14.5 How Good Is the Regression Line?

So far, we have discussed several methods to evaluate the fitted model. These included the scatterplot, the significance level of the slope, and the RMSE relative to the standard deviation around the overall mean of Y, called Std Dev[Mean] in the Fit Mean option. In this section, we discuss the very popular coefficient of determination, R², and show how to use residual plots to verify some assumptions of simple regression analysis.

14.5.1 Scatterplot

A scatterplot provides a visual measure of the goodness of the regression line. The closer the observations fit the line, the better the fit. Scatterplots are also useful for identifying alternative models, such as a curved line, or in identifying possible outliers.