Chapter 7 • Modeling Relationships of Multiple Variables with Linear Regression 177
Figure 7.9
Data Points, a
Regression Line and Residuals (dashed lines)
Constant Variation Sometimes the regression line
is not well suited to the data, as in Figure 7.10, which shows a fan effect This regression line is much better at predicting low values of the dependent variable than it is at predicting high values. It is therefore
not an appropriate model, regardless of the strength of the coefficients or their significance values. Ina
well-fitting regression model, the variation around the line is constant all along the line (Figure 7.5 is a good example.
30 20 10 0
120 100 80 60 40 20 0
Figure 7.10
Scatter Plot of a Fan Effect Chapter 7 • Modeling Relationships of Multiple Variables with Linear Regression 178
Share with your friends: 
