- Welcome to AntiPolygraph.org Message Board.
A forum for free and open discussion of polygraph-related topics
Posted by RuinedOverCVSA
- Dec 30, 2009, 05:33 PMHey guys.. it really is a science... except when I feel like disagreeing with the computerized results.. then its intuition... except its still a science! right?
I wonder how many chemists or any legitimate sciences can just change the formula to something whenever they feel like it?
Oh the irony.
Arbitrary is the word you need to use in describing your pseudo science.
98% Accuracy.. Except when someone takes more than 1 Poly of course.. then its just random: (IE: I've "failed" 1 CVSA, and 2 Polys.. Yet I PASSED 7 Polys and 1 CVSA) Hmmm 98% doesn't turn up numbers like that no matter what way you cut it. Lets hear your excuses this time Im expecting "Oh your 10 different examiners weren't properly trained" or something.. lmao. (Only had the same guy once)
I wonder how many chemists or any legitimate sciences can just change the formula to something whenever they feel like it?
Oh the irony.
Arbitrary is the word you need to use in describing your pseudo science.
98% Accuracy.. Except when someone takes more than 1 Poly of course.. then its just random: (IE: I've "failed" 1 CVSA, and 2 Polys.. Yet I PASSED 7 Polys and 1 CVSA) Hmmm 98% doesn't turn up numbers like that no matter what way you cut it. Lets hear your excuses this time Im expecting "Oh your 10 different examiners weren't properly trained" or something.. lmao. (Only had the same guy once)
Posted by yankeedog
- Dec 10, 2009, 06:53 PMGeorge,
If I understand Gordon's term, "optimization," try to look at the "optimization" this way. When you are listening to your favorite CD and you turn up the volume, you aren't altering or manipulating the data. You are just making it easier to hear. The data remains the same. If you are at a theater and the picture on the screen is out of focus, the projectionist puts the picture in focus by adjusting the lens. Focusing the picture does not alter or manipulate the data, it just makes it easier to see. Again, the data remains the same. With the computerized software in polygraph testing, we can increase or decrease the tracings so they are easier to see. The data does not change.
One of the reasons the algorithms may have a different conclusion than the examiner is the algorithm may read an artifact as a reaction, whereas, the examiner will see the cause of the artifact (deep breathing, movement, etc) and can exclude that data.
If my views are wrong or if I have made it too simplistic, I am sure Gordon can jump in and correct me.
Hope this helps.
If I understand Gordon's term, "optimization," try to look at the "optimization" this way. When you are listening to your favorite CD and you turn up the volume, you aren't altering or manipulating the data. You are just making it easier to hear. The data remains the same. If you are at a theater and the picture on the screen is out of focus, the projectionist puts the picture in focus by adjusting the lens. Focusing the picture does not alter or manipulate the data, it just makes it easier to see. Again, the data remains the same. With the computerized software in polygraph testing, we can increase or decrease the tracings so they are easier to see. The data does not change.
One of the reasons the algorithms may have a different conclusion than the examiner is the algorithm may read an artifact as a reaction, whereas, the examiner will see the cause of the artifact (deep breathing, movement, etc) and can exclude that data.
If my views are wrong or if I have made it too simplistic, I am sure Gordon can jump in and correct me.
Hope this helps.
Posted by George W. Maschke
- Dec 10, 2009, 06:48 AMGordon,
I apologize for any implication on my part that you engaged in any unethical behavior in your review of Ed Gelb's polygraph examination of Larry Sinclair. I have no reason to believe that you did so. I cited your review as an example of the fact that polygraphers can manipulate computerized polyraph tracings after the fact. But I see the original poster had asked specifically about changing a chart to "fake" a response. I did not mean to imply that your manipulation (or "optimization," as you termed it) was tantamount to "faking a response," and I regret that my first reply in this thread may well have left that impression.
