I don't believe polygraphs can distinguish between lies and truth.

However, out of curiousity, are they even objective enough so that the charts produced will be read the same by all polygraphers?

If a test is run, and the chart is given to 10 polygraphers, will all 10 (or at least the vast majority of them) come to the same conclusion as to "truthful" or "deceptive"?

Been wondering about this for a while.  I'm rather assuming most blind polygraphers WOULD come to the same conclusion on any given chart.  But not sure about it.

lane99, 

I have been involved in studies of examiners scoring charts, and we found examiners do have high agreement rates when scoring charts.  Properly trained examiners are in agreement on results of polygraph charts. 

Bill_Brown wrote on Aug 12^th, 2011 at 4:46pm:


Publication citation, please.

That CBS Expose was in 1986, over 20 years ago.  It was also a setup.

Bill_Brown wrote on Aug 13^th, 2011 at 9:49pm:


The Marin protocol is irelevent to interrater reliability. So are all the studies you cite. 

Teh Horowitz study you cite even concludes "The R-I test produced an unacceptable rate of false positive decisions." 

You dont offer a cit by Donald Krapohl. Maybe you can dig up something responsive? I can't, but your the expert. 

figs, 

Quote:


The Marin protocol requires the examiner score 100 known solution cases with a  minimum of 86% accuracy.  That is inter rater reliability.  This is to qualify under that system as an expert witness in court using paired testing.  And other studies I have seen and not quoted are in the same area of inter rater reliability. 

I will look later and find other studies, sorry I just don't have the time right now.  

figs, 

Found this on the APA website, another study to look at: 

[quote] A Replication and Validation Study on an Empirically Based Manual Scoring System1
Ben Blalock, Barry Cushman & Raymond Nelson
Abstract
This is a replication of a study validating the hand scoring system for comparison question polygraph examinations proposed by Nelson, Krapohl and Handler (2008). Nine polygraph examiner trainees at an American Polygraph Association accredited polygraph school used an empirically based three-position manual scoring system involving three evaluative criteria and a reduced set of basic rules to evaluate 100 confirmed event-specific single-issue criminal investigation polygraph examinations from the Department of Defense Polygraph Institute confirmed case archive. Average decision accuracy for the inexperienced examiners was 88% with 13.1% inconclusives. Sensitivity and specificity levels achieved by the trainees did not differ significantly, suggesting they achieved balanced accuracy characteristics using the empirically based scoring system. All nine of the inexperienced examiners scored the sample cases with sufficient accuracy to meet the accuracy requirements specified by the Marin protocol (Krapohl, 2005; Marin, 2000).      Results from this study parallel the results reported in the previous experiment and support the validity of an empirically based three-position manual scoring method. [quote]

We were discussing the topic of inter rater reliability.  I used the protocol known as the Marin protocol to demonstrate there is inter rater reliability demonstrated by this protocol.    


Certifies:
The Center certifies qualified examiners, in accordance with 
ASTM Standard E2324-04 (The Marin Protocol)
Examiners applying for certification must submit an end-to-
end videotape of an examination showing their proficient 
adherence to a validated methodology, and the interpretable 
chart produced by that examination. They must then blind-
score 100 sets of charts generated in real-world examinations 
in cases where the truth is now known, fifty of which are 
from examinations where the subject was deceptive and the 
rest from examinations where the subject was non-deceptive. 
The test sets are selected from a collection of more than 400 
sets. The applicant is certified if and only if he declares a 
conclusive result for at least 80 per cent of each group of 
fifty, and at least 86 of his conclusive results for each 
group are correct.