posted 05-22-2013 11:09 AM         

McClatchy - The Polygraph Files - http://www.mcclatchydc.com/polygraph/

posted 05-22-2013 04:01 PM            

[edit]

My first response was not intended to sound negative -- I'm sick and grumpy from long days and days of travelling and sleeping in airports and airplanes

That is helpful and interesting.

We have already been studying this, and will continue to provide as much information and technical detail as people want.

However, I'm not sure that a detrend filter that removes slow-wave and quadratic wave form isn't just a high-pass filter. A sine wave is a quadradic wave form. Butterworth filters are common solutions for this. Tonic EDA instability has been described by some of our engineers as very low frequency sine type wave. And many waveforms can be defined with a quadradic equation. So a filter for quadratic waveforms could be a fancy and confusing, but impressive, way of referring to a high-pass filter for the EDA waveform.

Of course, I'm no expert. I'm just sayin'.

r

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

posted 05-22-2013 06:29 PM            

I didn't quote Raskin and Kircher because I was running from memory, but they call them "quadratic." I'm okay with that term since we all remember our high school teachers forcing us to plot our answers for ax^2+bx+c and getting that curved plot (like a sine wave). I don't dare to go much beyond that description since I don't know why they differentiate it from the auto (high-pass) filter they also have that stabilizes the baseline.

To get this right, the detrend (filter) "...remove[s] global linear and quadric trends from the tracing."

They call the detrending feature "detrend" and the filtering feature a filter. They are both filters. It's important to note what they have found:

quote:

---

The measurements obtained from a detrended signal may differ slightly from those obtained from a signal that has not been detrended.... However, our research indicates that the diagnostic validity of the measurements is not affected by detrending."

---

posted 05-23-2013 06:00 AM            

"...global linear and quadratic trends..." still sounds like a high pass filter. But maybe also a slope filter. Someone should. ask.

I could say the same thing about the Lafayette with complete accuracy and honesty. The auto EDA removes global quadratic trends and linear trends. The Lafayette Detrended EDA does not remove linear trends but removes slope trends and only downward slope trends and at or only below the baseline arrow. So the correspondence between measured reaction features using Lafayette detrended and manual EDA is essentially perfect because it is the same. Not at all sure that would be the case if the "global linear and quadradic trends" filter is a form of high-pass FIR filter.

Anyway, Marisa Taylor either missed that point or intentionally lied when she implied we have been unsuccessful at resolving scoring differences between raw and managed EDA modes.

All of this is an example of why we need access and opportunity to understand the technical details. If not we are vulnerable.

All of this is why we did OSS 3 as an open source algorithm. There is always more to learn. And we can learn if we have access and opportunity. As with other forms of social conflict, a choice to not learn more - acting as if we don't have to learn more, or as if we already have ask the answers we need - may become a threat to our survival.

< sorry more typos than usual - posting from a phone is dangerous >

.02

R

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

posted 05-23-2013 10:33 AM            

You are receiving this newsletter because you are a Lafayette Instrument
Company customer. You may view this email
online.

RESPONSE TO RECENT ARTICLES
PUBLISHED ONLINE MAY 22, 2013

Dear Colleagues,

More than two months ago, Lafayette Instrument Company received questions
from a journalist regarding our EDA (electrodermal activity sensor), our
LX4000 polygraph system, and our business practices. We were not obligated
to respond to the questions, but we did respond hoping that we could
improve the quality and integrity of the subsequent article. In that
regard, we were unsuccessful. On or around May 20, 2013, a series of
articles were published online and we feel these articles selectively
ignored extensive amounts of data that we provided to the lead journalist.

The complete set of questions, and our responses, can be found here: http://www.lafayettepolygraph.com/eda-questions.pdf

Additionally, we provided a shorter EDA document to our customers and
partners in March (2013): http://www.lafayettepolygraph.com/eda-info.asp

To summarize:

There is nothing wrong with the LX4000. Implications in the news media
regarding LX4000 problems are based in misinformation. We are continuously
engaged in research and development to improve all of our products and we
are certain that the LX4000 and LX5000 devices represent the best
technology available to the profession. The EDA phenomena referenced in the
article involves the potential for occasional tracing variations between
Manual and Auto EDA; this is a known phenomenon that is documented in
psychophysiology literature, and is not unique to devices from Lafayette
Instrument. It was irresponsible and misleading to alarm people by labeling
a known phenomenon, a “glitch” It was similarly confusing to misuse the
term “skew” in an unscientific way. All EDA modes have their advantages and
disadvantages. LIC has taken the approach of helping customers evaluate
these advantages and select the EDA mode that best meets their needs.

We stand by our products and we will work with you to quickly resolve
issues or address further questions as needed.

For more information, or to get to know us better, please contact

*Terry Echard*
President and General Manager
terry@lafayetteinstrument.com

*Chris Fausett*
Vice President and
Polygraph Product Manager
chris@lafayetteinstrument.com

*Jennifer Rider*
Vice President and CIO
jennifer@lafayetteinstrument.com