The MSSISS 2011 keynote speaker was Dr. Terry Speed from UC Berkeley.
Models for Statisticians
Try typing “What is a model?” into Google (and hitting enter) and see what you get. It’s a jungle out there, and not all that you find is relevant to us as statisticians. You could repeat the exercise with “What is a statistical model?” and see answers ranging from a banality to category theory. We’ve all heard and probably repeated George Box’s dictum – he calls it a quaquaversal quotation –
All models are wrong, but some models are useful.
(Sometimes wrong is given as false.) I’ll start my talk by disagreeing with this statement. What is a model, what are models for, how should we use them, how can we avoid misusing them, and can we do without them (“design-based” vs. “model-based”)? These are a few of the questions I’ll touch upon during my talk, which will focus on questions, not answers. However, in my view models play a key role in providing answers, once we have questions. The general drift of my approach is that models mediate between questions and answers. As statisticians we use data, so my summary is that for statisticians, it goes like this:
questions + data —models–> answers.
I’ll close with a 1983 statement by the late Basil Rennie, a British mathematician who spent the latter part of his career in Australia. It raises the question of whether we need to deal with reality when discussing models. All thought and all communication is modelling, and most misunderstandings arise by someone confusing either a model with reality or one model with another.
Student Organizing Committee
|Brady West||Survey Meth.||bwest[at]umich[dot]edu|
Faculty Advisory Committee
|Michael Elliot||Surv Meth.||mrelliot[at]umich[dot]edu|