I am responding here to “nopoly4me” as he has requested.  I doubt seriously that my response will have any effect on his or any other person’s opinion of polygraph or its efficacy.  I should clarify a few things about the study at question here for those who have read only the abstract.  First, the polygraph test charts were scored using the CPS (Computerized Polygraph System) scoring software, not the polygraph examiners who conducted the tests.  Therefore, being “blind” to the condition of the participants in the study would not make a difference to the computer’s decisions.  The CPS scoring software produces a posteriori or “inductive” probability of truthfulness or CPSp|t.  Those values alone were used as the primary dependent measure for analysis.  A proposition is knowable “a priori” if it is knowable independently of experience. A proposition is knowable “a posteriori” if it is knowable on the basis of experience.  In the case of the CPS scoring software, decision probabilities are derived based upon experience gained from previous polygraph test where ground truth is known.  The software is then tested against a second batch of known condition or ground truth tests to determine it's "a posteriori" ability to provide a probability accuracy.  Others here have made the argument that even with the use of the computer as the scoring mechanism, the examiners could have “swayed” test results because they knew the condition of the participant.  Such an argument has little value.

     Second, the argument is being postulated here by some, “nopoly4me” included, that this study has provided some proof concerning the accuracy or lack thereof of polygraph.  Such is not the case.  The comparisons in this study were between the participants who were given information from TLBTLD and those who were not, under both guilty and innocent conditions.  This study subjected CPSp|t to a guilt (guilty, innocent) X information (informed, naïve) ANOVA.  ANOVA allows us to compare different things and form conclusions based upon the within class differences in those things.  It provides us the significance between those differences. This study revealed that CPSp|t values for guilty participants were significantly lower than were the CPSp|t values for innocent participants.   Stated simply, when TLBTLD information was provided to innocent participants, their correct condition classification accuracy was diminished.  We can infer that that difference in classification accuracy was diminished by the introduction of TLBTLD which was the variable. Kendall’s Tau-b analysis of the data indicated a significant relationship between the guilt criterion and the decisions for both informed, tau-b =.50, p=.009 and naïve participants, tau-b=.41, p=.046.  Tau-b values range from -1 (100% negative association, or perfect inversion) to +1 (100% positive association, or perfect agreement). A value of zero indicates the absence of association.  By any measure conducted there was significance between those who had information from TLBTLD and those who did not and the significance was that those innocent participants with TLBTLD were more likely to be misclassified as guilty by the CPS scoring software.

     Finally, Sergeant 1107 (based upon his understanding of a previous posting) is incorrect in his assertion that this study provides us with a 72% correctness of classification or a 72% accuracy rate for polygraph.  Polygraph accuracy was not determined in this study, nor was it a goal of the investigators.  To make such an assumption misreads the data and makes conclusions not supported by this study.  I would suggest that anyone who has an interest in the study to obtain a copy from the publisher and read it so that you may form your own conclusion about the value of reading TLBTLD prior to your polygraph.. Based upon this study TLBLD will decrease your chances of being classified as innocent and increase you chances of being misclassified if you are actualy innocent.  If you are guilty, read on as it made no difference in classification accuracy.