Residual Scatterplots

Figure 1. Residuals versus predicted values
Scatterplot with regression standardized residuals on the vertical axis and regression standardized predicted values on the horizontal axis

The standardized residuals are plotted against the standardized predicted values. No patterns should be present if the model fits well. Here you see a U-shape in which both low and high standardized predicted values have positive residuals. Standardized predicted values near 0 tend to have negative residuals.

  1. To produce a scatterplot of the residuals by the predictor Package design, from the menus choose:

    Graphs > Chart Builder...

    Figure 2. Chart Builder
    Chart Builder
  2. In the Chart Builder, select the Scatter/Dot gallery and choose Simple Scatter.
  3. Select Standardized Residual as the y-axis variable and Package design as the x-axis variable.
  4. Click OK.
Figure 3. Residuals versus package design
Scatterplot with standardized residuals on the vertical axis and Package design on the horizontal axis

The U-shape is more pronounced in the plot of the standardized residuals against package. Every residual for Design B* is negative, whereas all but one of the residuals is positive for the other two designs. Because the linear regression model fits one parameter for each variable, the relationship cannot be captured by the standard approach.

Next