Figure 11. Diagram showing how total variation (sums of squares) in our response variable is partitioned between what can be explained by our explanatory variable and what is left unexplained.

The variation we explain can be added to the variation left unexplained to calculate total variation. This is an illustrative example and not meant to imply that exactly 50% of the variation in coffee consumption has been explained, we will only know this once we obtain the results of our GLM.