In chapter 7 we examined path models as the logical extension of multiple regression models (chapter 6) to show more meaningful theoretical relationships among our observed variables. Thus, the two previous chapters dealt exclusively with models involving observed variables. In this chapter we begin developing models involving factors or latent variables and continue latent variable modeling throughout the remainder of the book. As we see in this chapter, a major limitation of models involving only observed variables is that measurement error is not taken into account. The use of observed variables in statistics assumes that all of the measured variables are perfectly valid and reliable, which is unlikely in many applications. For example, father’s educational level is not a perfect measure of a socioeconomic status factor and amount of exercise per week is not a perfect measure of a tness factor.