1997/1998 CANR BiG Seminar Series

In the Spring semester 1998, the CANR Biometry Group organized its inaugural Biometry Seminar Series. Featured speakers were distinguished statisticians, that have influenced the development of biometry and statistics in agriculture and natural resource disciplines for many decades.


Noel A. Cressie Title: Spatial Statistics and Spatial Mixture Modeling
Abstract: Dr. Cressie’s talk will addresses the problem of how to build hierarchical (mixture) models in situations where the parameters of interest are in fact random processes in space and time, and where Gaussian assumptions no longer necessarily hold. The seminar will conclude with an analysis of forest health data, carried out jointly with Mark Kaiser and Jaehyung Lee, where a spatial beta-binomial mixture model is fitted to the data.
The speaker: In the past two decades a paradigm shift occurred in many empirical sciences, emphasizing the innate spatial and temporal nature of the processes we observe in contrast to cross-sectional snap-shot images. Theory and tools for measuring, modeling, analyzing and interpreting these processes developed in parallel. Only few have contributed to the field of spatial statistics as Dr. Noel Cressie has. In the past two decades were defined, advanced and widely recognized through his leadership. In 1993, Cressie's seminal textbook "Statistics for Spatial Data" (Wiley and Sons, New York, 900 pp.) gave proof to this outstanding accomplishment.
Host: Oliver Schabenberger

George A. Milliken Title: Graphical representation of designed experiments with applications.
Abstract: Dr. Milliken will be speaking on graphical representations of designed experiments. Determining or identifying the structure of a designed experiment is necessary before an appropriate model can be constructed for analyzing the resulting data. Graphical representations of the four basic building blocks of designed experiments are presented and their analyses are motivated. The four basic building blocks are the completely randomized design, the randomized complete block design, the split-plot design and the strip-plot design. Examples will be used to demonstrate the use and value of the graphical representations.
The speaker: Dr. Milliken is a world renowned applied statistician who regularly consults for industry and agricultural/biological researchers. He has provided international leadership on the development and adoption of new statistical methods and critically important expositions on the proper use of statistical software like SAS for the analysis of experimental designs. His 2 volume series entitled "Analysis of Messy Data" coauthored with Dr. Dallas Johnson is a popular and invaluable resource for experimental scientists.
Host: Robert J. Tempelman


Robert R. Sokal Title: Local spatial autocorrelation in biological variables.
Abstract: Most biological variables are not randomly distributed over space, instead they are correlated over various spatial scales. Standard spatial autocorrelation (SA) statistics are global, measuring overall autocorrelation throughout the sampled range. Recently, SA methods have been extended to include the detectionof local SA at individual sampling stations.This seminar examines these new statistics and tests their applicability in spatially distributed  biological data.
The speaker: Dr. Sokal is one of the foremost authorities in the application of statistics to biological issues. He has authored several books including Biometry (3rd. Edition, 1995), which has been a leading text for several decades. He is an international leader in research focusing on geographic and spatial
distributions of genetic variation in humans and other species. Most relevant to this seminar, his work in spatial statistics is singularly responsible for
the introduction and motivation of spatial autocorrelation statistics in the biological sciences. The topic of the seminar promises to have similar impact.
Host: Bryan Epperson
 

Shayle R. Searle Title: Evolution of Analysis of Variance: from R.A. Fisher to generalized linear mixed models.
The speaker: Dr. Searle is a Professor Emeritus from Cornell University where he was a Professor of Biological Statistics in the Biometrics
Unit.  Presently he is a statistical consultant with a broad range of interests. He is world-renowned for his expertise in analysis of unbalanced data, and has authored several books including "Linear Models for Unbalanced Data", and "Matrix Algebra Useful for Statistics"
Host: Daniel B. Hayes
 

(C) 1998 CANR Biometry Group, Michigan State University.
The CANR BiG Seminar Series is sponsored jointly by the Michigan Agricultural Experiment Station and the
College of Agriculture and Natural Resources at Michigan State University. The support is gratefully acknowledged.
Last updated July 9, 1998