Inference for Multiple Linear Regression in JMP®
Chapter 9, Printout 2
- Launch JMP and create a new data table.
- Enter the data.
- Rename Column 1 to "x1" and enter the Air Flow data from the x1i column in Table 4.8 (p. 150).
- Add a second column, "x2," and enter the Cooling Water Inlet Temperature data from the x2i column in the table.
- Add a third column, "x1^2"
- Double-click the column header, "x1^2."
- In the dialog box that follows, select "New Property --> Formula."
- Click "Edit Formula."
- In the dialog box that follows, double-click on the words, "no formula." Enter ":x1^2"
Note: When you finish editing the function, JMP will display it in a graphically-editable form. Functions can also be entered graphically using the "Table Columns" list and the arithmetic buttons. This approach is more intuitive, though more time-consuming.
- Click "OK" to close the function dialog box.
- Click "OK" in the column properties dialog box.
JMP fills the "x1^2" column with the squares of the values in "x1."
- Add a fourth column, "y," and enter the Stack Loss data from the yi column in Table 4.8.
- Compute the multiple linear regression.
JMP computes a fit and displays summary information.
- Select "Analyze --> Fit Model."
- In the "Select Columns" list in the dialog box that follows, click "y"; then click the "Y" button to designate y as the response variable.
- Click "x1" in the "Select Columns" list and click "Add" to designate x1 as a factor variable. Do the same for x2 and x1^2.
- Select "Standard Least Squares" from the pop-up menu labeled "Personality" if it is not already selected. (This should be the default.)
- Select "Minimal Report" from the pop-up menu labeled "Emphasis."
- Leave the checkbox labeled "No Intercept" unchecked.
- Click "Run Model."
- Display confidence limits for parameters.
JMP adds two columns to the Parameter Estimates table.
- Right-click anywhere on the table under the label "Parameter Estimates."
- Select "Columns --> Lower 95%" from the resulting pop-up menu.
- Right-click on the table again, this time selecting "Columns --> Upper 95%" from the pop-up menu.
- Generate a predicted value for x1 = 60, x2 = 20
- Click the red triangle next to the label, "Response y." Select "Save Columns --> Prediction Formula" from the resulting pop-up menu.
- Again, click the red triangle next to the label, "Response y." Select "Save Columns --> StdErr Pred Formula" from the resulting pop-up menu.
JMP saves the formulas to new columns in the data table.
- Add a new data point to the table by entering 60 in row 18 of the "x1" column and 20 to row 18 of the "x2" column.
JMP computes a value for x1^2, a predicted y, and a standard error for the prediction.