STATISTICS SEMINAR Queen Mary and Westfield College THURSDAY 20 JANUARY, 2000 ___________________________________________________________________ Peter McCullach UNIVERSITY OF CHICAGO "WHAT IS A STATISTICAL MODEL?" ____________________________________________________________________ ALL ARE WELCOME The talk will START at 16.30, in the Mathematics Seminar Room (103) on Level 1, Mathematics Building, Queen Mary and Westfield College. Mile End Road, London The nearest underground station is Stepney Green. Turn left at the exit and walk 400 yards. _____________________________________________________________________ ABSTRACT The argument is made that a statistical model is more than a family of distributions on a sample space. Without a suitable extension to other units, to other treatments, to other sample spaces, and so on, no inference is possible. All useful statistical models are extendable, and this extension is invariably taken for granted in model formulation. I argue that it is helpful to make this extension explicit. Motivated by absurd examples, a case will be made in favour of the following propositions. (i) A statistical model is necessarily a functor on the category of statistical designs. (ii) A parameter space is a functor on the category of covariate morphisms. (iii) A sub-parameter is a natural transformation of functors. Some unexpected implications of (iii) will be illustrated by familiar models. ____________________________________________________________________________ For more information ask: Barbara Bogacka School of Mathematical Sciences Queen Mary and Westfield College Mile End Road London E1 4NS Tel: 020 7882 5497 e-mail: [log in to unmask] --------------------------------------------- The seminar information is kept on: http://www.maths.qmw.ac.uk/~rab/seminars.html %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%