For example, if you were trying to predict the birth weight of a baby from the weights of its parents, you might look at the last thousand babies born in a certain hospital and note their weights, and then note their parents’ weights, and you might find that if you divide the mother’s weight by 21.7 and the father’s weight by 43.4 and then average those two numbers, the result would correspond to the baby’s birth weight. There would be no significance in the number 21.7 or the number 43.4; the weights wouldn’t connect to any law of nature; they’d be statistical accidents. Regression equations
...more
