You must type this and all other homework assignments. Do not e-mail the assignment to me; turn it in early (at 322 Wolf) for a foreseeable absence, or turn it in late after an unexpected absence from class.
1. Look at the balance data set you used for the last homework. Pick one of the nominal variables with more than two values ("most disliked class" or "shoes worn during balance test") and one of the measurement variables (letters in first name, letters i last name, age, height, shoe size, or balance time). Sort the data by the nominal variable. Calculate the mean, standard deviation, and sample size for the measurement variable for each category of the nominal variable. For example, if you picked "most disliked class" and "letters in last name," you'd calculate the mean, standard deviation, and sample size of "letters in last name" for everyone who said "algebra," then you'd do the calculation for "calculus" people, etc.
2. Look at the standard deviations you found in question 1. What's your opinion, are the data homoscedastic enough to do a regular one-way anova, or should you do Welch's anova?
3. Pick the group with the largest sample size, and print a histogram of the measurement variable for that group. What's your opinion, are the data close enough to normal that you don't need worry, or should you try a data transformation?
4. Whether or not you though the data needed a transformation, look at the histograms for both the log and square-root transformations, and print whichever looks best. If you said in question 3 that your data needed transformation, did one of these transformations help?
5. Perform a one-way anova on your untransformed data set, using the spreadsheet linked on the one-way anova page. Report the F-statistic, the degrees of freedom, and the P-value.
6. Transform your dat using either the log or square-root transformation. If you said in question 4 that the square-root transformation looked the best, use it; otherwise, use the log transformation. Report the F-statistic, the degrees of freedom, and the P-value.
Optional SAS question. From now on, each homework assignment will have a SAS question. These are completely optional, only do them if you want to teach yourself SAS. You will not get extra credit, and there will not be SAS questions on the exams.
7. Follow the instructions for Mac or Windows to connect to the mainframe computer, Strauss, that has the SAS program on it. Don't spend too much time beating your head against the wall, it may be that the UD IT department has done such a bad job of making it easy to connect to the mainframe that you should give up. If you do get into Strauss, look at the introduction to SAS for basic instructions on writing a SAS program.
If you do get connected, look at the sample program on the one-way anova page, then modify that program to do a one-way anova on your untransformed data. Report the F-statistic, the degrees of freedom, and the P-value. Also print the SAS program (your .sas file) and the .lst file. If you didn't get a .lst file, or if the P-value you got was different from the spreadsheet value, print the .log file as well. Get an early start on this so you can e-mail me if you have problems.
Return to the Biological Data Analysis syllabus
Return to John McDonald's home page
This page was last revised August 21, 2016. Its URL is http://udel.edu/~mcdonald/stathw5.html