Trough urine samples were analyzed for sodium content for each of two collection periods, one before and one after administration of Mercuhydrin, for each of 30 dogs. The experimenter used 7 dogs as a control group for the study; these dogs were not administered the drug, but their urine samples were collected for two similar time periods.

a. Use ANACOVA (computer results are given next), with the first ("before") collection period measure of sodium content as the covariate, to find the adjusted mean sodium contents for the experimental and control groups for the second ("after") collection period.
b. Test whether the two adjusted means differ significantly.
c. What alternative testing approach (involving a t test) could be used for these data? Carry out this test.
d. Are the two testing methods (t test versus covariance analysis) equivalent in this problem? Explain by comparing regression models appropriate for each method.
e. What do the results of a lack-of-fit test, based on using the computer output given, indicate about the appropriateness of the covariance model used in parts (a) and (b)? The lack of fit test compares the value 6.3568 (= 21.0913/3.3179) to F30, 4, 0.95; see Chapter 15 for further discussion of lack-of-fit tests.