CURRENT RESEARCH TOPICS

‘Buehlerised’ conditional upper limits for difference between two correlated proportions. Imagine asking N individuals whether they are satisfied or dissatisfied on two occasions. These matched binary pairs lead to a 2x2 table, but the proportions in each row are correlated. We compare 5 standard approximate methods for constructing an upper limit for the difference in the probabilities. Buehler (1957) showed how to construct upper limits that are unconditionally valid and also optimal (i.e. minimal) subject to a given ordering. The aim of this project is to use Buehler’s method to adjust the approximate limits so that they are unconditionally minimal. There is a pretty clear winner. (This is joint work with my Doctoral Student, Max Moldovan and a submitted version is HERE.) Routines for calculating the SRLR based confidence limits in R and also Matlab are in a zipped file HERE. The paper is to appear in Statistics in Medicine in 2007.