Model Selection and Multimodel Inference in the Life Sciences
Dr. David R. Anderson
Thursday and Friday, 8 a.m. â€“ 5 p.m.
26th & 27th August 2010
Registration is $250 per person and includes a copy of Model based inference in the life sciences: a primer on evidence, lecture handouts, snacks during the workshop, and undoubtedly a wealth of knowledge. Please register early since space is limited. If interested please contact John Skinner (see contact information below).
ABOUT THE WORKSHOP
This 2-day session introduces a variety of general analysis methods based on Kullback-Leibler information. These new methods are useful in model-based inference in the analysis of empirical data in the sciences. The material focuses on science hypotheses, models, and model selection methods such as AIC, AICc, and QAICc. After introducing important background material, methods are introduced to make formal statistical inference from more than a single model (multimodel inference). These approaches include types of model averaging, incorporating model selection uncertainty into estimates of precision, dealing with model selection bias, ranking the importance of predictor variables, and confidence sets on models. The material is not deeply mathematical; the emphasis is on science concepts and philosophy and many examples are used to aid applications. These are informal sessions with substantial time for discussion and debate and getting side-tracked onto interesting related issues. This is not a course in how to derive mathematical models to represent various science hypohteses or management positions, although many examples will be provided relating to these issues. Likewise, it is not a course on estimation methods for model parameters, however, some time will be spent outlining the behind least squares and maximum concepts likelihood estimation.
Graduate students, post-docs, faculty and government scientists can make use of new approaches in this course in writing grant proposals, defending budgets, planning their research, analyzing their data, and reaching careful inferences based on quantified evidence (usually multimodel inference). Resource managers can base management decisions on better science and attempt effective resolution of some types of conflict over resource use with these approaches. The methods to be presented have potential application across the sciences, including the life sciences and social sciences (including human dimension issues). There is potentially a â€ślotâ€ť in it for virtually everyone. Participants in this course will develop an understanding of the advantages and deeper issues associated with model selection and multimodel inference. This short course will be based on the text book,
Anderson, D.R. 2008. Model based inference in the life sciences: a primer on evidence. Springer, New York, NY. 184pp.
Attendees will find this course highly interactive and fun. The days will go by quickly!
Attendees should have a reasonable background in applied statistics. Beyond the basic concepts of measures of variation, covariation, natural logarithms, and subscripted variables, people need a decent understanding of least squares â€śregressionâ€ť (e.g., Î˛ values, residual variance, residual sum of squares, R-sq, residuals), goodness-of-fit issues, sampling applications, and simple experimental designs.With the above background, people will find the quantitative material easy; it is the philosophical and conceptual issues that are challenging to nearly everyone.
This is not a modeling course. I will showcase a variety of models and attendees will gain some insights into modeling, but the short course does not focus on model building or the estimation of model parameters (via maximum likelihood or least squares). In addition, little is said about various random effects models.
If interested, please download a tentative course outline.
ABOUT THE INSTRUCTOR
David Anderson spent most of his professional life as a research scientist with the U.S. Department of the Interior. He holds a PhD in Theoretical Ecology from the University of Maryland and has worked in a wide variety of quantitative areas in the biological sciences. He has worked intensively on model selection and related subjects since 1990, begining with joint work with Drs. Jean-Dominique Lebreton and Jean Clobert (France) and Kenneth Burnham (USA). During this time he has published 18 journal papers and two editions of the Springer-Verlag book on model selection, multimodel inference, and closely related topics (see below). Much of this work has been done in close collaboration with Drs. Kenneth P. Burnham and Gary C. White.
Dr. Anderson has published 15 books and research monographs; 99 papers in peer-reviewed national/international scientific journals; 45 book chapters, government scientific report series, and conference proceedings and transactions; and 15 technical reports in ecology and other life sciences and statistical science. He was a Senior Scientist with the U.S. Geological Survey and is now president of Applied Information Company in Fort Collins, CO. Further information on Dr. Anderson can be found at http://www.cnr.colostate.edu/~anderson/.
LOCATION & SCHEDULE
This workshop will be held in the Bear Mountain Conference Room at the Alaska SeaLife Center, 301 Railway Ave., Seward, AK 99664
Beautiful Seward, Alaska is located 120 mi (~2-1/2 hour drive) south of Anchorage. There are several hotels within walking distance of the Alaska SeaLife Center but accommodations should be booked early since tourist season runs through September and hotels can become full for the season. Click here to see a map of hotels in Seward. All attendees will receive free admission to the Alaska Sealife Center. Some snacks and refreshments will be provided during the workshop.
The workshop will run approximately 8am - noon, 1pm - 5pm on Thursday and Friday.
Why not spend the rest of your weekend and explore Seward! We have one of the most road accessible glaciers in Alaska, tours boat that will take you to Kenai National Park, and you can find fishing charters almost every day.