The core of the method to treat the non-response suvey data, is to make some hypothesis and model the response probability. In this paper, we present a method of how to make factor anlysis when survey data compose of some item-nonresponses. We combine the factor anlysis model and response model, by treating the factors as latent variables which influence item response pobability,and assuming the response pobability has a logit model. After that, we give a two-step method to estimate the parameters. In the end of this paper, real data analysis illustrates that our model and estimation method are applicable. |