On maximum likelihood estimation for log-linear models with non-ignorable non-response
Non-response boundary solutions occur in log-linear models with non-ignorable non-response. We prove existence and uniqueness of maximum likelihood estimates under weak conditions, and their consistency and asymptotic normality. This contrasts with findings for boundary solutions in log-linear models for sparse tables.
| Year of publication: |
2005
|
|---|---|
| Authors: | Clarke, Paul S. ; Smith, Peter W.F. |
| Published in: |
Statistics & Probability Letters. - Elsevier, ISSN 0167-7152. - Vol. 73.2005, 4, p. 441-448
|
| Publisher: |
Elsevier |
| Keywords: | Boundary solutions Categorical data Consistency Informative non-response Normal approximation |
Saved in:
Saved in favorites
Similar items by person
-
Power of edge exclusion tests for graphical log-linear models
Fátima Salgueiro, M., (2006)
-
Markov chain Monte Carlo exact inference for analysing social networks data
McDonald, John W., (2007)
-
What do Bayesian methods offer population forecasters?
Abel, Guy J., (2010)
- More ...