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.
>>>For this homework assignment, you may use either spreadsheets or SAS; you do not need to use both.<<<1. Look at the blacknose dace data from homework 5. Pick one of the watersheds with a large number of samples (Double Pipe Creek, Patuxent River, or Youghiogheny River) and plot three frequency histograms for the data: one for the untransformed data, one for log-transformed data, and one for square-root transformed data. Say which transformation you think makes the data look most normal. (Note: because the data includes zeros, you should use the feature on the histogram web page to add 0.5 to each value before log-transforming.)
2. Perform Bartlett's test on the variances of the 12 watersheds, for untransformed, log-transformed (with 0.5 added), and square-root transformed data. Report the three P-values.
3. Based on the histograms and the Bartlett's test, say whether you think the untransformed, log-transformed or square-root transformed data fits the assumptions of the anova best, and say why you think it is the best. Say whether you think the data meet the assumptions well enough to do the anova, or whether you should do the Kruskal-Wallis test instead.
4. Perform a one-way anova on the log transformed or square-root transformed data (whichever you thought was better; if you thought the untransformed data were best, use either log transformed or square-root transformed for this question). Report the mean squares, degrees of freedom, F-statistic, P-value, and results of the Tukey-Kramer procedure. Plot a graph showing the back-transformed means and the Gabriel comparison intervals.
5. Perform a Kruskal-Wallis test on the data, and compare the result to the result of the anova.
6. Perform all of the steps above on the data you collected for homework 4. Because most of you had only 10 observations per group, you should look at a histogram of the residuals from the group means instead of just looking at a histogram of one group.
Return to the Biological Data Analysis syllabus
Return to John McDonald's home page
This page was last revised August 4, 2009. Its URL is http://udel.edu/~mcdonald/stathw6.html