Linear models are a powerful and useful set of methods in a large number of settings. Very briefly, there is some outcome measurement that is very important to us and we want to explain variations in its values in terms of other measurements in the data. The heights of several trees can be explained in terms of the trees' ages, for example. It is not a straight line relationship, of course, but knowledge of a tree's age offers us a large amount of explanatory value. We might also want to take into account the effects of measurements on the amount of light, water, nutrients, and weather conditions experienced by each tree. Some of these measurements will have greater explanatory value than others and we may want to quantify the relative usefulness of these different measures. Even after we are given all of this information, some trees will appear to thrive and others will remain stunted, when all are subjected to identical conditions. This variability is the whole reason for statistics existing as a scientific discipline. We usually try to avoid the use of the word “prediction” because this assumes that there is a cause-and-effect relationship. A tree's age does not directly cause it to grow, for example, but rather, a cumulative process associated with many environmental factors results in increasing height and continued survival. The best estimate we can make is a statement about the behavior of the average tree under identical conditions.