JMP activity: read in the data from the vitamin supplement ANCOVA
example discussed in class.  Fit the ANCOVA model.  Find the adjusted
means and partial F-test for diet effect in the output.

0) Save the file "vitsupp.txt" which is already in a convenient format for import into JMP.
1) open JMP
2) read the "Tip of the Day", they're usually good
3) Open Data Table, by clicking 
File
   Open
then browsing to the appropriate folder. 

In PROC GLM, it was important to declare DIET as a class variable  with a
CLASS statement.  To do the equivalent in JMP, 

Either double-click or right-click the column header for diet and select
Column Info.   Change the ``Modeling Type" to "Nominal".  Click OK.

To fit the ANCOVA model, click
Analyze
   Fit Model.  
      Select/make dark the "Y" variable on the left and click the Y button.
      Select/make dark the "diet" variable on the left and click the Add button,
      underneath "Construct Model Effects"
      Select/make dark the "Z" variable on the left and click the Add button.
         Click "Run Model"
            Find appropriate output.
               Note that nonsense output would have been generated if we
               hadn't made the change to the diet variable.
                  *For a nifty  plot of the adjusted means for  the four diets,
                  Click the small right triangle next to the "Diet" Window and
                  select "LSMeans Plot"
                  *Also, consider the contrast which compares diet 4 with the
                  average of the other three.  Click the little right triangle
                  again and select "LSMeans Contrast" then click "+" for diets
                  1,2 and 3 and "-" for diets 4.  JMP seems to normalize the
                  coefficients in a natural way.  Click on the blue triangle
                  next to "Test Detail" to see more output.

(In SAS you'd accomplish this with the following code:
PROC GLM;
   CLASS diet;
   MODEL y=diet z;
   ESTIMATE "diet 4 - mean(diet1,diet2,diet3)" diet 1 1 1 -3/divisor=3
RUN;
           Check out the parameter estimates.  Note that the parameterization
           used by default in JMP is different from that used by PROC GLM.
           JMP uses the factorial effects parameterization (mu + tau_i) 
           discussed in class.

)