tag:blogger.com,1999:blog-3240271627873788873.post4530912815844744064..comments2017-03-24T09:35:22.124-04:00Comments on Doing Bayesian Data Analysis: Lessons from Bayesian disease diagnosis: Don't over-interpret the Bayes factor, VERSION 2John K. Kruschkehttp://www.blogger.com/profile/17323153789716653784noreply@blogger.comBlogger3125tag:blogger.com,1999:blog-3240271627873788873.post-90323131455576647612016-03-12T16:01:10.425-05:002016-03-12T16:01:10.425-05:00For another treatment of uncertainty in the prior ...For another treatment of uncertainty in the prior probability along with uncertainty in the Bayes Factor (that is, the ratio of the test's sensitivity and specificity) see<br /><br />Mossman D, Berger JO: Intervals for posttest probabilities: a comparison of 5 methods. Med Decis Making 2001;21:498-507<br /><br />available here:<br />http://mdm.sagepub.com/content/21/6/498.full.pdfDouglas Mossmanhttps://www.med.uc.edu/psychiatry/department-directory/faculty/Index/Pubs/mossmad/noreply@blogger.comtag:blogger.com,1999:blog-3240271627873788873.post-7093174933992661522016-01-04T12:50:23.058-05:002016-01-04T12:50:23.058-05:00Right, the disease-diagnosis example is so over-us...Right, the disease-diagnosis example is so over-used (without further examples involving continuous parameters) that a lot of beginners think that Bayesian analysis only applies to discrete hypotheses.<br /><br />And yes, an even more realistic expansion of the model would not only include uncertainty in the base rates, as was done in this post, but would also include uncertainty in the detection rate and false-alarm rate.<br />John K. Kruschkehttp://www.blogger.com/profile/17323153789716653784noreply@blogger.comtag:blogger.com,1999:blog-3240271627873788873.post-36546123152257283292016-01-04T12:12:30.926-05:002016-01-04T12:12:30.926-05:00Thanks for a detailed post! :) For the longest tim...Thanks for a detailed post! :) For the longest time I actually thought Bayes factors referred to posterior odds, as that was how people used them... <br /><br />Btw, while it's a really good example in this post, I really don't like widespread use of the disease-diagnosis situation to introduce Bayes. In my mind, Bayes is a lot about modeling uncertainty, but the disease-diagnosis example starts out by giving these, for some reason, <b>extremely certain</b> detection/false-alarm rates, which in any "real" situation would be pretty uncertain too. So it's a good example <b>if</b> you want to confuse people, imho :) Rasmus Bååthhttp://www.blogger.com/profile/16575386339856902265noreply@blogger.com