Your explanation of how the amplitude of tracings may be adjusted comports to the understanding I had gathered regarding regarding computerized polygraph results. That is, the tracing as a whole is adjusted, and not reactions to individual questions. Changing the amplitude of a tracing can, however, make a small reaction appear large. In the case of the orginal poster's question, the polygraph operator could well have increased the amplitude of a tracing in order to present the examinee with a striking display on the monitor.
The reason that I put "optimized" in quotes is because it is the term that you used in your report. I also note that on your website you state that the day after you conduct a polygraph examination, you "conduct a thorough re-analysis of the charts optimizing the parameters on a large monitor, supplemented by computer analyses." (emphasis added)
But I note that the federal polygraph handbook makes no mention of any procedures for "optimization" of parameters. Nor have I found any criteria for such optimization in the polygraph literature. How do you determine what manipulations constitute "optimizition" of parameters versus a derogation from the optimum?
You indicate that such optimization can spell the difference between an inconclusive and a conclusive outcome. But are adjustments to the tracings that result in a conclusive rather than an inconclusive outcome necessarily an "optimization?" Why or why not?
I apologize for any implication on my part that you engaged in any unethical behavior in your review of Ed Gelb's polygraph examination of Larry Sinclair. I have no reason to believe that you did so. I cited your review as an example of the fact that polygraphers can manipulate computerized polyraph tracings after the fact. But I see the original poster had asked specifically about changing a chart to "fake" a response. I did not mean to imply that your manipulation (or "optimization," as you termed it) was tantamount to "faking a response," and I regret that my first reply in this thread may well have left that impression.
Your explanation of how the amplitude of tracings may be adjusted comports to the understanding I had gathered regarding regarding computerized polygraph results. That is, the tracing as a whole is adjusted, and not reactions to individual questions. Changing the amplitude of a tracing can, however, make a small reaction appear large. In the case of the orginal poster's question, the polygraph operator could well have increased the amplitude of a tracing in order to present the examinee with a striking display on the monitor.
The reason that I put "optimized" in quotes is because it is the term that you used in your report. I also note that on your website you state that the day after you conduct a polygraph examination, you "conduct a thorough re-analysis of the charts optimizing the parameters on a large monitor, supplemented by computer analyses." (emphasis added)
But I note that the federal polygraph handbook makes no mention of any procedures for "optimization" of parameters. Nor have I found any criteria for such optimization in the polygraph literature. How do you determine what manipulations constitute "optimizition" of parameters versus a derogation from the optimum?
You indicate that such optimization can spell the difference between an inconclusive and a conclusive outcome. But are adjustments to the tracings that result in a conclusive rather than an inconclusive outcome necessarily an "optimization?" Why or why not?
Posted by Gordon H. Barland
- Dec 09, 2009, 11:41 PMDrew,
How good to hear from you. I recall you and your research well. The work you and Dr. Carlton did on systolic time intervals (STI) broke new ground and was exciting. Too bad everything back then had to be done on a spread sheet and was so manpower intensive. But then, sine labore nihil – nothing of value comes without effort.
I cannot speculate on why the algorithm didn't come to the same conclusion I did.
May you have a merry Christmas, and may the new year bring you success and happiness.
Gordon
How good to hear from you. I recall you and your research well. The work you and Dr. Carlton did on systolic time intervals (STI) broke new ground and was exciting. Too bad everything back then had to be done on a spread sheet and was so manpower intensive. But then, sine labore nihil – nothing of value comes without effort.
I cannot speculate on why the algorithm didn't come to the same conclusion I did.
May you have a merry Christmas, and may the new year bring you success and happiness.
Gordon
Posted by Drew Richardson
- Dec 09, 2009, 03:45 PMGordon,
As you may recall, I was likely the first student at DoDPI some two decades ago to collect digitized data during a polygraph exam and perhaps one of the few to ever simultaneously collect both analog and digitized and to use his own (non-polygraph industry) algorithm for detecting trends and change points in times series data, i.e., cumulative summation (cusum) analysis. In fact I believe we worked together on many such analyses during the time of my dissertation.
