LieBabyCryBaby wrote on Dec 7
th, 2006 at 4:02am:
Digithead,
You can be condescending and spout all the complicated math that you want, but it makes no difference.
Wow, I must've hit a nerve.
I guess you suffer from math phobia but I was unaware that addition, subtraction, multiplication and division were considered "complicated math."
LieBabyCryBaby wrote on Dec 7
th, 2006 at 4:02am:
It's all very basic, not complicated. If we test 100 people, and we are correct 90% of the time, we can be wrong 10% of the time and it doesn't matter. From the viewpoint of the powers that be in most agencies, a screening process that gets it right 90% of the time, or even 80% of the time, does what it is supposed to do--it screens.
Except you are making the base rate rate fallacy. In the real world, we never know the truth beforehand, so the real measure of polygraph employment screening accuracy is whether the prospective employee is deceptive given that the test is positive. With your mythical world of 90% accuracy, I have
clearly shown you that when the base rate is low the majority of positives will be false. This will result in too many people wrongly accused of lying with no recourse to clear their names. How can you defend this as a cost of doing business? How can you continue to claim that something is accurate when it's
right only 1 out of 12 times for the very thing it was designed to detect? And given that we're dealing in hypotheticals, I contend that the base rate of deception is even lower and that polygraph accuracy is nowhere near 90%, causing the false positive rate to be even more dramatic than this example.
And when this is spun into the context of sex offender screening, you have no defense against the 10% false negative rate. They're simply a danger to the community rather than "a cost of doing business" or "an acceptable loss."
LieBabyCryBaby wrote on Dec 7
th, 2006 at 4:02am:
So, let's assume that out of that 100 people we get 90 right and 10 wrong. Of course any agency is going to hope that all of those 10 wrong are false positives, not false negatives. Why? Because that is simply 10 people who don't get the job. But if 10 false negatives get the job, then an agency has 10 out of 100 people on the job who got through the process despite having various relevant issues to hide--the issues the agency cares about most. Screening out 10 false positives is simply the cost of doing business, you see. What they don't want are criminals working for them. Where those 10 false positives came from, there are plenty of other qualified candidates to choose from. If you want to raise those 10 false positives to 20, then it's not as attractive, but still an agency will have 80% of its employees who are the type of employees the agency wants.
Sorry, you're still making the base rate fallacy. But continue your handwaving it eases your cognitive dissonance...
LieBabyCryBaby wrote on Dec 7
th, 2006 at 4:02am:
Now, of course we want to look for the most accurate screening methods we can find. But looking and finding are two different things. While we are looking, we use what we have. Before DNA, we used what we had, despite whatever inaccuracies, simply because we needed to use something, right? And I still don't think law enforcement agencies are ready to give up on fingerprinting or ballistics or eyewitness accounts, for example, simply because they aren't 100% accurate. Neither are they willing to give up on the polygraph when it is one of those useful tools and the best thing currently available.
You're using misdirection again. Fingerprinting, ballistics and eyewitness accounts are examples of pattern recognition. Similarities in patterns, false memories, or fuzzy data account for their mistakes. These can also be verified with other evidence. CQT Polygraph is testing for the presence of deception based on a comparison to a
probable lie. Not similarities in swirls, ridges, or descriptions. It requires a different standard in which to evaluate its accuracy.
LieBabyCryBaby wrote on Dec 7
th, 2006 at 4:02am:
Finally, one million people doing the wrong thing doesn't make it right. But when those one million people are doing a very tough job, they use the best tools they have, and they'll keep using them until someone invents something better. Got any inventions in mind, Digithead?
I concur that law enforcement is a very tough job. Which is why I moved from the field into academia because I saw all of the junk being used and felt that I could do a better job at getting rid of it via good research and evaluation.
And somebody (Lykken, RIP) did invent something better than CQT polygraph. It's called the guilty knowledge test and it rests on the scientific bedrock of cognitive response rather than CQT's emotional response.
Trust me, LBCB, I'm doing all I can to improve law enforcement and crime control policies. It may take me my whole career, but I hope to be able to dump CQT polygraph onto the trash heap of phrenology and other pseudosciences once used in law enforcement and criminology.
ETA: Change "wrong" to "right only". My mistake...