In SPSS, how do I find outliers in my regression?

In regression analysis, sometimes a few outlying observations can have an extra large effect on the estimated coefficients; thus, you may wish to identify and delete these observations from your dataset.

One important measure of influence is called the leverage. SPSS calculates centered leverages which lie between 0 and (n-1)/n, where n is the number of observations. The mean value of this measure of leverage is p/n, where p is the number of independent or explanatory variables. If all the observations have roughly equivalent influence on the estimated value of the coefficients, the leverages would be close to p/n. Thus, you might like to discard the observations for which leverage is greater than 2p/n. You can use the /SAVE LEVER option with the regression command to get the leverage values, for example:

In addition to the regression output in the output window, it will save leverage values as an additional variable in your dataset.

You may also calculate the leverages using the SPSS menus:

1. From the Analyze menu, select Regression, and then Linear.
   
   
2. In the dialog box that appears, click Save.
   
   
3. In the next dialog box that appears, check Leverage values.

This will save leverage values as an additional variable in your dataset.

For more about statistical and mathematical software, email the UITS Stat/Math Center, visit the center's web page, or phone 812-855-4724 (IUB) or 317-278-4740 (IUPUI). The center is located in Bloomington at 410 N. Park Avenue, and is open for consultation by appointment Monday-Friday 9am-5pm.

Comments/Questions/Corrections