I have no issue with the manipulations you describe for purposes of enhancing observational clarity while at the same time maintaining the retention of the intact original digital data.
I have no real input into your questions for George or his answers to you. I do have a couple of other questions though.
What do you think is the source of the discrepancy in the results that you and George have discussed?
Are you comfortable with the very high statistical confidence limits that are routinely associated with the results of digitized polygraph exams? I personally am not aware of anything related to behavioral assessment(s) that would remotely suggest any credibility for the high levels of certainty that routinely accompany polygraph exam results.
Best Wishes, Drew
As you may recall, I was likely the first student at DoDPI some two decades ago to collect digitized data during a polygraph exam and perhaps one of the few to ever simultaneously collect both analog and digitized and to use his own (non-polygraph industry) algorithm for detecting trends and change points in times series data, i.e., cumulative summation (cusum) analysis. In fact I believe we worked together on many such analyses during the time of my dissertation.
I have no issue with the manipulations you describe for purposes of enhancing observational clarity while at the same time maintaining the retention of the intact original digital data.
I have no real input into your questions for George or his answers to you. I do have a couple of other questions though.
What do you think is the source of the discrepancy in the results that you and George have discussed?
Are you comfortable with the very high statistical confidence limits that are routinely associated with the results of digitized polygraph exams? I personally am not aware of anything related to behavioral assessment(s) that would remotely suggest any credibility for the high levels of certainty that routinely accompany polygraph exam results.
Best Wishes, Drew
Posted by Gordon H. Barland
- Dec 09, 2009, 12:27 PMGeorge, you sidestepped a direct answer to both questions, professing ignorance. That surprises me. I have obviously overestimated your knowledge of the polygraph, their operating systems, and their scoring algorithms.
The operating systems of all North American made digital polygraphs (and likely all digital polygraphs worldwide) are explicitly designed to store the raw physiological and audio/video data in a protected file that cannot be tampered with once each chart is completed. It is necessary for those data to be inviolate so the test can be reviewed for quality control or evidentiary use. An examiner cannot alter the original data post hoc. Even when the exam is copied for review, the reviewer receives a true copy of the original data. The audio/video is time-locked to the charts, so that the reviewer can easily verify that the questions had not been manipulated, such as by asking a different question than the one marked on the chart. Thus, the correct answer to my first question is no, an examiner cannot change a DI chart to an NDI chart (or vice versa) by manipulating the results after the test is over. My wording may have been imprecise; instead of "results" I should have said "raw data."
What examiners and reviewers can change is the size and position of the physiological channels (respiration, electrodermal activity, cardio, plethysmograph, and movement) for optimal viewing on the screen for analysis. In the old days with paper charts from analog polygraphs, when the test was over, the charts had to be interpreted as is. If the sensitivity of the electrodermal channel were too low so that the reactions could not be seen; or if the sensitivity were too high so that the pen hit the pen stop or went off the top of the chart, it was impossible to know how big the reactions were. Sometimes the pens snagged, which prevented both channels from recording properly. Occasionally the ink would stop flowing or flow too rapidly from the pen, and the tracings could smear and become illegible. Not infrequently two channels overlapped with each other (for example, if the electrodermal pen moved into the respiratory tracings) until the pens could be manually re-centered. In every case, when the chart was over, you were stuck with what you had.
With digital technology, these problems are a thing of the past. The examiner or reviewer can re-adjust the amplitude of each channel, if needed. Now before you say, "Aha! There's the smoking gun," you must understand that the examiner cannot enlarge the response to any one question in isolation; the enlargement (or reduction) is applied equally for all questions for that channel on that chart. For example, if the beat-by-beat amplitude of the cardio tracing was so high during the exam that it ran into the other tracings, the reviewer could reduce the amplitude after the test, but it would reduce the cardio on all questions equally. The examiner can also reposition a tracing upward or downward on the computer monitor to avoid overlapping channels. Optimizing the channels is usually done in an effort to avoid an inconclusive result, on cases going to court, or when the charts are being analyzed for countermeasures.
