RSS Ordinary Meeting (organized by the Research Section) "Sampling bias and logistic models", by Peter McCullagh (University of Chicago) Wednesday 6 February 2008, 5pm RSS Headquarters, 12 Errol Street, London Summary This paper considers various forms of random-effects models for binary data. It concludes that parameter attenuation is a statistical illusion attributable to sampling bias and ambiguous notation. A new random-effects model makes it clear that the conditional distribution of Yi given Xi=x is not the same as the marginal distribution of Y-values in stratum x. Implications for likelihood calculations and estimating equations are described. The paper can be downloaded and more information can be found from: http://www.maths.lancs.ac.uk/~wite/RSS/ DISCUSSION CONTRIBUTIONS ON THE DAY ARE INVITED. PLEASE CONTACT THE SECRETARY OF THE RESEARCH SECTION ([log in to unmask]). Ernst Wit Secretary Research Section ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Professor Ernst Wit Chair of Biometrics, Head of Medical Statistics Department of Mathematics and Statistics Lancaster University Lancaster LA1 4YF, UK Tel. 00 44 (0)1524 593 954 Fax. 00 44 (0)1524 592 681