posted 07-01-2008 05:34 PM            

stim 1 of 1

I think Bob said this was a blind stim.

chart 2 page 1 (chart 1 was the stim)

chart 2 page 2

chart 2 page 3

chart 3 page 1

chart 3 page 2

chart 3 page 3

chart 4 page 1

chart 4 page 2

chart 4 page 3

chart 5 page 1

chart 5 page 2

chart 5 page 3

chart 6 page 1

chart 6 page 2

chart 6 page 3

Thanks Bob.

r

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

posted 07-01-2008 06:09 PM            

Anyone want to bother scoring this stinky mess?

I ran it through OSS-3.

Guess the result?

r

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

posted 07-01-2008 06:09 PM            

I got -8, DI from memory it was a single issue. It was kind of all over the place.
That was just a brief look.

I am waiting for someone with a little more time to chime in.

Tks Nels.

posted 07-01-2008 06:11 PM            

C2
R-I-C
I-C-R

C3
R-I-C

C4
R-I-C
I-C-R

C5
R-I-C
R-I-C-R

posted 07-01-2008 06:31 PM            

Utah Zone

Introductory Question
Sacrifice Relevant
N1
C1
R1
N2
C2
R2
N3
C3
R3

Rotate the RQs and CQs, on each chart. In this case the Ns are also rotated.

r

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

posted 07-01-2008 06:36 PM            

You may have been looking at the first set of graphics. Parts of those images were truncated.

I fixed them.

r

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

posted 07-01-2008 09:00 PM            

NO Opinion schedule chat with examinee re:CMs

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

posted 07-02-2008 11:01 AM            

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

posted 07-02-2008 11:35 AM            

What can you tell us about his demeanor during the test. Was he gamey? Cooperative? Superficial? Obsequious? Calloused? Arrogant? Mesmerizing? Dumb? Articulate? Amiable? Despicable?

Its a great case for illustrating more about OSS-3 and computer scoring algorithms.

OSS-3 is not intended to interpret uninterpretable data.

Kircher & Raskin (1988) describe removing artifacts from the data on which they performed feature development, using discriminate analysis and all-subsets regression methods. So, there is no reason to assume that Kircher features would be robust against artifacts (any more than any other feature set), and there is good reason to assume they are not. Features are ultimately just measurement numbers. Alter the numbers, and you alter the result.

For that reason it is a huge mistake to use a computer scoring algorithm in a click-and-score manner. Lafayette's long awaited OSS-3 includes a nice chart reviewing feature that makes it rather quick to review every question for interpretable data. Just mark anything data that don't pass the sniff-test, then score.

When Barry talks about computer assisted scoring, this probably what he means. It is irresponsible of our profession to limit our scoring models to arcane, simplistic, and unjustifiable methods - simply because we understand them, and are afraid of fancy math. The rest of the scientific community is not afraid of advanced math, and the smarter folks will think we are rather bone-headed for neglecting to learn about things like alpha thresholds, p-values, the inflated alpha problems associated with multiple simultaneous significance tests (spot-scoring rules), the importance of omnibus methods like ANOVAs, and other solutions such and Bonferonni or Tukey.

So, there is no need to cause ourselves an aneurysm with the hard math - just let the computer do it. However, it is also important to remember that we, the examiners, are in control. The computer at present, cannot tell good interpretable data from artifacted and uninterpretable data. We do that. Then let the computer do the heavy lifting.

Here is the result from OSS-3 after marking the artifacted and uninterpretable segments, and movements.

Don Krapohl has suggested to include a message, under INC conditions, to collect additional test charts if possible. I agree. There is evidence to suggest that additional test charts can reduce INCs, and increase accuracy, though I don't think it will help in this case.

This result is INC because the Test of Proportions (google it) indicates the pattern of attacks at various questions is different, at statistically significant levels, from that which we would expect if the activity occurred randomly. Non-random attacks are strategic attacks. In this case the odd-ratio is 999 to 1 that these attacks are strategic and intentional (non-random).

This subject attacks a number of the Neutral questions in addition to the CQs.

You can mark those too, in the OSS-3 tool.

Paul Menges, at DACA, thought of using the Test of Proportions in this manner, before it occurred to me. While we were incorporating it into the OSS-3 algorithm, he and Dale Austin suggested that including all of the other questions in the RQ group would increase the power of the Test of Proportions, when the subject attacks the CQs. They are correct. So, the default condition is that behavioral artifacts at CQs are evaluated against behavioral artifacts at all of the RQ and other questions together. That is a complication when the subject attacks the Ns. So, there is an option to not use all of the other questions, and compare behavioral artifacts at CQs only to artifacted RQs. In this case it makes no difference, because the subject attacks the CQs so persistently.

