 |
 |
Dear Allstaters, Imagine a clinical test with a continuous outcome R(x) which estimates some random variable RR. The test procedure is in inself a source of variability, and the distribution of estimates (R) is highly non-normal (because of a truncated measurement range, digit preference, and all sorts of other nasty properties). Clinically, a change in RR is thought to have occured (90% of the time) when the follow-up estimate R(x+i) falls outside the range delimited by the 5th and 95th percentiles of R. How would you refer to this interval? Any sources? Other comments? Thanks ******************************************************************* Paul H Artes | e-mail: [log in to unmask] Department of Ophthalmology | office tel: (902) 473-3240 Dalhousie University | fax: (902) 473-2839 Halifax, Nova Scotia, CANADA | *******************************************************************