Data Analysis with Microsoft Excel: Updated for Office 2007

370 Statistical Methods

Normal Errors and the Normal Plot

What about the assumption of normal errors? Usually, if there is a problem

with non normal errors, extreme values show up in the plot of residuals ver-

sus predicted values. In this example there are no residual values beyond 25

in absolute value, as shown in Figure 9-8.

How large should the residuals be if the errors are normal? You can de-

cide whether these values are reasonable with a normal probability plot.

To make a normal plot of the residuals:

1 Return to the Multiple Regression worksheet.

2 Click Single Variable Charts from the StatPlus menu and then click

Normal P-plots.

3 Click the Data Values button, click the Use Range References option

button, and select the range C29:C109. Click OK.

4 Click the Output button and specify the new chart sheet Residual

P-plot as the output destination. Click OK.

5 Click OK to start generating the plot. See Figure 9-11.

Figure 9-11

Normal

P-plot of

the residuals

The plot is quite well behaved. It is fairly straight, and there are no extreme

values (either in the upper right or lower left corners) at either end. It appears

there is no problem with the normality assumption.