Course Description
Statistical concepts and principles for both experimental and observational investigation in the ecological sciences. Students will explore the fundamental and advanced issues in ecological/environmental measurement processes that will aid scientific discovery. Students will learn hypothesis driven search skills and weight of evidence portrayals of research findings.
Additional Requirements for Graduate Students:
Graduate students will be required to identify a project
approved by their faculty adviser that will aid their thesis or
dissertation research by demonstrating their understanding of
the principles of statistical thinking in ecology. This
demonstration can either be in the form of a 20 minute
presentation, or a written statistical methods section toward
their thesis or dissertation that describes the field/lab
methods that will be used to adequately address sampling and
measurement variation in their data collection scheme. With
either choice, the graduate student will also be required to
discuss issues that may or may not fully be addressed by the
intended data analysis method.
Athena Title
STAT THINK ECOL
Non-Traditional Format
This course will be offered as a Maymester at the UGA Savannah River Ecology Laboratory (SREL) Conference Center, near Aiken, SC. Class size will be restricted to 15-16 students.
Prerequisite
(STAT 2000 and ECOL(BIOL) 3500-3500L) or permission of department
Semester Course Offered
Not offered on a regular basis.
Grading System
A - F (Traditional)
Course Objectives
Students will learn to give statistical thinking to their ecological research interests before collecting one single data point. They will learn the philosophy of the Hypothetico- deductive loop for scientific inquiry, how to craft sampling plans, and common experimental design flaws or missing features. They will learn to distinguish between measurement and sampling uncertainty, the essential concept of variance partitioning, and the estimation of sampling variability via randomization methods (e.g., bootstrap sampling) and simulation. Finally, they will learn the value of portraying the weight of evidence in ecological research via data display that may or may not require a formal statistical inference.
Topical Outline
Lecture 1. Statistical Thinking A. It’s the science of science 1. Peeling the artichoke (on falsification) 2. Eureka! (On verification - HD loops) 3. What about corroboration? B. What scientists really want to know 1. The frequentist in most of us 2. The Bayesian in all of us Lecture 2. The Essential Beginning A. Formulating the right problem 1. What you think is the problem vs. what is the problem 2. It’s a matter of perspective, and a function of experience 3. The role of consulting - and not just with statisticians B. Scope of inference 1. What you can say vs. what you can't say 2. It’s in how you collect the data Lecture 3. Distinguishing Multiple Sources of Variation A. Random variation: measurement processes B. Systematic variation: natural factors C. Bias Lecture 4. Identifying the Experimental Unit A. Measurement scale B. Scale transformation Lecture 5. Purposeful Investigation A. Discovery oriented 1. It’s more about natural history 2. Taxonomically descriptive 3. What we find over here B. Hypothesis driven 1. It’s more about ecology 2. Contextually interpretive 3. Pattern recognition Lecture 6. Sampling Plans A. Addressing specific questions B. Regarding accuracy vs. precision C. The more variable, the more samples D. Avoiding confounding Lecture 7. Weight of Evidence A. Summary data display vs. tabular presentation B. Role of significance testing 1. P-value reliance 2. The erroneous concept of asymptotic certainty Lecture 8. The Statistical Model A. Explaining relationships B. Predicting additional measurements Lecture 9. Drowning in Measures A. N-p dilemma: N = sample size, p = no. response variables measured B. Dimensionality reduction 1. Multivariate methods may or may not apply 2. Indices may be more helpful, but less statistically rigorous 3. The trade-off