Abstract
Novel necessary and sufficient conditions for the identifiability of the linear mixed model are derived. These conditions either relax or generalize previously reported conditions. The novel conditions are translated to criteria that can be checked for most commonly employed parametrizations of the random effect’s covariance matrix of linear mixed model.
Original language | English |
---|---|
Journal | Communications in Statistics - Theory and Methods |
Early online date | 2023 |
DOIs | |
Publication status | E-pub ahead of print - 2023 |
Keywords
- Injectivity
- non linear parametrization
- normality