George,

The OTA's findings and statements that polygraph is better than chance was based on all available credible research and only acceptable field studies were included.  Some suggested research studies were eliminated due to validity and structural problems found upon "peer-review".  

The Bersh study, although archaic, does much to enlighten the general public and scientific community by producing what I believe to be the first field research study of the ZCT (1961 Backster).  There were changes made to the ZCT that could have increased the Bersh study accuracy even further, standardized numeric scoring criteria being one.  

Bersh examiners' did not solely conduct a "clinical" evaluation of subjects for deception as Lykken suggests.  The examiners used a global scoring method of evaluation.  This method does uses the charts to discern if one is showing deception to a particular question.  Global scoring also includes observations of the subject prior, during, and following the exam, and all the available investigative material.  It does puzzles me that Lykken would state, "clinical impressions or behavior symptoms, which, we know from the evidence mentioned above, should not have permitted an accuracy much better than chance."  It is a well-known fact that psychologists quite frequently use this very method to come to their professional opinion.  Sometimes, if not often, psychologists' opinions are regarding weather a client "truly" believes or is being "truthful" about something they say has happened to them.  The psychologist then gives his professional opinion to the aforementioned.  I have both seen and heard psycholgists testify to these opinions in court.  Unlike polygraph, psychology rarely has physiological data to base or support their inferences.

As for the results of the study, the OTA compares Bersh?s to Barland and Raskin?s study.  The OTA does note that the two studies have some inherent differences.  However, the OTA considered the studies similar enough to compare.  The OTA states, "Assuming the panel's decisions, the two studies' results are strikingly different.  Barland and Raskin attained accuracy rates of 91.5 percent for the guilty and 29.4 percent for the innocent subjects; comparable figures in Bersh's study are 70.6 percent guilty correct and 80 percent innocent correct."  My math shows a combine accuracy rate of 81.05 percent for guilty, 54.7 percent for innocent, and 67.87 percent overall accuracy for the two studies.  The OTA then wrote, "It is not clear why there should be this variation?.."  They go on to give some possible but miss some technical reasons for the differences in the findings of the two studies.  Most obvious, Bersh?s study used ZCT and R&I, the global scoring method, and eliminated inconclusive exams.  Ground truth is the most difficult element to establish in a polygraph research study because it is subjective to the interpretations and opinions of the peer-reviewer.

The R&I question format has proven to be a less accurate technique when compared to the ZCT or CQT in specific criminal issue examinations studies.  This is arguably the reason why the Army Modified the General Question Technique (MGQT) to include comparison questions, zone/spot scoring, and total chart minutes.   How the two question formats in Bersh?s study compared or differed in accuracy would be interesting.

The available scientific research for polygraph shows that a greater percentage of inconclusive exams are found in the innocent.  Thus it would be prudent to ascertain that Barland and Raskin's study may have produced similar, if not better, results in the truthful and in the deceptive when compared to Bersh, if inconclusive results were set aside.  The scientific community often holds inconclusive results against polygraph when reviewing its scientific validity and accuracy.  However, polygraph examiners view inconclusive results as not enough in the chart tracings to determine an opinion.  An inconclusive can be attributed to many variables.  One example of inconclusive chart tracings may be found in an exam where the examinee has problems remaining still or intentionally moves.  Even Farwell the Brain Fingerprint?s inventor says his instrument will produce inconclusive results if the examinee does not remain still during the examination.  

Lykken also states, "Because the exams were clinically evaluated, we can be sure that every test that led to a confession was scored as deceptive."  He makes this statement without any supportive documentation, and/or reference to a specific incident within the study where this actually occurred.  There is no evidence to support his opinion on this issue.  If a confession were to be obtained prior to chart data collection, the exam would have been considered incomplete by the examiner.  This is not the case in point in Bersh?s study because inconclusive and incomplete exams were not included.

Lykken's argues that Bersh's study is "fatally flawed" because of his prior assertions.  He writes, "That the polygraph test frequently produces a confession is its most valuable characteristic to the criminal investigator, but the occurrence of a confession tells us nothing about the accuracy of the test itself."  I agree that a confession is a valuable tool in a criminal investigation.  I disagree with his fallible knowledge of the use of a properly documented confession to ascertain conformation of the polygraph data results.  A proper confession covers the elements of the crime and includes information that only a person who committed the crime would know.  When this information is present in a confession, it would undoubtedly confirm the data.  The question here is not weather the confession can be used to confirm the polygraph chart data but what standard was used in deeming statements made by examinee's as a confession.  However, this point is not asserted and/or proven in Lykken's argument, thus it would appear to be a nonexistent flaw.

Horvath's research study provides good data in areas but had missing information that might of hindered the overall accuracy results. Barland submits that Horvath's original examiners were 100 percent correct in their opinions.  Barland notes that some special charts administered in 32 percent of the cases were removed from the files of considered deceptive subjects.  These special charts were most likely removed to avoid pre-judgment by the research evaluators. I do not think his study invalidated polygraph in anyway.  The study in fact provided valuable insight into the possible effect incomplete chart data might have on accurate review.  Horvath's study still produced better then chance results considering there was a 50% chance of the reviewers being correct and they were overall 64% correct.

Lykken states, "The original examiners in these cases, all of whom used the Reid clinical lie test technique, did not rely only on the polygraph results in reaching their diagnoses but also employed the case facts and their clinical appraisal of the subject's behavior during testing."  This statement is partially true but not completely factual.  The examiners in this study used scoring of the charts along with the global information present.  The global scoring method in no way goes against the chart data results and contrariwise uses other information to confirm the chart data.  Lykken goes on to purport, "Moreover, some other suspects, judged truthful using global criteria, could have produced charts indicative of deception."  This is an illogical statement.  He never stipulates what scoring method or criterion might have produced a deceptive result.  He cannot prove or disprove his assertions.  I could easily conclude that, if given all the data available to the original examiners and the same scoring method, the reviewers would have concurred with the original examiners in 100% of the cases.  Neither Lykken nor I can prove or disprove our assertions because this variable was not present or measured.  However, Barland had access to and reviewed the data after the missing variable was discovered.

The point I am making is that no matter how meticulous one accounts for variables there will most likely be ones that need further research to answer.  This is true in any research including physiology, psychology, and medical.  One cannot control for and/or predict every possible variable.  Because a variable is in question does not invalidate the findings or methodology.  Because DNA's sample database is relatively small in comparison with the total population that inhabits the earth, does not lead scientists to doubt the accuracy and/or scientific validity of its methods or findings.

I have read chapter one of The Lie Behind The Lie Detector and have found no reference to what standardization and control the CQT lacks.  You state that it lacks these elements but give no examples or criterion for standardization and control.  I would think this would be hard for you to do, as even the scientific community is quite subjective in their opinion of what constitutes acceptable standardization and control for scientific validity.  Can you reference, for comparison purposes, any other scientific method that has been accepted and its basis for acceptance?  Can you reference, for comparison purposes,  any other scientific method that was rejected based on comparable factors you might use to make this statement?