To test one or more custom hypotheses involving any model parameters, select Custom Test from the Estimates menu. In this window, you can specify one or more linear functions, or contrasts, of the model parameters.
The results include individual tests for each contrast and a joint test for all contrasts. See Figure 2.26. The report for the individual contrasts gives the estimated value of the specified linear function of the parameters and its standard error. A t ratio, its p-value, and the associated sum of squares are also provided. Below the individual contrast results, the joint test for all contrasts gives the sum of squares, the numerator degrees of freedom, the F ratio, and its p-value.
Click the Done button to perform the tests. The report changes to show the test statistic value, the standard error, and other statistics for each test column. The joint F test for all columns is given in a box at the bottom of the report.
Provides a power analysis for the joint test. This option is available only after the test has been conducted. For details, see Parameter Power.
Figure 2.26 shows an example of the specification window with three contrasts, using the Cholesterol.jmp sample data table. Note that the constant is set to zero for all three tests. The report for these tests is shown in Figure 2.27.
The Cholesterol.jmp sample data table gives repeated measures on 20 patients at six time periods. Four treatment groups are studied. Typically, this data should be properly analyzed using all repeated measures as responses. This example considers only the response for June PM.
1.
|
2.
|
Select Analyze > Fit Model.
|
3.
|
4.
|
5.
|
Click Run.
|
6.
|
7.
|
In the Custom Test specification window, click Add Column twice to create three columns.
|
8.
|
To see how to obtain these values, particularly those in the third column, see Interpretation of Parameters in Statistical Details.
9.
|
Click Done.
|
The results shown in Figure 2.27 indicate that all three hypotheses are individually, as well as jointly, significant.