All of this you experienced examiners can tell by looking at the charts. What we have done is to use common statistical methods for verifying that - so that it can be mechanically replicated using measurement based features. What is still missing is an automated process for IDing those artifacts with a measurement and statistical procedure - we're working on that.

When the Test of Proportions is significant for non-random attacks, the default condition is to force the result to INC, because we all know that truthful people sometimes do stupid things. There is an option to also allow an SR result when the Test of Proportions is significant. (NSR results are not allowed, though there is an option to not use the Test of Proportions.)

Here is the result with the Allow SR result option.

The result of allowing an SR result when the Test of Proportions is significant for deliberate distortion, is a reduction of INC results, increase in sensitivity to deception, and some anticipated increase in FP errors.

The first concern should be whether we have enough interpretable data to render a valid and reliable opinion. If the data are uninterpretable, the second concern is, of course - why not. If the artifacts are deliberate, we do not really know why, except that the examinee has some 'splainin to do. If the observed artifacts are not significant, and there is still sufficient interpretable data, then the correct thing to do is to score the test to an opinion, collecting addition charts, if possible, when the scored results are INC.

Here is the measurement and transformation tables. You can see the values I marked as artifacts. At times, I did not mark both pneumos. That is in attempt to still achieve a score. Too much missing/artifacted/uninterpretable data will case the algorithm to force an INC result, because scoring nothing is inherently unreliable.

You can see that the third and fourth charts were not scored. This is because the CQ pneumo data were worthless, and there are only one EDA and Cardio CQ measurement to score. OSS-3 requires at least two interpretable charts (three must be collected), two interpretable components, and two CQ values. Using a single chart, single component, or single presentation of a QC violates the testing/measurement principle of taking multiple measurements under similar conditions.

Anyone want to guess how this test would measure and score under blind click-and-score procedures?

r

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

posted 07-03-2008 11:10 AM            

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

posted 07-03-2008 11:41 AM            

How about just that? For example, you can say that there were a great number of artifacts, and you marked and calculated (or you used software to calculate) the odds that such artifacts (which you'd also explain or expand upon somewhere) occurred by chance. You can then cite the statistic and your conclusion (i.e., they aint accidental). Then you could add that if those artifacted data are not considered - as they should not be - there are significant reactions to the RQs, meaning you can't conclude that he didn't do it. I think that answers the original question as best as you can. (You can also cite the statistics with the limited data and state how that should be considered by the reader.)

Obviously you can't just say you scored the charts and say DI, NDI or INC. You can, however, articulate what you see and know and leave it at that.

posted 07-03-2008 01:00 PM            

An algorithm is simply a defined procedure, whether handscoring or computer scoring. Our procedures require an assumption that we are measuring autonomic responses. That assumption cannot be fully supported in this case. Therefore, it can be argued that this case cannot be scored for truth or deception.

For this reason, the default setting for OSS-3 is to for the result to INC when the Test of Proportions (for random occurrence of artifacts) is significant.

I tend to regard those reactions to he RQs as autonomic, while some of the reactions to the CQs are arguably a result of behavioral activity.

The Test of Proportions can score this test, not for deception, and not for the examinee's motive or intention.

The Test of Proportions compares the occurrence of artifacts at CQs to the artifacts at RQs, and tests the significance of difference between that pattern of occurrence and what we would expect if they occured randomly. (null-hypothesis = no different from random). When the probability value for is very low that the data fit a random pattern, we infer they are non-random or intentional/strategic (that's why we call it inferential statistics).

The Test of Proportions, in OSS-3, is a CM algorithm within the algorithm. With a p-value of <.001, you have an odds ratio of 999 to 1 that the pattern of distortion is deliberate.

The Test of Proportions does not score the data, it scores whether an when the examiner marked an artifacted an uninterpretable segment. It does not care why the data are uninterpretable - only that the examiner marked it.

In the end, scoring to DI, if possible, or reporting INC is both a value judgment and matter of policy. Its not wrong either way, but field practices should be aligned to best serve the information and decision-support needs of the referring agency or professional. That is why OSS-3 includes the option to score to DI/SR if possible, even when the ToP is significant.

