Week of April 12 - 16

Monday, April 12

 2:30 GROUPS AND GEOMETRIES SEMINAR  - Room 459 MSC
      Sergey Shpectorov, Mathematics and Statistics, BGSU 
      "Orthogonal and symplectic groups and geometries"

 3:30 Refreshments
 3:45 COLLOQUIUM  - Room 459 MSC
      ***** Note different day *****
      Nathan Feldman, Michigan State University
      "Pure subnormal operators have cyclic adjoints"
      Abstract: In this talk we shall discuss various classes of
        linear operators on Hilbert space, including normal and
        subnormal operators.  We shall be interested in the cyclic
        behavior of these operators and we shall discuss and answer an
        old problem about subnormal operators.  Several examples will
        also be given.

Tuesday, April 13

 4:00 STATISTICS SEMINAR  - Room 459 MSC
      Arthur Yeh, Applied Statistics and Operations Research, BGSU

Wednesday, April 14

 2:30 GROUPS AND GEOMETRIES SEMINAR  - Room 459 MSC
      Sergey Shpectorov, Mathematics and Statistics, BGSU
      "Orthogonal and symplectic groups and geometries"
                                                             
 3:30 ALGEBRA SEMINAR  - Room 459 MSC
      Cecile Huybrechts, Queen Mary College (University of London)
      "The flavor of diagram geometry"
        This talk is supposed to be very elementary, with many
        examples and nice pictures.

 3:30 STATISTICS SEMINAR  - Room 304 MSC **** Note room ****
      G. P. Patil, Distinguished Lukacs Professor, BGSU
      "Environmental sampling and observational economy with emphasis
       on encounter sampling, composite sampling, ranked set sampling,
       and adaptive cluster sampling"
      Abstract:  
      Encounter Sampling: Surveys for monitoring changes and trends in
        our environment and its resources involve some unusual
        conceptual and methodological issues pertaining to the
        observer, the observed, and the observational process.
        Problems that are not typical of current statistical theory
        and practice arise.  In statistical ecology and environmental
        statistics, the theory of weighted distributions provides a
        perceptive and unifying approach for the problems of model
        specification and data interpretation within the context of
        encounter sampling.  Appropriate statistical modeling
        approaches help accomplish unbiased inference in spite of the
        biased data and, at times, even provide a more informative and
        economic setup.

      Adaptive Sampling: Several ecological and environmental
        populations are spatially distributed in a clumped
        manner. They are not very efficiently sampled by conventional
        probability based sampling designs.  Adaptive sampling is
        therefore introduced as a multistage design in which only the
        initial sample is obtained using a conventional probability
        based procedure.  When the variable of interest for a sampling
        unit satisfies a given criterion, however, additional units in
        the neighborhood are selected in the next sampling stage.
        This procedure is repeated until no new units satisfy the
        criterion, or the conditions of a stopping rule are satisfied.

        With the recent growth of geographic information systems
        (GIS), spatial data coverages for landscapes are becoming
        universal.  Such information, obtained mainly from digitized
        maps and remotely sensed sources, may provide a powerful aid
        to adaptive cluster sampling for increasing the efficiency of
        sampling clustered populations from across a two-dimensional
        surface.
     
      Observational Economy: Sampling consists of selection,
        acquisition, and quantification of a part of the population.
        While selection and acquisition apply to physical sampling
        units of the population, quantification pertains only to the
        variable of interest, which is a particular characteristic of
        the sampling units.  A minimum requirement is that
        identification and acquisition of sampling units be
        inexpensive as compared with their quantification.

      Composite Samples: Composite sampling has its roots in what is
        known as group testing.  An early application of group testing
        was to estimate the prevalence of plant virus transmission by
        insects.  In this application, insect vectors were allowed to
        feed upon host plants, thus allowing the disease transmission
        rate to be estimated from the number of plants that
        subsequently become diseased.  In light of recent
        developments, composite sampling is increasingly becoming an
        acceptable practice for sampling soils, biota, and bulk
        materials.

        A recent breakthrough with composite samples may be worth
        mentioning.  The individual sample with the highest value,
        along with those individual samples comprising an upper
        percentile, can now be identified with minimal retesting.
        This ability is extremely important when "hot spots" need to
        be identified such as with soil monitoring at a hazardous
        waste site.

      Ranked Set Samples: Ranked set sampling is a little known method
        of sampling that allows the use of auxiliary information for
        improving upon the performance of simple random sampling.  The
        primary requirement is the ability to rank small sets of
        sampling units with respect to the variable of interest
        without actually measuring that variable.  Subjective
        judgment, prior experience, visual inspection, and concomitant
        variables are among the types of auxiliary information that
        may be used to achieve the ranking.  The method does not
        prescribe any specific form or structure for the auxiliary
        information and the method is accordingly quite robust.
        Errors in ranking are permitted, although the better the
        ranking, the better the performance of the method.

        Ranked set sampling (RSS), induces stratification of the whole
        population at the sample level, and provides a kind of double
        sampling estimator that is robust.

Friday, April 16

 3:15 Refreshments
 3:45 COLLOQUIUM  - Room 459 MSC
      Rod Little, University of Michigan
      "Multiple imputation for missing data in clinical trials"
      Abstract: Multiple imputation is a useful tool for handling
        missing data in statistical analysis, but it has received
        limited use in clinical trials. I review the basic concepts,
        theory and application of the method. I then discuss strengths
        and weaknesses of multiple imputation compared with
        alternative approaches to missing data in clinical
        trials. Multiple imputation has a number of useful properties
        for clinical trial settings. In particular, the method (a)
        corrects the major deficiencies of single imputation methods,
        (b) promotes uniform treatment of the missing values across
        analyses, (c) allows the incorporation of information into the
        imputations that is not used in the main analysis, (d) limits
        the effects of model misspecification to the imputations
        themselves, and (e) allows the assessment of sensitivity to
        plausible alternative imputation models. These features are
        illustrated using an application of multiple imputation to a
        clinical trial on Tacrine for the treatment of the Alzheimer's
        Disease, previously discussed in Little and Yau (1996).

      About the Speaker: Dr. Roderick Little is Professor in the
        Department of Biostatistics and Statistics at the University
        of Michigan, Ann Arbor. He is also Chairman of the
        Biostatistics Department.  He is the author of numerous
        research papers, and is co-author with Donald Rubin of
        Statistical Analysis with Missing Data.  He was a former
        Editor of the Journal of the American Statistical Association.