ABSTRACT
Data measured with errors occur frequently in many scientific fields. Standard regression models assume that the independent variables have been measured exactly, in other words, observed without error. Those models account only for errors in the response variable. This chapter focuses on the Bayesian regression models with errorsin-variables using integrated nested Laplace approximations (INLA). The authors’ discussion is based on two types of measurement error: classical measurement error and Berkson measurement error. A fundamental issue in specifying a measurement error model is whether an assumption is made on the distribution of the observed values given the true values or vice versa. Ignoring measurement error can result in biased estimates and lead to erroneous conclusions in data analysis. The chapter discusses the Bayesian analysis of a generalized linear mixed model with errors-in-variables using INLA. The prior settings are defined in the different entries of the list hyper.
