الفهرس 
A latent class model for multivariate data subject to missingness
AbstractOnly 14 pages are availabe for public view
 |  | from | 112 |  |  |
| | | from | 112 | | |
Abstract
In social sciences, such as educational testing and psychometrics, interest is often in measuring constructs or concepts, such as attitudes, behavior or abilities, which cannot be directly measured. These are referred to as latent (unobserved) factors or variables and can be measured through a number of manifest (observed) variables or items. These manifest variables may be subject to missingness. The observed items and the latent variables are linked together by statistical latent variable models. Both manifest and latent variables can be either categorical or continuous resulting in different latent variable models. The approach proposed in this thesis uses latent variable models to capture a latent phenomenon, while incorporating a missingness mechanism to account for possibly nonrandom forms of missingness. In this research, we consider models where both observed items and latent variables are categorical because such variables are often met in social studies, resulting in latent class models