&nbsp;Seminar Notice<DIV>
<DIV></DIV><BR>
<DIV>&nbsp;Statistical Laboratory</DIV>
<DIV>&nbsp;Iowa State University</DIV><BR>
<DIV>
		DATE AND TIME:&nbsp;<b> Friday, January 28, 2005, 4:10 p.m.</b></DIV><BR>
<DIV>
		PLACE:&nbsp; <b>319 Snedecor</b></DIV><BR>
<DIV>
		SPEAKER: <font color="#000099"><b>Howard Bondell</b></font>, Department of Statistics, Rutgers University, New Brunswick, NJ</DIV><BR>
<DIV>TITLE: Robust Logistic Regression via the Case-Control 
Formulation</DIV><BR><BR>
<DIV>ABSTRACT</DIV><BR><BR>
<DIV>It is well known that the maximum likelihood fit of the logistic regression 
parameters can be greatly affected by atypical observations. Several robust 
alternatives have been proposed in the literature and implemented in standard 
statistical software packages.&nbsp; However, upon considering the model via the 
case-control viewpoint, it is clear that current techniques can exhibit poor 
behavior in many common situations.</DIV><BR>
<DIV>A new robust class of estimation procedures is introduced.&nbsp; The 
estimates are constructed via a minimum distance approach after identifying the 
model with a semi-parametric biased sampling model.&nbsp; The approach is 
developed under the case-control sampling scheme, but is applicable under 
prospective sampling as well. Estimators resulting from this minimum distance 
methodology are shown to compare favorably with existing methods used in 
logistic regression.</DIV><BR>
<DIV>An alternative version of the current "bounded influence" class is also 
introduced in order to alleviate the difficulties found to be associated with 
the existing procedures.</DIV><BR>
<DIV>These new approaches can be highly efficient if the model is true, while 
remaining robust to small deviations in the model.&nbsp; Thus they can be used 
to fit the logistic regression model if it is appropriate for the bulk of the 
data, even in the presence of atypical observations</DIV><BR>
<DIV>COFFEE:&nbsp; 3:45 p.m., 104 Snedecor Hall</DIV><BR>
	<DIV>
		</DIV>
</DIV>