.02

r

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

posted 07-03-2008 03:16 PM            

I would use N.O.
What I would do is articulate the reasons I was unable to render an opinion. In other words there appears to be a pattern of artifacts throughout these charts rendering a fair interpretation of the test data impossible. Based on tracings provided by a movment sensor placed in the examination chair,and the artifacts location and frequency I have concluded that these artifacts are most likely the result of an attempt to deliberately manipulate the results of the exmination.

Changing the subject I noticed that this guy pulls the "pucker cord" in time with his answer. On some of the C.Q.s this seems to interfere with what may be an honest reaction. On several of the RQs there appears to be a loss of homeostasis prior to stimulus. Anticipatory reaction or just coincidence?

The way I look at it, CM's just add fly feces into the pepper shaker. We always take a risk when we don't know for sure whether or not it's there. But if we know it's there we're probably not going to sprinkle it on the potato salad and eat it.

Ahh I see I've forgotten to eat lunch again.

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

posted 07-04-2008 08:57 AM            

"Inconclusive" is an adjective which describes the test result. inconclusive is not itself the test results

An "opinion" is a noun, the quality of which is unknown without addition information in the form of descriptive numbers or adjectives. To say there is "no opinion" is to say there is "no noun" - there is nothing. What there is in actuality is an uncertain result. "Result" is a noun. "Uncertain" is another adjective. "No opinion" means, linguistically, that there is nothing, no noun, no opinion, no Volkswagen beetle, no rubber-ducky. Its all the same. Nothing.

The point at which we sound stupid, is when we forget that "inconclusive" is simply an adjective, not a thing, person, or place.

When we allow ourselves to become materialistic and concrete, you'll here us examiners say things like "he got a no-opinion," as if "no opinion" were a noun (person place or thing).

The reason "inconclusive" and "no opinion" are the synonymous, in the practical sense, is that "inconclusive" means there is "no opinion."

In the past, I had assumed that the reason "no opinion" emerged was due to the tendency for field examiners to cave-in to pressure to provide something helpful to the referrer, even when the test result is inconclusive (descriptive adjective). I've been at ATSA conferences and heard therapists rattle off poly-jargon like "inconclusive leaning toward deception," and "inconclusive leaning toward truth," with such ease and familiarity that I assume it is part of the consultation lexicon with their polygraphers.

In reality, this field-practice jargon (polybabble), represents a serious breach of professionalism. Test results are the product of testing, measurement, and math. Its not mysticism, or magic. Its just stimulus and response - just like like nearly every other kind of testing. The jargon "leaning toward..." is the substitution of a personal opinion, when the test result is inconclusive (adjective).

I like "no opinion" because it corrects the tendency to want to substitute a personal opinion (unsupported by data) when the result is inconclusive.

In written language, "no-opinion" looks silly. You'll see phrases like "when the results are no opinion" or "when there are no opinion results," and "so-and-so's test results are no opinion." Its just awkward, bad, and excessively concrete materialistic language usage.

No one talks like that or writes like that outside of the polygraph profession.

To say "so-and-so's test results are no-opinon" is equivalent to saying "the test result is no smelly socks," or
"no paisely neck-tie." Of course they are not. But the results are also not nothing. The results are inconclusive (adjective - describes a noun). The noun is either the opinion, or the result, or perhaps both. But there is certainly a noun (thing) somewhere, in the form of a result - which is uncertain or inconclusive.

inconclusive

uncertain
http://www.merriam-webster.com/dictionary/no%20opinion

opinion

"Inconclusive" simply means that the present data, and present state of the science do not support any particular conclusion and that the examiner can render no opinion regarding the results. "Inconclusive" and "no opinion" are not themselves a thing (nouns) or result.

I know there have been conversations among some examiners about stipulating the difference between "inconclusive" and "no opinion." None of those conversations are based in science. They are based in either 1) someone's ego-driven need to be an "expert", 2) someone's ego-driven need to be an authority, or 3) someone's ego-driven insecurities when facing uncertainties and ambiguities. There are a lot of remaining uncertainties and ambiguities in polygraph, and every other field of science. As professionals, we simply have to tolerate them, seek real answers based in the scientific study of data, and resist the tendency to make up answers just because we want them now. We also have to resist the influence of people who would offer to relieve us the anxiety surrounding our uncertainties and ambiguities, by conjuring up and selling us simplistic and unscientific answers and unscientific explanations - simply because we want the answers now, or because someone wants an authority figure to be in charge of all the remaining uncertainties.

Adopting unscientific solutions, excessively concrete jargon, and pretending to have answers which we do not have will probably cost us in the long game.

Happy 4th everyone.

r

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