Author
|
Topic: Interesting situation
|
rnelson Member
|
posted 05-13-2009 09:18 PM
Here are some charts from an exam that someone recently showed me. I have the examiner's permission to post this here.Look at the measurement indicators and centering marks at C7. Here is another example: There is not really anything wrong with the system or the software here, but it would be best to avoid centering or adjustment for at least 15 seconds (or whatever is the specified measurement period) after each question onset. Mostly, this is a great example of why, if we are going to make responsible use of computerized measurement and scoring systems, we want to know exactly what the computer looks at and measures - just as we want to know exactly (or at least be able to look it up in some citable published reference) what the 'puter does with those measurements. We also want features that make it easy to check for these things. The new measurement indicators work well. r
------------------ "Gentlemen, you can't fight in here. This is the war room." --(Stanley Kubrick/Peter Sellers - Dr. Strangelove, 1964)
IP: Logged |
skipwebb Member
|
posted 05-14-2009 06:53 AM
This is a perfect example of the answer to the age old question "Why does my scoring software call the charts X when I clearly have them as Y.IP: Logged |
blalock Member
|
posted 05-14-2009 07:45 AM
I wonder if there is a way to have the software prohibit examiners from centering until the kircher window is closed... Kind of like the option of not starting the next stimulus (question) onset until a pre-determined amount of time (20-30 seconds).------------------ Ben blalockben@hotmail.com IP: Logged |
Barry C Member
|
posted 05-14-2009 10:20 AM
That's a good idea Ben. I really didn't see the need to make any adjustments here. Moreover, they were in the scoring window for hand scores as well. It looks like the EDA was on detrend, meaning no adjustments were necessary anyhow. (That might not be the case as there's little data to make the determination, but it does look like it hits a low point and goes no father.)IP: Logged |
rnelson Member
|
posted 05-14-2009 10:52 AM
Those nifty little measurement indicators, along with the shaded scoring windows, make it quick and easy to ID this problem.They should be required on all systems, because they remove a lot of wondering and head-scratching. And that's what polygraph is supposed to be about - not having to wonder about some things. .02
------------------ "Gentlemen, you can't fight in here. This is the war room." --(Stanley Kubrick/Peter Sellers - Dr. Strangelove, 1964)
IP: Logged |
Poly761 Member
|
posted 05-19-2009 10:31 PM
In reviewing the first chart segment that was posted (R6-C7 & R8), I assigned the following scores:R6-C7 Pneumo +1, EDA -2 and Cardio "0" R8-C7 Pneumo +1, EDA -2 and Cardio "0" Your thoughts.
IP: Logged |
detector Administrator
|
posted 05-20-2009 01:00 AM
Ray,Does it matter at all how the data is recorded by any particular instrument? In other words, is it possible to have the centering be a non-issue in regards to OSS measuring the data based on how the software accounts for centering? Just curious. A second question. The red measurement markings on these. Are those automated or placed there manually? My point is...I like em. Never thought of drawing them that way. I've always counted chart lines, but this seems like a simple way to show overall amplitude and duration. ------------------ Ralph Hilliard PolygraphPlace Owner & Operator Be sure to visit our new store for all things Polygraph Related http://store.polygraphplace.com IP: Logged |
rnelson Member
|
posted 05-20-2009 07:56 AM
P7,Not saying I'm the final authority on scoring, but I'd give 0 to those pneumos. The computer will find something with which to give a score. Visually, however, they are pretty, but uninformative. No timely suppression of amplitude for 2 or 3 cycles, no slowing of rate for 2 or 3 cycles, and no timely and temporary increase in baseline for 2 or 3 cycles. The EDA is easy to give - scores to, ignoring effect of the centering mark at the CQ. The Cardio is unimpressive. The computer finds something minimal in both RQs. To me, they are just descending tracings. The CQ has a little bit of increase right at the answer. The segment scored by the computer bothers me, not just because of the centering adjustment but because it occurs about 2-3 seconds after the answering mark. I'd be inclined to be cautious and give it a zero. Ralph, the OSS-3 software is not designed to account for centering marks, only artifact marks. Keep in mind that OSS-3 is based on Kircher features - meaning that in building OSS-3 we did not do feature development. There is no reason to do so, because there is 20+ years of publication and replicated study and showing that Kircher features work well. Perhaps there are better features somewhere, but they are undocumented and/or unpublished. Plus, our current archive of confirmed case samples, coupled with the ongoing limitations of our current instrument and software availabilities, means that we cannot get to the data to study it and find out. The data are most often stored in a proprietary binary format that is useless for research. All we get is pictures of the data on the computer screen or printed paper. With all that in mind, achieving the Kircher measurements themselves, and submitting it to the algorithm is the task of each instrument manufacturer. They do this according to the information published in a couple of important journal articles and other sources. Centering could be accounted for when taking the measurements. It does not appear to be done that way in this case, as those nifty Lafayette measurement indicators clarify so nicely. Lafayette puts those on the displayed chart automatically when you run their OSS-3 tool. r
------------------ "Gentlemen, you can't fight in here. This is the war room." --(Stanley Kubrick/Peter Sellers - Dr. Strangelove, 1964)
IP: Logged |
fidibus Member
|
posted 05-25-2009 10:59 AM
At our school, AzSPS, we teach that you should wait for the entire 25 seconds, adjust wait five seconds, then begin the next question. LauraIP: Logged |
thenolieguy4u Member
|
posted 06-03-2009 11:52 AM
Hi, I noted that with many charts I look at in QC that some Examiners do not keep the variables equal at question onset with the Pneumos at the same starting point, GSR and cardio level or coming down slightly. To get the best possible use of the algorithm products we need to keep the variables the same at Q onset. IP: Logged |
Barry C Member
|
posted 06-03-2009 12:20 PM
What do you mean by equal at onset?IP: Logged |
Barry C Member
|
posted 06-03-2009 12:35 PM
Okay, I think I get it, but it isn't really correct. (It may be better, but until tested we don't know that.)Let's look at the CPU algorithm as first presented by Kircher and Raskin. They trained the algorithm on "Utah" tests. Utah tests are designed to compensate for the inherent bias (against the truthful) in the CQT, thus the reason for the N C R question order. That is, at the CQ question onset the tracings should be at their low-points. By the time of the RQ presentation, arousal should be somewhat elevated leaving less arousal potential (the Law of Initial Values) for the RQ (to try to balance the bias, so to speak). Since algorithms have been re-trained on other test structures, that probably doesn't matter much. On average, the algorithm is going to work just fine regardless. (That's what the data shows.) Can we squeeze a little more out of them? Maybe, but this suggestion could lead to the opposite effect intended. It's an empirical question for which one could find an empirical answer, but I wouldn't put the value of any findings at the top of the list. IP: Logged |
blalock Member
|
posted 06-03-2009 12:36 PM
I was trying to decipher that as well. Can you explain it more clearly? Your experience at doing QC could help all of us on here.------------------ Ben blalockben@hotmail.com IP: Logged | |