Advanced Biostatistics: Course Description and Objectives

All biological research requires the use of statistics to determine whether or not the data collected in an experiment deviate significantly from null expectations. Unfortunately, biology students spend far too little time investigating what statistical tools they have at their disposal, and instead invest their limited time in learning more biology. While it is certainly necessary to learn as much biology as possible, an undesirable consequence of this choice is that a student’s rudimentary knowledge of statistics relegates their hypothesis testing to methods that they already know: if a student only knows a t-test, the design of all of their experiments begins to resemble that of a t-test! This is extremely limiting, as the range of biological questions they want to ask is essentially hampered by lack of tools, not lack of biological knowledge and intuition. This seminar course is designed as the first-step in alleviating this difficulty.

This course will review the basic univariate and multivariate statistics commonly used in evolutionary and ecological research. The goal of the course is to give students a general idea of what statistical methods are commonly used in evolutionary ecology, which methods are appropriate for which types of data, and to provide a general knowledge of how the methods work. While this course is NOT intended to be a substitute for multivariate courses in statistics, it IS intended to provide sufficient detail so that students may pick up a recent issue of Ecology, Evolution, American Naturalist, etc. and be able to understand the statistical methods used in the articles. This way, students may more critically evaluate the literature, and hopefully, be able to use these methods in their own research as the need arises.

To make the mathematics accessible, I will use a ‘think-first, learn equations second’ approach; beginning with data types and hypotheses, progressing to the choice of statistical method, and finally ending with equations and necessary practical details. Our topics will begin by reviewing basic univariate statistics (ANOVA, regression, correlation), and making the ‘jump’ to multivariate general linear models (GLM: MANOVA, MANCOVA, multivariate regression). We then review exploratory methods, including ordination approaches (CVA, PCA, PCoA, MDS), clustering (UPGMA, WPGMA, K-means), and multivariate 'correlation' (canonical correlation, and PLS). We will also discuss common problems biologists make in data analysis, and if time permits, discuss several ‘specialty’ topics such as the phylogenetic comparative method, resampling methods, spatial statistics, and meta-analysis.

The course will meet once a week for 1 ½ hours. However, I firmly believe that the best way to learn new techniques is from a combination of discussion and hands-on experience. Therefore, there will also be a computer lab each week. In these laboratory periods, we will perform the statistical methods covered in the previous lecture, using example data sets. It is hoped that the exercises in laboratory will reinforce the concepts discussed during lecture, so that the students may better apply these methods to their own research. In computer lab, we will use a combination of software, including Excel (with Pop-Tools), JMP, and R (for statistical programming).

Return to Syllabus