Dan:

I think you would have made a fine president of the APA.

The problem is the polygraph establishment and their inflexible mindset.

Regarding the polygraph accuracy percentages, I've seen numbers from the low eighties to the high nineties.  This lack of consistency is deplorable from the scientific viewpoint.

In fact, there are two accuracy rates: the percentage of innocent subjects that correctly pass the polygraph, and the percentage of guilty subjects that correctly fail the polygraph.  As the threshold is changed (composite numerical score, something between about plus/minus 20), one accuracy percentage increases but the other percentage necessarily decreases, and vice versa.  In other words, a tradeoff must be established.  Do you set the threshold to minimize false negatives at the expense of increasing false positives, or the other alternative?  I see no reason that the threshold be set to equalize the two accuracy rates.  And how do you deal with the issue of a very small base rate (less than 1 in about 1000 or so)?

Regards, Evan S