Biometry - BIO 6xx (4 hours)
Statistics in Ecology, Evolution, Behavior and Genetics
(rough draft on new course proposal)
2a. Prefix and number: BIO, 600 Level
2b. Title: Biometry: Statistics in Ecology, Evolution, Behavior and Genetics
2c. Actual contact hours: 4 hours per week lecture/lab
2d. Grading system: Letter
2e. Credit hours: 4
2g. Course description: This class emphasizes application and presentation of the statistical procedures that are commonly encountered in the fields of ecology, evolution, behavior and genetics. Students will be presented the full experience of designing experiments, analyzing data (provided by the instructor or the student), and presenting the statistical results in standard scientific format (tabular, graphical). Students will work with datasets from their own research specialties.
This class is intended to complement, rather than replace, classes in statistics offered by the Statistics Department. It assumes the student already has had at least a beginning-level graduate class in applied statistics, and preferably two such classes. It covers a lot of ground very quickly, and draws on a diverse array of procedures (e.g. experimental design, univariate techniques, multivariate techniques, time series, survival analysis) from several different courses in applied statistics from the Statistics Department.
2h. Prerequisites: STA 570 or equivalent, STA 671/672 strongly recommended, or consent of instructor
4. Cross listed: with Entomology and Forestry
5. Requested starting date: Spring Semester 2010.
6. Offered: Spring Semesters
7. How often offered: Alternate years
8. Why this course is needed: The range of statistical procedures used by researchers in the fields of ecology, evolution, behavior and genetics spans several courses in the Statistics Department, which include STA 671 Regression and Correlation, STA 672 Design and Analysis of Experiments, STA 673 Distribution Free Statistical Inference and Analysis of Categorical Data, STA 675 Survey Sampling, STA 677 Applied Multivariate Methods, STA 679 Design and Analysis of Experiments II, STA 626 Time Series Analysis, STA 630 Bayesian Inference, and STA 635 Survivability and Life Testing (Note: although STA 626, 630 and 635 cover techniques that are widely used in ecology, evolution, behavior and genetics, these particular courses are geared for Statistics graduate students, rather than applied graduate students from departments like Biology, and have a more rigorous set of prerequisites than the other courses I listed, STA 671, 672, 673, 675, 677, 679). Most students in Biology take only one or two semesters of Statistics Department courses (STA 570, and maybe STA 671/672), and thus never get formal exposure to many of the procedures that they will encounter in the professional literature, and that they will need to master for their own research. The proposed Biometry course will cover some procedures from each of these courses, and thus give students exposure to the broad range of statistical methods used in their fields. Students will be encouraged to pursue higher level courses in the Statistics Department where greater depth in a specific topic would improve the quality of their research, and if appropriate, to obtain the Graduate Certificate in Applied Statistics from the Statistics Department.
9a. Instructor: Craig Sargent
9b. NO new facilities are required.
10. Anticipated Enrollment: 10-15 every two years (based on enrollments in other BIO 600-level courses)
11. Intended Audience: Biology graduate students in the fields of ecology, evolution, behavior and genetics, and similar graduate students from other departments across campus (e.g. Entomology, Forestry).
14. Applicability to requirements for degrees and certificates: Would count as a 600-level elective for graduate degrees in Biology (MS, PhD), and possibly as an elective for the Applied Statistics Certificate in the Statistics Department.
16. This course would NOT change the degree requirements of any program on campus.
17. Major teaching objectives, syllabus, reference list. (In progress, but see below)
This class emphasizes application and presentation of the statistical procedures that are commonly encountered in the fields of ecology, evolution, behavior and genetics. Students will be presented the full experience of designing experiments, analyzing data (provided by the instructor or the student), and presenting the statistical results in standard scientific format (tabular, graphical). Students will work with datasets from their own research specialties. Students will be introduced to several software packages (SAS mainframe, SAS PC, JMP (Mac), Systat, R), and allowed to use whichever best suits their level of comfort or needs. After this class, students should feel comfortable reading the statistical procedures from research publications in their field; designing, analyzing and presenting their own research experiments; and, delving more deeply into the applied statistical literature as necessary.