The computer scoring algorithm normally agrees with the examiner's scoring. When there is a difference, it usually is when the examiner considers the exam inconclusive but the algorithm recommends a decision. Only rarely does the algorithm come to the opposite conclusion than the examiner's decision. When that does happen, the examiner's evaluation is generally accepted as the "official" result; the algorithm is considered a supplement, not a replacement for the human evaluator. Keep in mind that there are perhaps five different scoring algorithms commonly available in North America, and they don't always agree among themselves. Each analyses the data differently.
There is nothing nefarious or unethical about optimizing the charts for analysis. This is a common practice within many disciplines where digital images are analyzed by trained professionals, such as dental x-rays, mammograms, and microscopy. The reviewer can zoom in or out, enhance the contrast, etc. in order to arrive at a more detailed analysis.
My optimization of the chart for analysis had absolutely no bearing on whether the algorithm agreed or disagreed with my conclusion. To suggest otherwise (and you put quotation marks around optimize as if to imply it was unethical) betrays a surprising ignorance of polygraph procedures. Your implication about my integrity calls to mind a Chinese proverb: "One's suspicion of others if often based on one's knowledge of oneself."
Peace.
Gordon
The operating systems of all North American made digital polygraphs (and likely all digital polygraphs worldwide) are explicitly designed to store the raw physiological and audio/video data in a protected file that cannot be tampered with once each chart is completed. It is necessary for those data to be inviolate so the test can be reviewed for quality control or evidentiary use. An examiner cannot alter the original data post hoc. Even when the exam is copied for review, the reviewer receives a true copy of the original data. The audio/video is time-locked to the charts, so that the reviewer can easily verify that the questions had not been manipulated, such as by asking a different question than the one marked on the chart. Thus, the correct answer to my first question is no, an examiner cannot change a DI chart to an NDI chart (or vice versa) by manipulating the results after the test is over. My wording may have been imprecise; instead of "results" I should have said "raw data."
What examiners and reviewers can change is the size and position of the physiological channels (respiration, electrodermal activity, cardio, plethysmograph, and movement) for optimal viewing on the screen for analysis. In the old days with paper charts from analog polygraphs, when the test was over, the charts had to be interpreted as is. If the sensitivity of the electrodermal channel were too low so that the reactions could not be seen; or if the sensitivity were too high so that the pen hit the pen stop or went off the top of the chart, it was impossible to know how big the reactions were. Sometimes the pens snagged, which prevented both channels from recording properly. Occasionally the ink would stop flowing or flow too rapidly from the pen, and the tracings could smear and become illegible. Not infrequently two channels overlapped with each other (for example, if the electrodermal pen moved into the respiratory tracings) until the pens could be manually re-centered. In every case, when the chart was over, you were stuck with what you had.
With digital technology, these problems are a thing of the past. The examiner or reviewer can re-adjust the amplitude of each channel, if needed. Now before you say, "Aha! There's the smoking gun," you must understand that the examiner cannot enlarge the response to any one question in isolation; the enlargement (or reduction) is applied equally for all questions for that channel on that chart. For example, if the beat-by-beat amplitude of the cardio tracing was so high during the exam that it ran into the other tracings, the reviewer could reduce the amplitude after the test, but it would reduce the cardio on all questions equally. The examiner can also reposition a tracing upward or downward on the computer monitor to avoid overlapping channels. Optimizing the channels is usually done in an effort to avoid an inconclusive result, on cases going to court, or when the charts are being analyzed for countermeasures.
