chapter  3
Simple Models: Models of Error and Sampling Distributions
Pages 15

In this chapter, our focus is on the error term εi, which should be included in the full statement of our models. For the simple model, the complete statement is:

MODEL: Yi = β0 + εi

This model says that were it not for random perturbations represented by the εi, all the Yi would equal β0 exactly. What are these random perturbations? They are anything that causes the observed datum Yi to deviate from the true value β0. For the internet access example, the delivery or installation of equipment may have been unusually slow in some states; other states may have had either unusually good or unusually poor record-keeping procedures; still others may have had unusually high or low numbers of individuals whose work involved internet use. Any of these or countless other factors may have caused the observed percentages of households with internet access to deviate from the single value β0 predicted by the simple model.