|
|
|
This lab is designed to do several things.
The data for this experiment are real. They came from a student named David Merrell, who collected them as part of a high school science project. I was sufficiently interested in what he did to write and ask him for the data, which he sent. I have a web page that will tell you a bit more
Merrell raised three groups of mice. One group (Group 1) was raised in the absence of music, one group (Group 2) heard 10-12 hours of Mozart every day, and a third group (Group 3) heard 10-12 hours of Anthrax every day. They were originally taught to navigate a maze (unspecified) for food, and then were placed back in the maze each week and the time to run the maze was collected. (Actually there were three trials each week, and the data file that I present gives those three times. I have also calculated, and present, the mean and the median running time per week. You should use the medians. (Why do you think I said this?)
These data are available in a file called Anthrax.sav, which is located on Gumby. I'll explain again how to get to Gumby. There are many variables, but the ones that you are interested in are Group, which is near the beginning, and Median1 -- Median4, which come at the end. Can you guess what these variables represent?
I want you to lay out a set of hypotheses about group differences and about the effect of time. I then want you to read in the data and compute the statistics that are needed to test your hypothesis. We have not covered a statistical test for these differences, but that is irrelevant. The differences are very clear without any statistical test.
The first way to look at the data is to tell it that you want to graph a boxplot, that you want a clustered boxplot that is a summary of several variables. You tell it that on the first dialog box that comes up. Then you want to plot median1 -- median4 by group. This will give you a nice pretty picture.
Now calculate the means of each group on each of the relevant variables. You want Statistics/compare means/means.
Plot the data to show the relevant means. (What are these means of?) You may want to calculate the means, create a new data file with the means in it, and then plot the new data file. (The way that I would do it is to select the Graph object from within Word, and use the results you have to modify their graph to give you what you want-- (Insert/Object/Graph). I'll illustrate this in class.
I have also included the weights of each mouse when they were received from the breeding lab and at each of the four weeks of testing (wt0 to wt4). (No, I did not make a typo, although he may have.) You can look at these variables as well, and consider what they have to say about the results. Why would weight be at all relevant?
Write up this study as if you had run it. Where you don't know what Merrell did, simply make it up as if you did. (You don't have to be correct, and there are some things I couldn't tell from his write-up.) Do not use any statistical tests, such as Anova, but make reasonable assumptions that what you think you see on paper would be reflected in the test if you ran it. Plot your graph of times and/or weights using the graphing feature in Word.
So much for a misspent youth.
Last revised: 09/05/01
© David C. Howell (1999)
