\magnification=1200
\baselineskip=20pt
\nopagenumbers
\font\big=cmr12 scaled \magstep2

\centerline{\bf STANFORD UNIVERSITY}
\centerline{\bf DEPARTMENT OF STATISTICS}
\centerline{\big DEPARTMENTAL SEMINAR}
\bigskip

\baselineskip=12pt
\centerline{3:30 p.m., Wednesday, March 12, 2003}
\centerline{Sequoia Hall Room 200}
\centerline{(Cookies at 4:30 in 1st Floor Lounge)}

\bigskip
\baselineskip=15pt
\centerline{\sl Jiayang Sun}
\centerline{\sl Case Western Reserve University}
\bigskip
\centerline{\bf Mixture Problems and Biased Sampling}
\bigskip

In the first part of this talk, I'll describe a new test for testing 
the homogeneity in a mixture distribution.  Our new test, D-test, with 
a weighting function, is competitive to the recent modified likelihood 
ratio test which is distribution-free  and   locally  most  powerful, 
asymptotically. In addition, the D-test is more in agreement with 
human visual perception and  has closed form expressions in terms 
of parameter estimates, while the likelihood ratio type tests do not. 
The convergence rate of the D-test is established.  Its application will
be illustrated via an analysis of a financial data set and a 
medical data set.  Some new visualization tools 
will be suggested for choosing the optimal weighting function and its 
potential in data mining will be discussed. 

In the second part of the talk, if time permits, I'll talk about 
new procedures for estimation when survival data come with both 
censored information and selection biases. 

The first part of talk is based on the joint work with R. Charnigo 
and the second part is with B. Wang. 


\bye