two should be scored.)

How do you get a -1 on the EDA? The response on CQ4 is larger (greater amplitude) and more complex (which is a tie-breaker or point addition).

posted 02-18-2006 07:36 AM            

I've posted the Identifi computer score at
http://www.raymondnelson.us/c/021606_ID.html

I've seen much better examples of controlled breathing than this. In fact I would be reluctant to try to prove it.

There will always be variabiliy in respiratory rate. In this test, you can see the upper and lower Pneumos remain synchronized and stable throughout. I'm much more bothered when I seen spreading or converging pneumos - which can result from slipping sensors or the subjects attempt to voluntarily regulate his abdominal and thoracic expansion/contraction.

The Cardio anomallies are interesting. This subject is diagnosed with a heart murmur (mitral valve prolapse), which allows leakage back into the left ventricle upon cardio contraction. The result is those extrasystole segments when his heart skips a beat (or whatever) in attempt to reset the closing action of the valve.

Following the skipped heartbeat, there is always a few seconds of rise in BP, then the normal trend can be observed. Its entirely involuntary. Axciton has a feature to remove these spike, but that also adulters the data. I've seen this before. I'll find some other examples.

The exam is recorded on video, and there is no finger or hand movement.

Interpreting PVE or extrasystole charts seems somewhat speculative to me, as there are no published guidelines, that accurately reflect the cause and effects of this anomally. I ignor the first few seconds of tracing after the skipped heartbeat, then look at the trend of the segment after that. Of course, this condition causes some data to be uninterpretable.

You can also see the CM Sensor, at the bottom of the chart, which reflects only normal oscillation from breathing (I like to see that as no-body ever sits perfectly still, and the respiratory oscilation tells me the sensor is working).

I get NDI scores from these charts.

posted 02-18-2006 09:00 AM         

I printed out your charts to take a 'look see'; but there is an error in the chart displays. What you have listed as Chart 1- Page 2; is the same chart segment as what you have listed as Chart 2-Page 2.

Scoring therefore should be the same for both charts :-) and unforntuately not accurate overall for a decision :-)

posted 02-18-2006 11:32 AM            

I had asked for the type of exam used as all I was receiving were the questions I evaluated. I wasn't able to scroll across. I do not include "scoring" of a SR in any total(s).

I was instructed to score the GSR 8-10 seconds prior to the start of the question. I don't evaluate anything during and especially nothing after a question is answered. With SR3-C4 I agree there is good sympathetic, para-sympathetic activity & duration at C4, but, too late. (If) a comparison during and after this question is evaluated there is not enough ratio to use which only leaves the "late" & minimal sympathetic/para-sympathetic activity.

I would not score the cardio as the increase in pneumo volume, in my opinion, created the increase/decrease in heart rate and increase in BP. This distortion does not allow for proper evaluation of SR3-C4 (if it were proper to evaluate).

After reading C1,Pg2 & C2,Pg2 are the same, I looked at my evaluation and became concerned due to the different scores. I likely violated my own rule of scoring conservatively and pushed a "staircase supression" at C2,Pg2. For reason(s) unknown I apparently ignored the good cardio response when scoring C1,Pg2.

posted 02-18-2006 02:02 PM            

END.....

posted 02-18-2006 03:13 PM            

1 Neutral
2 Symptomatic (or Neutral if you don't use the Sym)
3 Sacrifice Rel
4 Comparison
5 Strong Relevant
6 Comparison
7 Strong Relevant
8 Comparison
9 Sypmtomatice (or Neutral)

Score R5 and R7 to the strongest adjacent CQ, component by component. For example, you could score R5's cardio to C4's cardio if C4's cardio is stronger than C6's. Then, you could score R5's EDA to C6's EDA if C6's EDA is stronger than C4's. In other words, you bounce between to two if necessary.

posted 02-18-2006 08:31 PM            

I've fixed the chart display page 2 from chart 2 was displayed inacorrectly as page 2 of chart 1.

Its now correct.

I was unaware that you could run a BiZone with neutrals in place of the symptomatics.

I would prefer to use the Utah, or AFMGQT, but when I know the people that will QC a test will to see very standard zone formats, with time-barred controls, and relevant questions that are time-delimited (on or about the date of the allegation), that is what I do.

