* * * * PLEASE NOTE UNUSUAL LOCATION * * * *
SNEDECOR LECTURE

Statistical Laboratory
Iowa State University
 
DATE AND TIME: Monday, October 10, 2005, 4:10 p.m.
 
PLACE: 1213 Hoover
 
SPEAKER: Steve Portnoy, Department of Statistics, University of Illinois, Champaign
 
TITLE: Regression Quantiles:  Reveling in the Charm of Variety
                                   
 
ABSTRACT
Consider the ubiquitous problem of analyzing a response, Y, in terms of one or more explanatory variables, x.  Since the early 19th century, this has been approached using the paradigm of Gaussian errors: estimate the mean response as a function of x and summarize all population variability in terms of a single standard deviation parameter.  However, the dangers of this approach have become increasingly clear in recent past.  Gaussian "least squares" methods are highly sensitive to rather minor departures from distributional assumptions (for example, they fail spectacularly in the presence of outliers).  Perhaps more fundamentally, they ignore important sources of population heterogeneity. More than 100 years ago, Francis Galton suggested that the percentiles (or quantiles) of the response provide a far more complete and satisfactory approach. For linear statistical models, regression quantiles were introduced by Koenker and Bassett in the late 1970's, and underwent an extensive development over the past thirty years.  A vast array of generalizations and complements are undergoing active investigation today.  The basic ideas will be presented using historical examples, including work of Boscovich in 1755 that predated the development of least squares methods.  After reviewing the case of traditional linear models, some recent work on generalizations will be presented in terms of examples for censored survival data.
 
COFFEE:  3:30 p.m., 104 Snedecor Hall
 
Seminar schedules and abstracts are available via WWW:  http://www.stat.iastate.edu/

Jeanette La Grange
Department of Statistics
102 Snedecor
Iowa State University
Ames, IA  50010-1210
515 294-3440 (office)
515 294-4040 (fax)
http://www.stat.iastate.edu/directory/staff/jeanette.html