Hi Another thought ... if one took such a test and received a "medical report" as positive, would one have to report it as a pre-existing condition for health insurance, life insurance, or other enterprises that screen out risky characters? I'm not sure how health insurance works in the States, having access to more available Canadian healthcare. Or if you did not report it, would it be grounds for denying a later claim?
To firm up the connection to teaching, in what kinds of courses could / should instructors discuss these issues? Statistics and probability? Any course that includes material on decision making, including intro, cognitive, ...? Critical thinking courses? And are our students prepared cognitively to appreciate the point? Would the best approach be to have students actually work out the frequencies (for a number of cases) and then determine the final proportion / percentage? I've done the latter over the years for lie-detector tests (same problem of false positives), but am not convinced students took the message to "heart." Take care Jim James M. Clark Professor of Psychology 204-786-9757 204-774-4134 Fax j.cl...@uwinnipeg.ca >>> Joan Warmbold <jwarm...@oakton.edu> 14-Jun-09 5:27:19 PM >>> As per Stephen's point about the high positive test rate for both AIDES and breast cancer, I was amazed at the high positive rate for AIDES as discussed by Stanovich. "Casscells, et. al. (1978) gave a variant of the following problem to 20 medical students, 20 attending physicians and 20 house officers at four Harvard medical School teaching hospitals: /Imagine that the HIV virus that causes AIDS occurs in 1 in every 1,000 people. Imagine also that there is a test to diagnose the disease that always indicates correctly that a person who has HIV actually has it but that the test wrongly indicates that HIV is present in 5% of those tested. What is the probability that the individual actually has the HIV virus? / What answer would guess most gave, assuming that they knew nothing else about the individual's personal or medical history? The most common answer given was *95%* whereas the correct answer is approximately *2%.* Such a strong misunderstanding of the application of statistics by physicians is scary. According to Stanovich, the physicians vastly overestimated the probability that a positive result truly indicated the disease because of the tendency to overweight the case information and underweight the base rate information--i.e., only 1 in 1,000 are HIV-positive. That is, of 1,000 people, only one will actually be HIV-positive and the other 999 will be false positives, making the probability of false positives amazingly high. When we discussed this in class, it became clear that students had friends (family members, etc) who had been told they had tested positive for HIV without being told that the chances of this being a false positive a actually quite high. Instead, they were simply asked to return for another test, thinking that their chances of having the AIDS virus was fairly high as the physicians themselves apparently tend to think this is the case. And I so clearly recall my own sister when she tested positive for breast cancer and her and her husband were very worried/concerned until her physician could return from his 2 week vacation to conduct a biopsy as they were given absolutely no information about the high false positives for breast cancer via mammograms. So the idea of of using a quickie, simple test for Alzheimer's with a relatively high false positive rate would seem to be very unwise idea indeed. Come on folks, can't we see this leading to intense paranoia for folks who don't perform well on this particular test at one particular point in time? I would never recommend any type of short, quickie test for Alzheimer's' but, instead, a far more reliable series of tests, observations and interviews. Joan jwarm...@oakton.edu. Christopher D. Green wrote: > Okay, without looking at Gig's books and articles, trying to do it off > the top of my head: > > .93x13=12.09 (12 out of 13 is good) > .86=(1000-13)=848.82 (849 out of 987 means 138 false alarms for every > 12 hits). > So, the probability of actually having Alzhiemer's based on a positive > test here is only 12/138=8.69% > Is that right? > > Now, that sounds bad, like Claudia said, but for any low-probability > event like Alzheimers, you always going to have way more false alarms > than hits. It's the same for HIV and breast cancer tests as well. > > Chris > =============== > > sbl...@ubishops.ca wrote: >> For those of you who are Gerd Gigerenzer fans (and who isn't these >> days), here's an exercise for the reader involving a new screening >> test for Alzheimer's. Actually, feeling that one never knows when it >> will strike, it's just a cheap trick to get you to check my own >> calculations. >> >> There's a new BMJ report of a self-administered test for Alzheimer's. >> Takes only 5 minutes. The authors conclude "It is a powerful and >> valid screening test for the detection of Alzheimer's disease". Wow! >> >> They report sensitivity of 93% [probability of correctly detecting >> Alzheimer's] and specificity of 86% [probability of correctly >> rejecting diagnosis of Alzheimer's]. >> >> An accompanying editorial helpfully notes that the prevalence of >> dementia [which would mostly be Alzheimer's] is 13 per 1000 in people >> aged 65-69. >> >> Questions: >> >> 1) Using this "powerful" test, for every patient correctly identified >> as having Alzheimer's, how many patients will be incorrectly so >> identified with this devastating diagnosis? >> >> 2) Do you think this test is as useful as the authors claim? >> >> The first person to correctly respond will receive a free orientation >> to time and place. >> >> Sources: >> >> Article: >> >> Brown, J. et al (2009). Self administered cognitive screening test >> (TYM) for detection of Alzheimer's disease: cross sectional study. >> BMJ, 338: b 2030 [on-line first] >> >> Free at http://www.bmj.com/cgi/reprint/338/jun08_3/b2030 >> >> Editorial: >> >> Nicholl, C. (2009). Diagnosis of dementia. BMJ 338: b1176. >> >> You get a sufficient peek at it, plus, despite its warning that >> payment is required, free access to the rest when you click on "full >> text of this article" at >> >> http://www.bmj.com/cgi/content/extract/338/jun08_3/b1176?papetoc >> >> Stephen >> ----------------------------------------------------------------- >> Stephen L. Black, Ph.D. Professor of Psychology, Emeritus >> Bishop's University e-mail: sbl...@ubishops.ca >> 2600 College St. >> Sherbrooke QC J1M 1Z7 >> Canada >> >> Subscribe to discussion list (TIPS) for the teaching of >> psychology at http://flightline.highline.edu/sfrantz/tips/ >> ----------------------------------------------------------------------- >> >> --- >> To make changes to your subscription contact: >> >> Bill Southerly (bsouthe...@frostburg.edu) >> >> > > > --- > To make changes to your subscription contact: > > Bill Southerly (bsouthe...@frostburg.edu) > --- To make changes to your subscription contact: Bill Southerly (bsouthe...@frostburg.edu) --- To make changes to your subscription contact: Bill Southerly (bsouthe...@frostburg.edu)