In this case, this test is not intended to resolve the felony allegations, for which the subject was relieved of his duties and has been detained for three months. This test was intended to determine the likelihood of other similar, though unknown issues.

r

posted 02-18-2006 09:35 PM            

Thanks for the information. Unusual to see an exam (Bi-Zone) where so much change is permitted between controls for scoring.

What is the difference between the Bi-Zone and ZCT. Are the relevants at 5&7 in the Bi-Zone used similar to a ZCT, same question (lengthened & strengthened) at R7? Or, are two relevant questions of one issue permitted? Sequence of both is the same except the Bi-Zone has the option of allowing use of an irrelevant/neutral question in place of the symptomatic (ZCT).

I'd probably opt for use of neutrals/irrelevants in the Bi-Zone theoretically giving the examinee more chance to relax during each test.

END.....

posted 02-18-2006 09:42 PM            

I like the bi-zone for some purposes, as it presents a good single issue format, with a robust theoretical framework and good ability to achieve resolution with a clearly defined single issue of concern.

The bizone is not so good for screening, as the time-bars are more troublesome when there is no allegation or known incident, and it is helpful to have more relevant questions to broaden the sensitivity of the test to several (mixed) issues of concern.

r

posted 02-19-2006 07:13 AM            

When reported the Bi-Zone is a "true single-issue test" my understand is (only) one issue can be dealt with, 2 questions?

5. Did you stab Joe?
7. Did you stab Joe in Harry's bar?

Or, can two questions about a single issue (stabbing) be asked?

5. Did you stab Joe?
7. Did you hide the knife used to stab Joe?

END.....

posted 02-19-2006 08:10 AM            

quote:

---

5. Did you stab Joe?
7. Did you stab Joe in Harry's bar?

Or, can two questions about a single issue (stabbing) be asked?

5. Did you stab Joe?
7. Did you hide the knife used to stab Joe?

---

My reasons are that it is conceivable that a person could lie to one of the second questions, while being truthful to the other. This is therefore not a single issue test.

One of the testing principles is to attempt to stimulate a reaction to an issues that the subject has some response potential to. The subject's response potential, is theoretically related to: 1) some knowlege of the issue due to direct involvment, 2) fear of being caught and consequenced for lying, and 3) a conditioned response potential resulting from behavioral involvement in the issue. Other pysychological factors also play a role in the subjects response potential to polygraph quesitions, including the threat value of the issue itself (even if truthful), the complexity of the question language and issue, and other experiential events that cause a subject to focus primarily on a single test questions (prior testing or interrogation experience, having read about the polygraph, or a naive therapist telling a PCSOT subject that only one question matters...) It is assumed that behavioral involvement in the issue under investigation represents the most robust response potential. It is also assumed that if we attempt to stimulate a response, with a robust, behaviorally descriptive, polygraph test questions, and get nothing or not much, then the subject is likely not involved in that behavioral event.

Like other predictive testing and measurement sciences, the polygraph depends upon math - predictive statistics. We're just lucky we don't really have to know much about the math to conduct tests in the field. We count reactions, look for a numerical threshold and make a decision (actually, we offer an empiricallly based professional opinion). Our opinions are empirically based because they are (can be) supported by predictive statistics (statistical significance). That is we can demonstrate (or those pinheaded statisticians can) that there is a statistically significant likelihood that a certain numerical result would not occur by chance. The formulae of interest here has to do with the standard-error of mean-difference, across a bimodal distribution of truthful and non-truthful normative scores (assuming, of course, a normal statistical distribution of both truthful and non-truthful scores). If we assume the distribution of scores to be normal, and if we adhere to commonly accepted practices in the social and measurement sciences, then we assume that 95 percent of test scores will occur with the normal range of the distribution of scores (both truthful and non-truthful). Up to five percent (in a perfect mathematical model) of scores can be predicted to occur outside the normal range, and these will most likely be our inconclusive scores. Human variability is wide enough that we can easily anticipate an inconclusive rate greater than five percent - perhaps double that. So our observed inconclusive rate can be expected to be in the range of 5 to 10 percent.

Now, with a single issue test, this offers a resolution rate (inverse of the inconclusive rate) of 90 to 95 percent - not bad.

