Events and Meetings of Italian Statistical Society, Advances in Latent Variables - Methods, Models and Applications

Font Size: 
Measurement error modeling in regression models with IRT measures as covariates
Michela Battauz, Ruggero Bellio

Last modified: 2013-06-14

Abstract


This contribution summarizes some techniques to adjust for the effect of the measurement error introduced when a latent variable measured by means of an IRT model is inserted as a covariate in a regression model. Two general classes of methods are available for the adjustment, which can take place following either a structural or a functional approach. Advantages and limitations of both these approaches are presented, with some suggestions for practical applications. An example with empirical data having a multilevel structure is provided.

Full Text: PDF