Class will meet twice weekly for 2 hours per session. These sessions will be a mixture of lecture, demonstration, and hands-on student interaction with statistics. There will be regular homework assignments (ungraded), four graded take-home exams (worth 15% per exam), and a final project (worth 40%).
Emphasis on the topics (below) will depend on the backgrounds of the students in the class. Students who have had only STA 570 will spend more time on items 1-5 than students who also have had STA 671 and 672.
There will be considerable reading for this course, including chapters from three textbooks (Gotelli & Ellison, Scheiner & Gurevitch, Sokal & Rohlf, see below), handouts and online material.
Textbooks:
Gotelli and Ellison. A
Primer of Ecological Statistics. Sinauer.
(link) Author websites: Gotelli,
Ellison
Scheiner and Gurevitch, eds. Design
and Analysis of Ecological Experiments.
Sokal and Rohlf. Biometry.
W.H. Freeman. (link)
Author Websites: Sokal,
Rohlf
Software:
SAS
(PC, Mainframe)
JMP (Mac, PC, Mainframe)
Systat
(PC)
R (PC, Mac,
Linux)
Lecture Topics:
1.
Introduction
1.1. Hypothesis Testing, Software (SAS, JMP, Systat, R)
1.2. Type I and Type II error, Power, Data Types
1.3. Parametric versus Nonparametric Statistics and
Assumptions
2.
Parametric Statistics
2.1. Tests for Normality
2.2. Data Transformations
2.3. The t-test and introduction to Analysis of
Variance
3.
Experimental Design
3.1. Power Analysis and Effect Size
3.2. Optimization of Resources
3.3.
Pseudoreplication
4.
Univariate Analysis of Variance
4.1. One Way
4.2. Factorial
4.3. Model I, II and Mixed Models
4.4. Nested
4.5. Split Plot
4.6. Blocking
4.7. Repeated Measures
4.8. Recreating all of the above designs out of a
completely factorial model without replication
4.9. Analysis of Covariance
5.
Regression and Correlation
5.1. Assumptions of Regression versus Correlation,
Model I and II Regression
5.2. Linear Regression
5.3. Piecewise Linear Regression
5.4. Nonlinear Regression
5.5. Polynomial Regression
5.6. Multiple Regression
5.7. Multiple and Partial Correlation
5.8. Path Analysis
5.9. Logistic Regression
6.
Parametric Statistics, Multivariate
6.1. Analysis of Variance
6.2. The Multivariate "Normal" (matrix) Equation: [β]
= [XtX]-1[XtY], and its application to
univariate and multivariate models
6.3. Ordination
6.3.1.
Principle Components Analysis
6.3.2.
Factor Analysis
6.4. Cluster Analysis
7.
Nonparametric Statistics
7.1. Categorical Data
7.1.1.
Goodness of Fit
7.1.2.
Contingency Tables
7.1.3.
Log-Linear Models
7.1.4.
Nonmetric Multidimensional Scaling
7.2. Ranked Data
7.2.1.
Analyzing Ranks in lieu of ANOVA
7.2.2.
Correlation
8.
Special Topics
8.1.
Randomization
8.1.1.
Monte Carlo
8.1.2.
Jackknifing
8.1.3.
Bootstrapping
8.2.
Meta-analysis
8.3.
Time Series
8.3.1.
Fourier Analysis
8.3.2.
Autocorrelation
8.3.3.
Cross Correlation
8.3.4. Time
Series Techniques Applied to Spatial Data
8.4.
Survival Analysis
8.4.1.
Logistic Regression
8.5.
Bayesian Inference
References:
Syllabi for similar
courses at other universities
Washington
State University
Texas
A&M
Statistics
Department
Course
Listings
Graduate
Certificate in Applied Statistics