Experimental Design & Probability
Biology 174H is a course in experimental design and statistics. It differs from many other statistics courses by integrating statistical concepts into the whole process of doing science, rather than something you just do after you've collected your data. After a brief introduction to probability and the basic concepts of statistical inference, the course focuses on the Analysis of Variance as a tool for asking questions of nature and designing informative experiments to answer them. We cover the mechanics of the technique, including identification of treatments & appropriate controls, why you need to have replicates of everything, nested designs, orthogonal designs with interactions (you'll learn what all these terms mean during the course), computational methods, and how to detect significant differences among subsets of treatments after the overall analysis. We also delve a bit into the mathematical basis for the technique (mostly just algebra) in order illustrate its underlying elegance.
However, the primary emphasis of the course is on the proper interpretation of the results and their implications for further studies. These days, anyone can obtain computationally accurate results from computer software, but fewer people truly understand what those results actually mean. So we go beyond just rote interpretation of P-values and explore what these techniques can tell you about the effects of your experimental treatments (and more importantly what they cannot). You'll discover that the usefulness of these analyses hinges on the logic of your initial question and the clarity with which it has been defined.
We also cover topics that are not typically part of an introductory statistics course, such as statistical power and determination of the appropriate underlying model for any design arising from an appropriately articulated question. These are important tools for building better experiments and doing better science.
By the end of the course I hope you'll find that statistical considerations are not just something you do after you've completed an experiment, but are instead an integral part of articulating your question, clarifying your thinking, and designing the most effective experiment given the ever-present constraints of time & resources.