That's a very well thought out question and observation.  I agree.  We can't determine too much from only the self-reported CM users.  (We can determine some things, depending on how much data we have, but it still requires some speculation.)

The research shows many examiners aren't good at detecting CMs, but thankfully the research shows, as this topic points out on post number one, it doesn't matter since they don't work.  We don't need to speculate.  We have data.

You'll find that the only CMs that were effective (in the Honts study touted here so often) were those for which the subjects had hands-on training - not likely in the field, unless one hires the gentleman from the other site, but his suggestions are laughable.  Then, it's still unknown if one could augment CQ reactions that look real AND are greater than the RQs AND be timely AND fail to produce any of the CM signatures. Good luck.

With that said, I've seen some examiners fooled.  They shouldn't have been, but they were.  Not all examiners are created equal, and some would score what are literally impossible involuntary physiological "reactions."

I probably know more about the polygraph research than most, and I wouldn't want to try to "beat the box" if getting caught had a price.  I don't believe I could do it (which is exactly what the research shows).