BOWLING GREEN STATE UNIVERSITY
          DEPARTMENT OF MATHEMATICS AND STATISTICS CALENDAR
 
                       Week of October 27 - 31

Monday, October 27

 3:30 INVITED STATISTICS SEMINAR  - Room 459 MSC
      Lev Klebanov, BGSU and St. Petersburg State University
                             for Architecture and Civil Engineering
      Model Construction in Statistical Estimation Theory

Tuesday, October 28

11:30 ALGEBRA SEMINAR  - Room 447 MSC
      Sergey Shpectorov, Dept. of Mathematics and Statistics, BGSU. 

Wednesday, October 29

 6:00 KME EVENT  - Room 459 MSC
      Jim Albert, Dept. of Mathematics and Statistics, BGSU. 
      "The Statistics of Baseball"
      Pizza and pop will be provided.

Thursday, October 30

 3:30 SCIENTIFIC COMPUTING SEMINAR  - Room 459 MSC
      TBA

Friday, October 31

 3:30 Coffee
 3:45 COLLOQUIUM  - Room 459 MSC
      Robert L. Strawderman, Dept. of Biostatistics, University of Michigan
      "Approximately exact inference for the common odds ratio
       in several 2 by 2 tables"
      Abstract:  The conditional maximum likelihood estimator of the
        common odds ratio in a sequence of independent 2 by 2 tables
        is known to be superior to the Mantel-Haenszel estimator in
        terms of asymptotic efficiency and has the further advantage
        that its exact distribution is known.  

        However, a long-standing barrier to the widespread use of this
        estimator has been computational intractability; in
        particular, the calculation of significance levels, confidence
        sets, and power based on the exact distribution requires fast
        and efficient algorithms.  An important class of such
        algorithms form the basis of StatXact (Cytel, 1992), a
        software package able to solve various aspects of the exact
        inference problem, including that for a sequence of several 2
        by 2 tables in real time.
        
        In this talk, we develop an alternative methodology by
        establishing a useful Lugannani-Rice-type saddlepoint
        approximation to the exact distribution of the conditional
        maximum likelihood estimator.  The approximation is derived
        from an interesting representation for hypergeometric random
        variables recently developed in Kou and Ying (1996a,b), and
        provides fast, accurate calculations of p-values, confidence
        sets and power functions.  The primary computational burden is
        in determining the roots of a certain well-studied polynomial,
        which need be done numerically but only once for each table.
        Consequently, the required computational effort is typically
        minimal; for example, all of the examples were done using code
        written by the authors entirely in S-plus.

        Joint work with Marty Wells, Cornell University

-----------------------------------------------------------------------
This announcement and a schedule of future colloquia are available on
the Worldwide Web; see http://www.bgsu.edu/departments/math/.
If you wish to be placed on the e-mail distribution list, or have 
comments or material for the calendar, send email to