When we employ mixed issue tests (in which it is conceivable that a subject could lie to one question while being truthful to another), then we employ vertical decision thresholds (question by question), compared with horizontal decision thresholds (for the set of questions). A sound theoretical model for vertical (by question) scoring would again tell us that test scores exist in an assumedly normal distribution. Only, now we have intruduced the possibility of measurement error and variability for each distinct question/issue (mixed issue tests). In this case the inconclusive rate can be estimated by raising the resolution rate (90 to 95 percent) to the exponent of the number of distinct questions/issues. With two distinct questions, and a per question resolution rate of 90 to 95 percent, our observe resolution rate can be expected to fall as low as 81 percent. Simply put mixed issue tests are more likely to result in inconclusive or incompletely resolved test results.

In some situations the trade off is tollerable, and event desirable - such as in screening situations in which there is no known incident and we want the test to be very sensitive to any issue that we might want to explore further.

In diagnostic testing situations, I believe we are better to leave the Bizone test intact as a single issue test format. It offers a robust theoretical framework for achieving resolution to a single issue of concern.

While we rarely concern ourselves with the underlying statistics, we are negligent not to incorporate the sound principles of other statistical measurement sciences then we become vulnerable to accusations that we practice an isolated or schismatic voodoo science, in a kind of black-box that is not consistent with or aligned with, and does not employ the same scientific priciples as the greater scientific community.

posted 02-22-2006 10:59 PM         

Ray, my numerical eval/decision on your posted charts was NDI (R5 +6; R7 +5; and +11 overall); and I would not have reported CM’s being used (although I suspect a good probablity exists). After I numerically evaluated the charts, I then printed out your Identify Report- and was glad to see my final scoring was pretty close;unforntunately though- Identify and I didn’t agree very well on where the points were earned- particularly in the cardio where I zero’d them when PVC’s occurred just before or during question presentation. (Makes me wonder where I’d stand on a QC by someone else ?) Thanks for posting the Identify report- although I don’t use that program. I do like the capbability of making a more direct comparison to manual scoring which Polyscore does not provide. I noticed Identify gives only one value for respiration; does it combine both Thoracic and Abdominal to arrrive at that numerical value ? Or does it predominately evaluate just the thoracic? as Polyscore does.

Thought I would write some comments about EDA responses and PVC’s.

In regards to EDA, Barry you wrote “On computer tests, the EDA does not lag behind as it does with an analog. The EDA reaction is on-time if it starts at least 1/2 a second after question onset.” True EDA does not lag behind, but Barry I would question 1/2 second as being ‘on-time’ after question onset. Before you chastise me- I recall a an APA seminar Don Weinsten gave a presentation and DodPi considers EDR as timely if the response occurs at question onset (I presume there has been no change in their viewpoint since 2004). Research however indicates the time frame from stimulus onset to the point where the EDR begins (a latency period) is anywhere from 1 to 5 seconds; Edelberg reported 1.8 seconds as being the characteristic value in comfortable ambient temperature. (Maybe this is ‘why’ Polyscore gives a 2 second delay from question onset before evaluating any EDR responses?) I like to see at least a 1- second delay, preferably 1.5- 2 seconds. A response exactly at question onset- or within a 1/2 second- I believe is more of an anticipatory response of the upcoming question or some other cause (again makes me wonder where I’d stand on a QC by someone else ?) If we look backwards from the EDR response 1-1.8 seconds, that is when the ‘thought’ (possiblibly countermeasure? or inside/outside noise, pain or what have you) ‘caused’ that response. One way of checking and determining the latency period for the client being tested is to have him take a very rapid and deep breath in the beginning or end of the chart (or during the acquaintence chart) - then measure the time from deep breath onset to the EDR response, thereby establishing his/her latency period. A very good book I’d recommend: Electrodermal Actvity, Wolfram Boucsein 1992 Plenum Press NY

Ray, I also don’t agree that premature ventricular contractions (pvc’s) are being caused by mitral valve prolapse. In fact I suggest the mitral valve prolapse is not affecting the cardio tracing of the charts at all; mitral valve prolapse leaks blood backward in the heart from the left ventricle into the left atrium during systole- so stroke voume output may be slightly smaller. A prolapse of the aortic valve however, may affect the cardio tracing as the blood leaks back into the ventricle from the aorta after systole instead of the ‘normal rebounding’ of the blood off the closed aortic valve which causes the dicrotic notch we see. A prolapse of the aortic valve, depending on the severity, may cause the disappearance of the dicrotic notch.

