PHILOSOPHY OF THE COURSE (With material from R. A. Pimentel)
Biology is concerned with understanding the complexities of natural processes. In order to unravel interactions, biologists need quantitative tools. Biology, in contrast to many other fields, is a quantitative, experimental discipline. The documentation of natural patterns, design of robust experiments, and testing of hypotheses all require statistical concepts. However, an understanding of statistics is not enough when large amounts of data are involved. An additional tool is required: a computer.
Anyone can learn to run statistical programs on a computer and obtain output. Unfortunately, for the output to be meaningful two criteria must be meet: 1) the data must be collected properly (REMEMBER: garbage in = garbage out); and 2) the proper statistical analyses must be chosen. The difference between a number cruncher and a data analyst is that the latter produces real results. Knowing how to do things correctly can only be accomplished in one way: a thorough understanding of the material presented.
In contrast to other scientific fields, understanding statistics must involve work beyond that of listening to lectures. Exposure to real data and problems (hopefully your own!) is the only means of obtaining this goal. I will make every effort in this class to expose you to as many data as possible. For your part, it is essential that you understand what is presented and that you know what you are doing before you try to do it. Thus, assignments must be thoroughly read prior to solving them on the computer. You will not learn statistics on the computer, but only after intensive examination of an exercise.
GUIDELINES FOR WRITING UP HOMEWORK ASSIGNMENTS
Write ups of daily assignments are one way to accomplish these goals. By explicitly following each step of an analysis and writing out your interpretation you will clarify thoughts that were previously unclear. For all assignments, include the following in your homework:
An abstract of your results: this should describe, in 1-2 paragraphs, what the goals of your homework were, what you did, what you found, and what it means.
In addition, the following information should accompany each statistical problem, and should be written directly on the output.
- Describe the problem. What was the purpose of the exercise and how were the data collected. Describe the experimental design and any variables that were uncontrolled (i.e., were samples obtained at different times).
- State a Null Hypothesis. In a single statement make a clear, concise null hypothesis. State it in two ways: in strict statistical terms, using parameters, and in scientific terms. (This step may not apply to all assignments)
- List the assumptions of the statistical analysis. Even though you may do this a hundred times do not skip this step! If any assumptions can be tested be sure to run the appropriate test prior to the main test and include it in the output.
- Determine the level of significance. This is always done prior to a test and is traditionally set at alpha = 0.05.
- Provide a statistical interpretation of the results. State a clear acceptance or rejection of the null hypothesis and list all of the statistics involved.
- Provide a scientific interpretation of the results. Using scientific terms state a clear explanation of the significance of the analysis. THIS is the most important part! Discussion should be focused on what you have learned from your analysis. What do your results mean? How can they be improved? Did you accept the null hypotheses because it is true or because your sample size was too low or the system too variable?
All of the above should be written in as few words as possible. One or two sentences is usually sufficient, except for the interpretation, which should be at least a paragraph (after all this is why we're doing it the first place)
NOTE ON WORKING TOGETHER ON ASSIGNMENTS: It is OK, and in fact I encourage you to work with your classmates on your homework. However, you should individually write up your assignments. If any homework assignments are reasonably identical neither party will receive credit for the assignment.