Outliers can be established either via standard deviation criter= ia or by user definition. This topic describes how to create user-defined o= utliers.

To create user-defined outliers, follow the steps below:

- Open the Performance Models window.
- Right-click the node of the data set on which linear regression will be=
performed and select
**Analyze Model**. The system displays t= he Analyze Model window with a condition versus age plot of all data points= in the data set. - In the Graph tab, right-click and select
**Edit Outliers**= . The system displays the Edit Outliers dialog box. This dialog box allows = you to specify the criteria that determine whether a data point is an outli= er. You can insert as many records in this dialog box as desired. - In the Edit Outliers dialog box, define the conditions that determine w=
hat data points are outliers. Functions are selected by clicking the down a=
rrow. Values are typed directly in the provided field. To insert additional=
records (criteria), right-click and select
**Insert**. (You c= an also delete records by using the**Delete**command.) When = the dialog box shows the desired criteria, click**OK**. The s= ystem closes the dialog box and colors the data points identified as outlie= rs black. - To perform regression analysis without the identified outliers, click t= he Models tab. The system displays a message saying that the definition of = the outliers has changed and gives you the option of continuing or aborting= .
- Click
**Yes**to continue. In the right pane of the Models tab, locate the record showing 0-Sel= ected Model.

**Note**: The system only performs regression on the model ide= ntified with a Model Run type of 0-Selected Model. If the model so identifi= ed is not the one on which you wish to perform regression analysis, right-c= lick the record that shows the desired model and select=20**Move for Analysis**. This changes the Model Run type of this= model to 0-Selected Model, which allows it to be analyzed.- Right-click the row identified in step 7 and select
**Define Outl= iers**. The system updates the table to show the current amount of o= utliers in the NUM OUTL column. (Note: The value in the NUM SAMPLE column i= s also decremented by the amount of outliers found.) - Right-click the row identified in step 7 and select
**Calculate E= xcluding Outliers**. The system performs the regression analysis on = the data set (minus the outliers) utilizing the current model. Click the Gr= aph tab to view the results of this analysis.