Premature venticular contractions are a heart electrical conduction problem and result from either [1] the sinus node of the atrium (the heart pacemaker) intermittantly failing to ‘fire’ for some reason, thereby causing a ‘skipped beat’ in the atrium; since the sinus node does not ‘fire,’ the normal nerve impulse does not travel to the atrio-ventricular node to set the stage for a normal ventricular contraction (can be, but not necessarily, second-degree heart block ) Or [2] the sinus node of the atrium does ‘fire’ and the impulse reaches the atrio-ventricular node, but the A-V node does not ‘fire’ as normal again being a ‘skipped beat.’ (PVC’s can be a ‘serious heart problem’ if a heart blockage is occuring- or benign as it can also be caused by smoking a cigarette, drinking coffee, lack of sleep, cocaine use, or emotional states)

IF the sinus node of the atrium does not ‘fire’ the nerve impulse at all- or the sinus impulse never actually reaches the atrio-ventricular node (it is referred to as third degree heart block), then utimlately the atrio-venticular node (or from other locale within the ventricles) ‘fires’ on its own trying to establish itself as the ‘pacemaker.’ IF the atrio-ventricular node actually becomes the ‘pacemaker,’ the heart rate will be very slow (25-45 bpm).

We can not tell based on our cardio tracings which of the two reasons it truely is (whether the sinus node or the atro-ventricular node is causing the ‘skipped beat’), an EKG is needed to make that determination. Our cardio tracings simply reflect ventricular contractions (not the atrial contractions) and therefore we can only say a ‘normal and rhythmic’ ventricular contraction didn’t take place.

When either the atrial-sinus node or atrio-ventricular node ‘skips a beat’, it creates ‘extra diastolic time’ for blood to continue to flow from the atriums into the ventricles (causing the downward spike on the chart)- which results in overfill/stretching of the ventricles, and when the ventricular contraction does take place there is a signicant increase in ‘stroke volume’ being pumped out into the aorta (the upward spike imediately following- and completing the V-like appearance)

The blood pressure arousal many times seen imediately following a PVC- I speculate may result if the initial increased stroke volume ‘stretchs and distends’ aorta- brachial arteries, and then subsequent heart beats adds additional volume to the already ‘stretched and distended’ arteries causing a temprorary increase in blood pressure.

Aren’t I kind of a disagreeing fellow today ? - must not have had enough of something lately.

posted 02-23-2006 07:03 AM            

That is a very good description of the PVC events observed in these and some other test charts.

I can't claim to know much more about this, except that the subject reports that he is diagnosed with a heart murmur and prolapsed mitral valve.

I think there is a great deal of complexity surrounding these problems, 've consulted with GPs and a cardiac doctor on another case (the syncope case)- medical experts seem to scratch their heads just as much as we do on these things.

Your information seems to make sense.

In the end, I don't score the tracing 4 to 5 secs following PVC events. I've had other charts with such recurrent PVCs that I wouldn't score them.

The electrodermal latency is another concern. I'm uncomfortable with developing anecdotally based (read: ideosyncratic) guidelines about latency, and prefer to stick to the established standards. Strictly speaking the EDA does not "lag" it occurs timely with the stimulus and other reactions - it is recorded on the charts in a offset position on analog instruments (and I believe this offset is different for Lafayette and Stoelting instruments). The Utah rules, if I recall specify an onset latency of .5 seconds. In the absence of a solid empirically based consensus about the meaning of EDA latency, I believe we are wisest to stick to established standards of practice.

This does get complicated with individuals such as with a 23 year-old criminal suspect - who went through school as a special ed student with a low average IQ of 80, borderline intellectual functioning (not mentally retarded, but really dumber than a box of rocks), and a specific learning disorder (neurologically based perceptual-communicative disorder) that includes a receptive langugage lag of four to seven seconds. In this case, these issues are well documented and diagnosed with the aid of neuropsychological testing. His answers are quite delayed. Is that a countermeasure? I doubt he's smart enough. So, what to do? I run the test as normally as I can attempt to explain these anomallies and render a cautious (deceptive) opinion. I won't assert that he's a completely normal polygraph subject. In this case the polygraph was requested by the court prior to sentencing. I don't believe the test results were introduced to the court.

