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

Font Size: 
The role of fixed-effects approaches in latent variable modeling
Anders Skrondal

Last modified: 2013-06-17

Abstract


Individual-specific effects are usually treated as random in latent variable modeling. In this talk we consider the merits of an alternative approach where individual-specific effects are treated as fixed. In particular, we consider how fixed-effects approaches can address two major challenges in applied statistics: unobserved confounding and data not missing at random.