LOOKING FOR OUTLIERS


If you have a large number of cases, you may not want to look at the values of the residuals for all of them. Instead you may want to look only at the cases with "large" residuals. Such cases are called outliers . This is easy to do with the REGRESSION procedure. Just leave off the keyword ALL on the CASEWISE subcommand. In most computer software applications, you will find that if you do not tell the program to print all cases, it prints only those whose standardized residuals are greater than 3 or less than -3. By looking at the characteristics of the outliers, you can see situations where the model does not work well. Typical outliers will appear with large residuals in the form of Figure 9.5. The effect on the actual regression line is shown in Figure 9.6 and finally the outliers outside the pattern are shown in Figure 9.7.

click to expand
Figure 9.5: Outlier with large residual .
click to expand
Figure 9.6: Outlier that tilts the regression line.
click to expand
Figure 9.7: Outliers outside pattern of explanatory variables .



Six Sigma and Beyond. Statistics and Probability
Six Sigma and Beyond: Statistics and Probability, Volume III
ISBN: 1574443127
EAN: 2147483647
Year: 2003
Pages: 252

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net