Yes. That not only means you're misrepresenting, but you're actively lying, since you know that you're misrepresenting.
Sigh, ironically enough for this thread, I did not lie nor did I misrepresent. My view of the evidence is different than yours.
Except that they cite the accuracy numbers estimated from the metaanalysis in the range of 0.81 to 0.91 repeatedly throughout the report.
I think this qualifies as another lie.
Now who's misrepresenting? The NAS did not give a summary number of accuracy. They did not give one because they were concerned that it would appear to be an endorsement of the polygraph when they expressly wanted to avoid that (see Fienberg, et al 2005 Statistical Science article I cited earlier). They also believed that the variability present in the studies they did use to be nonrandom and that all of the studies they used probably overestimated real accuracy (Ibid).
That's easy. Because the NAS cites compelling evidence that it is.
I suggest that you reread the NAS report again, because the evidence in there is less than compelling.
To quote Fienberg et al (2005: 12) again: "It may be harmless if television fails to discriminate between science and science fiction, but it is dangerous when government does not know the difference." They go on to say that the science does not support the contention that the CQT polygraph could have any degree of high accuracy.
If you had asked me about polygraphs before I started researching them, I probably would have parrotted CFLarsen. The positive report that the NAS gave to the polygraph is one of the single most influential pieces of evidence that changed my mind.
If that's what counts as positive evidence for you, you have a very low threshold.
I'm astonished that anyone can read a report where the NAS goes out of its way, not once, but about six times, including stressing in both the report's conclusion and its executive summary, that polygraph testing can distinguish lies from the truth with better than chance accuracy and conclude that they do not believe that polygraphs work.
You have left out the most important part of their conclusion about CQT: That it is for specific incidents only. They also said that lab-based accuracy estimates overestimate accuracy for specific incidents and that estimates from field studies also overestimate accuracy because of sampling and measurement bias...
Once again, in specific incidents, the CQT format approaches its theoretically sound cousin, the GKT...
Because those are two independent questions; one is whether the technology works at all, and the other is whether the technology works robustly enough to be applicable to specific problems. The NAS clearly separates those questions and answers them "yes" and "no" respectively (where the specific problem is screening, of course).
We have a different definition of what "works" is...
That's no different than someone asking "if hybrid technology works, why don't we see hybrid heavy goods vehicles?" Just because something works or is useful does not mean it's universallly useful.
I agree.
Better for what? I agree that GKT is almost certainly a more promising line of research and may produce better polygraphs in the future; it has a better theoretical model, and there are some claims of better performance in the lab. However, CQT has a much longer track record and a much longer and better-documented history of positive evidence behind it. CQT is much more mature technology and there are much clearer protocols defining how to use it.
It already does produce "better results."
More mature? You need to do more research. The GKT was developed by Lykken in the late 50s as an outgrowth of similar research performed in the 40s; the CQT replaced the Relevant/Irrelevant test about the same time. The reason that CQT became more prevalent is that GKT requires more work on the part of the polygrapher, not because it was better.
And just because something has been around awhile doesn't make it better.
Clearer protocols? How do I begin? You do know that in CQT polygraph, it is expected that the subject will lie to a comparison question, right? So if the polygrapher picks the wrong comparison question (e.g. the person is truthful to it), then its results are wrong. How is this more mature and better than a test designed to avoid false positives?
If you need a polygraph today to investigate an incident that happened last week, I would recommend CQT -- for one thing, you will be able to find a CQT-trained polygraph operator in almost any major metropolitan area, and the equipment is immediately available. If you are trying to build a polygraph/forensics consultation firm, I would recommend a mixture, leaning initially to CQT as that's what the clientele will be expecting you to use (and will find more persuasive in your reports), and educate your clients about the advantages of GKT over time (as more data comes in). If you are trying to build a polygraph research lab, then GKT is a promising technology and you will be in a position to develop the protocols and track record you need within your lab.
I wholeheartedly disagree...
Plus there's the whole 1988 Federal Employee Polygraph Protection Act which bars its use in employment with exceptions for law enforcement and national security agencies.
It's the difference between research-quality products and commercial-quality products. iBM has designs (in the lab) for computer chips that can run at a zillion teraHertz and use no power whatsoever -- but those designs are produced on ordinary Intel-based computers, because even IBM doesn't trust the new technology enough to rely on the research machines to do any actual work. CQT is a functional technology with some serious limitations in application; GKT at this point is barely out of vaporware. I like it. It looks nice. It looks elegant. Ask me again in ten years if it looks useful.
Sigh, did you read the NAS part about the woeful state of CQT polygraph research?
Sigh, have you done any research into the state of GKT application? It's widely used in Japanese law enforcement.
As for usefulness of the GKT, see Ben-Shakhar and Elaad. 2003. The Validity of Psychophysiological Detection of Information With the Guilty Knowledge Test: A Meta-Analytic Review.
Journal of Applied Psychology 88(1):131–151.
Since it's obvious that there's no way I'm going to change your mind, I'll give up on the dueling quotations/citations and implore you to continue to do research into the matter with the hope that maybe you'll actually change your mind...
Regards...