The computer scoring algorithm normally agrees with the examiner's scoring. When there is a difference, it usually is when the examiner considers the exam inconclusive but the algorithm recommends a decision. Only rarely does the algorithm come to the opposite conclusion than the examiner's decision. When that does happen, the examiner's evaluation is generally accepted as the "official" result; the algorithm is considered a supplement, not a replacement for the human evaluator. Keep in mind that there are perhaps five different scoring algorithms commonly available in North America, and they don't always agree among themselves. Each analyses the data differently.
There is nothing nefarious or unethical about optimizing the charts for analysis. This is a common practice within many disciplines where digital images are analyzed by trained professionals, such as dental x-rays, mammograms, and microscopy. The reviewer can zoom in or out, enhance the contrast, etc. in order to arrive at a more detailed analysis.
My optimization of the chart for analysis had absolutely no bearing on whether the algorithm agreed or disagreed with my conclusion. To suggest otherwise (and you put quotation marks around optimize as if to imply it was unethical) betrays a surprising ignorance of polygraph procedures. Your implication about my integrity calls to mind a Chinese proverb: "One's suspicion of others if often based on one's knowledge of oneself."
Peace.
Gordon
Posted by George W. Maschke
- Dec 08, 2009, 12:09 AMQuote from: Gordon_H._Barland on Dec 07, 2009, 03:56 PMI have two questions for George. First, do you honestly believe that an examiner can change an NDI chart to DI (or vice versa) by manipulating the results after the test is over?
As I've explained in the discussion thread about Ed Gelb's polygraph examination of Larry Sinclair, your review suggests that such post-test manipulation is possible.
QuoteSecondly, the clear implication of your comment, George, is that I engaged in unethical behavior. Although we've never met, you know my reputation. Do you honestly believe that I did or would do that?
Your review of Gelb's polygraph examination of Larry Sinclair clearly indicates that you did manipulate a chart before scoring it, and the resultant outcome was diametrically opposed to that returned by the PolyScore algorithm. I do not pretend to know what your motivations for manipulating the chart were, but I don't think you've provided an adequate explanation of your reasons for doing so, nor do I believe that you have you adequately explained how you came to score a chart as "deception indicated" that the PolyScore algorithm scored as "no deception indicated" with a putative confidence level of >99%.
Posted by BBernie
- Dec 07, 2009, 09:46 PMQuote from: Gordon_H._Barland on Dec 07, 2009, 03:56 PMAnonymous wrote: "My question is can the examiner change the chart on the computer to "fake" a response...." to which George replied, "Yes. Polygraph operators can manipulate charts after the fact to produce a desired result. For a recent example, see Gordon Barland's review of Ed Gelb's polygraph examination of Larry Sinclair. Barland "optimized" the electrodermal channel to produce an outcome diametrically opposed to that reached by the PolyScore computerized scoring algorithm."
I have two questions for George. First, do you honestly believe that an examiner can change an NDI chart to DI (or vice versa) by manipulating the results after the test is over?
Secondly, the clear implication of your comment, George, is that I engaged in unethical behavior. Although we've never met, you know my reputation. Do you honestly believe that I did or would do that?
Gordon
So basically you're saying "Trust me!". Now I have a question for you. Why, pray tell, should we trust you??
Posted by Gordon H. Barland
- Dec 07, 2009, 03:56 PMAnonymous wrote: "My question is can the examiner change the chart on the computer to "fake" a response...." to which George replied, "Yes. Polygraph operators can manipulate charts after the fact to produce a desired result. For a recent example, see Gordon Barland's review of Ed Gelb's polygraph examination of Larry Sinclair. Barland "optimized" the electrodermal channel to produce an outcome diametrically opposed to that reached by the PolyScore computerized scoring algorithm."
I have two questions for George. First, do you honestly believe that an examiner can change an NDI chart to DI (or vice versa) by manipulating the results after the test is over?
