ABSTRACT
This chapter deals with modelling measurement error in both the dependent and independent variables of a structural multilevel model. The observed data consists of responses to questionnaires or tests and contain measurement error. It shows that measurement error in both dependent and independent variables leads to attenuated parameter estimates of the structural multilevel model. Therefore, the response error in the observed variables modelled by an item response model and a classical true score model to correct for attenuation. The Gibbs sampler used to estimate all parameters of both the measurement model and the structural multilevel model at once. The chapter divided into several sections. The first section describes a substantive example in which model can apply. Then, several different measurement error models for response error discussed. After describing combination of the structural model with different measurement error models, fitting these models discussed. Finally, it shows that the regression parameters estimates of the structural multilevel model attenuated when measurement error ignored.