The unfortunate fact is that criminals are simply not always healthy normal-functioning individuals.

r

posted 02-23-2006 07:33 AM            

That, however, conflicts with research. The Utah criteria, which are the most researched (and what DoDPI will essentially be going to once the change is approved), demonstrated that it takes that long (.5 sec) for a person to respond (in the EDA channel) to any question. If such responses were simply ORs as you suggest, then we'd see it in every question, which, in the end, need not influence how they're scored (since every question's reaction should begin the same way).

It takes about two seconds before we see anything in the finger pulse amplitude channel, which DoDPI is just starting to play with, and it takes about .5 sec to see a reaction in the cardio channel. (It takes about 8 seconds to see a chemical reaction in the cardio channel, which means the question was a killer! So if you see chemical reactions in a CQ, suspect CMs, as they rarely occur in CQs.)

To think that any reaction starts immediately doesn't make sense as the stimulus must be processed, and then an electrical signal must be sent out to cause the reaction we see. That takes time, albeit, a very short time.

Here's the Utah scoring criteria when PVCs are present:

"Physiological abnormatlities, such as [PVCs], may also render the cardiovascular reaction if they occur in the scoring window. PVCs are contractions of the left ventricle that occur before the left atrium has contracted and filled the left ventricle, causing very little blood to be pumped into the aorta. This is followed by a relatively long pause before the next ventricular contraction. During this pause, the drop in [BP] produces a distinct downward deflection in the cardiovascular tracing. If two or three PVCs occur within the scoring window, the signal is usually so distorted that it is not possible to score it. However, a cardiovascular reaction that occurred before the PVC may be scored. It is usually also possible to score the reaction if it contains only one PVC, although the subsequent rise in the tracing that is teh recovery from the PVC should not be scored as a reaction."

If I typed that right, you'll see you can score the zone if there is only one PVC in one or both of the questions being compared, but do so conservatively.

posted 02-23-2006 08:05 AM            

quote:

---

BTW, Anti-climax... was coined by Backster.

---

I know that. 'twas a lame attempt at humor.

My daughter told me I'm the most prosaic man she knows. I had to remind her that she's nine, and not allowed to use words like 'prosaic.'

I have the Bell article on the Utah rules (as .pdf) - I could put them up for others if people want.

posted 02-23-2006 09:07 AM            

It does work.

I also have the 1989 security screening report.

r

posted 02-23-2006 01:03 PM            

It's a great program as it scores very much like hand scoring. It's a good QC tool when you can't get somebody to look at your charts right away. I highly suggest examiners get it and use it - not as a primary means of scoring, but as a quick way to look things over before you interrogate or kick a person loose. Within seconds you will know if you should be taking a closer look to see if you missed something.

I've emailed Keith and asked about the above two questions. If he's around, he usually gets back to me rather quickly, so I'll let you know how it scores the EDA and pneumos.

I know only looks at line-lenght, but I don't know how it decides what to look at. I'm guessing it does it the same way OSS does, but I'm not sure.

posted 02-23-2006 03:19 PM            

Identifi does seem sensible to us field examiners. The interface is a bit quirky at times (but those are just programming interface/elegance gripes).

For a really frustrating time, I'm trying to install my polygraph software on a Linux system. (As if I don't have enough to do.) Has anyone done this yet.

I also transfered my examination database from Microsoft Access to MySQL - it has every question I've ever asked anyone (only a few thousand tests), in addition to every test subject, and contact information, test results, medical and background info, and more. It can merge to openoffice.org forms and produce written reports rather nicely and quickly, with questions answers, data, and subject and contact info.

I've put up the Utah rules at
http://www.raymondnelson.us/qc/1999_Utah_numerical_scoring.pdf

and another article on the Utah and DoDPI scoring at
http://www.raymondnelson.us/qc/2000_Utah_and_DoDPI_Scoring.pdf

As Barry indicated, these things should be considered must-read.

Also, note the directory and index were moved, due to site and server changes.

The new index is
http://www.raymondnelson.us/qc/index.html

the login remains the same

user: polyguest
password: torquemada

posted 02-23-2006 03:57 PM            

The EDA rise must occur between question onset and 6 seconds after the end of the question for Identifi to score it. Line-length is measured out 15 seconds, but he didn't say if he averages it or just goes with the largest CQ to the respective pneumo in the RQ, but I suspect that's it as it is supposed to be most similar to a hand score.

posted 02-24-2006 09:48 AM            

To EBVAN: I am also currious about what features signify the mental imagery?