Secondly, the clear implication of your comment, George, is that I engaged in unethical behavior. Although we've never met, you know my reputation. Do you honestly believe that I did or would do that?
Gordon
I have two questions for George. First, do you honestly believe that an examiner can change an NDI chart to DI (or vice versa) by manipulating the results after the test is over?
Secondly, the clear implication of your comment, George, is that I engaged in unethical behavior. Although we've never met, you know my reputation. Do you honestly believe that I did or would do that?
Gordon
Posted by BBernie
- Dec 06, 2009, 01:33 AMQuote from: PhilGainey on Dec 06, 2009, 01:01 AMQuoteit was such a massive, ridiculous, overexaggerated hump/response that there is no way that it can be accurate. Eespecially as I really had no heightened response to that particular question - I was not lying, nor worried about anything, nor doing anything that I wasn't doing for any of the other questions etc.
She could manipulate the chart to look like snoopy's nose, but it still doesn't mean you are lying or being deceptive. And if they ask you why you are reacting, they are asking you to speculate on a physiological response you have no direct control over. You might as well say, "I don't know, ask my central nervous system" or ask my "lymbic system".
But just for that reason you will be considered as "deceptive" because of the reaction (or that's what I assume the polygrapher would interpret).
Posted by T.M. Cullen
- Dec 06, 2009, 01:01 AMQuoteit was such a massive, ridiculous, overexaggerated hump/response that there is no way that it can be accurate. Eespecially as I really had no heightened response to that particular question - I was not lying, nor worried about anything, nor doing anything that I wasn't doing for any of the other questions etc.
Understand that "response" does not equal "lying". Did she say it does?
She could manipulate the chart to look like snoopy's nose, but it still doesn't mean you are lying or being deceptive. And if they ask you why you are reacting, they are asking you to speculate on a physiological response you have no direct control over. You might as well say, "I don't know, ask my central nervous system" or ask my "lymbic system".
Posted by George W. Maschke
- Dec 05, 2009, 09:44 AMQuote from: anonymouse on Dec 04, 2009, 05:38 PMMy question is can the examiner change the chart on the computer to "fake" a response, then use that "fake" response to pretend that the machine caught me lying, and then hope that I "confess" to something?
Yes. Polygraph operators can manipulate charts after the fact to produce a desired result. For a recent example, see Gordon Barland's review of Ed Gelb's polygraph examination of Larry Sinclair. Barland "optimized" the electrodermal channel to produce an outcome diametrically opposed to that reached by the PolyScore computerized scoring algorithm.
Posted by anonymouse
- Dec 04, 2009, 05:38 PMI took a polygraph that was done using a computer (rather than a traditional polygraph machine). After the exam finished, the examiner spent at least five minutes clicking things with his mouse, and then turns the screen to me and points to a large "hump" on one particular question, then claims that I'm showing a huge response. He then accused me of lying in my response to that particular question.
My question is can the examiner change the chart on the computer to "fake" a response, then use that "fake" response to pretend that the machine caught me lying, and then hope that I "confess" to something?
It was such a massive, ridiculous, overexaggerated hump/response that there is no way that it can be accurate. Eespecially as I really had no heightened response to that particular question - I was not lying, nor worried about anything, nor doing anything that I wasn't doing for any of the other questions etc.
It all seems a little too convenient for my liking - a sizeable response on one question on an otherwise "normal" chart, followed by questioning about whether I had something I was hiding.
My question is can the examiner change the chart on the computer to "fake" a response, then use that "fake" response to pretend that the machine caught me lying, and then hope that I "confess" to something?
It was such a massive, ridiculous, overexaggerated hump/response that there is no way that it can be accurate. Eespecially as I really had no heightened response to that particular question - I was not lying, nor worried about anything, nor doing anything that I wasn't doing for any of the other questions etc.
It all seems a little too convenient for my liking - a sizeable response on one question on an otherwise "normal" chart, followed by questioning about whether I had something I was hiding.