***Please excuse cross postings***
***Do Not Respond to the List***
Please direct inquires to Dr. Joe Witt ([EMAIL PROTECTED]; 304.876.7447)
Multivariate Statistical Analysis Techniques for Ecological Data
Course # FIS4400
DATE: January 22 - 26, 2007
LOCATION: National Conservation Training Center (Shepherdstown, West
Virginia)
TUITION: $1050
COURSE DESCRIPTION:
This course covers a variety of descriptive and inferential multivariate
statistical methods that are useful for analyzing biological and habitat
data. Applications include Emphasis is placed on technique assumptions,
applications, and interpretation of results. Participants use multivariate
procedures presented in class to analyze ecological data provided by the
instructor. Case studies covering multivariate analysis of terrestrial and
aquatic field data are discussed. Attendees also may bring their data to
supplement analysis exercises.
Objectives:
â¢Identify the basic concepts of matrix algebra, eigenvalues, and
eigenvectors, and multivariate normality;
â¢Use methods for displaying relationships and position (principal
components analysis, factor analysis, biplot displays, correspondence
analysis, multidimensional scaling, and cluster analysis);
â¢Apply techniques for group separation (MANOVA, canonical variate analysis,
discriminant analysis, logistic regression);
â¢Use techniques for determining relationships between sets of variables
(canonical correlation analysis and canonical correspondence analysis); and
â¢Analyze repeated measures data.
FOR INFORMATION CONTACT:
Dr. Joe Witt (304) 876-7447 Fax: (304) 876-7234
[EMAIL PROTECTED]
National Conservation Training Center Website "http://training.fws.gov/";
or
Department of Interior employees must register through DOI Learn.
Others not employed in the department of Interior may go to
http://training.fws.gov/ and click on "Courses" > "Catalog of Training" >
"Search the catalog as a guest" . Next, search on FIS4400. "Multivariate
Analysis for Ecological Data" will come up. Click on the course title and
then the "Apply" button. Alternatively, contact Joe Witt via email for an
application form that can be printed and faxed.
Closing Date for Applications: November 27, 2007
*****************************************************************************
Draft Course Outline:
Monday
8:30 AM
Welcome and Administrative Details
Introductory materials and mathematics
Lecture 1: Introduction and matrix algebra
Overview of methods
The language of multivariate analysis
Lab 1: Introduction to SYSTAT
Lecture 2: Eigenvalues and eigenvectors, Multivariate normality
The principal tool for summarizing multivariate information
Multivariate normal distribution
Displaying relationships and position
Lecture 3: Principal components analysis
Correlations and covariances
Decomposition of a square symmetric matrix
Linear composites
How many components
Plot of scores
Assumptions
Examples
Lab 2: PCA
Tuesday
8:00 AM
Displaying relationships and position (continued)
Lab exercise discussion
Lecture 4: Factor analysis
Factor analysis model
Methods of estimation
Rotation of factors
Scores
Assumptions and problems
Examples
Lab 3: Factor Analysis
Lecture 5: Biplot display
Singular value decomposition
Properties
Graphical display
Lecture 6: Correspondence analysis
Count data and analysis
Decomposition
Nonlinear relationships and the arch effect
Problems and concerns
Examples
Lab 4: Correspondence analysis
Lecture 7: Multidimensional Scaling
Distance measures
Approximating distances using Euclidian distance
Plotting coordinates
Non-metric scaling
Assumptions and their role
Examples
Lab 5: MDS
Wednesday
8:00 AM
Displaying relationships and position (continued)
Lab 5: MDS (continued); Summary of ordination
Lecture 8: Cluster analysis
Why use it?
Group structure
Distance measures
Hierarchical methods
Other methods
Three choices/assumptions
Examples
Lab 6: Cluster analysis
Group Separation: Testing, Display and Prediction
Lecture 9: Manova
Multivariate tests
Two groups â T-square test
Eigenvalues (again)
Assumptions
Examples
Lab 7: MANOVA
Thursday
8:00 AM
Group Separation: Testing, Display and Prediction (continued)
Lab exercise discussion
Lecture 10: Canonical Variate Analysis
Why use it?
Scores and graphs
Interpretation
Assumptions
Examples
Lab 8: CVA
Lecture 11: Discriminant analysis
Prediction of group membership
Linear and quadratic models for normal data
Classification tables
Assumptions and alternative approaches
Examples
Lab 9: Discriminant analysis
Lecture 12: Logistic regression analysis
Predicting probability of group membership
Maximum likelihood method
Tests and fit measures
Prediction
Comparison with discriminant analysis
Assumptions
Examples
Lab 10: Logistic regression
Friday
8:00 AM
Group Separation: Testing, Display and Prediction (continued)
Lab exercise discussion
Relationships between sets of variables
Lecture 13: Canonical correlation analysis
Relationships between sets of variables
Relating linear combinations
Interpretation
Assumptions
Examples
Lab 11: CCA
Lecture 14: Canonical correspondence analysis
Relating categorical variables and environmental variables
Graphical displays
Detrending
Lab 12: CANOCO program
Lecture 15: Analysis of repeated measures
What are repeated measures
The univariate and multivariate views
Split plot model and univariate analysis
Multivariate analysis
Contrasts for variables
Another approach
Examples
12:00 PM Course Evaluations & Course Ends
