Use of posterior predictive assessments to evaluate model fit in multilevel logistic regression
Assessing the fit of a model is an important final step in any statistical analysis, but this is not straightforward when complex discrete response models are used. Cross validation and posterior predictions have been suggested as methods to aid model criticism. In this paper a comparison is made be...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Published: |
EDP Sciences
2009
|
| Subjects: | |
| Online Access: | https://eprints.nottingham.ac.uk/1273/ |
| _version_ | 1848790574269726720 |
|---|---|
| author | Green, Martin J. Medley, Graham F. Browne, William J. |
| author_facet | Green, Martin J. Medley, Graham F. Browne, William J. |
| author_sort | Green, Martin J. |
| building | Nottingham Research Data Repository |
| collection | Online Access |
| description | Assessing the fit of a model is an important final step in any statistical analysis, but this is not straightforward when complex discrete response models are used. Cross validation and posterior predictions have been suggested as methods to aid model criticism. In this paper a comparison is made between four methods of model predictive assessment in the context of a three level logistic regression model for clinical mastitis in dairy cattle; cross validation, a prediction using the full posterior predictive distribution and two “mixed” predictive methods that incorporate higher level random effects simulated from the underlying model distribution. Cross validation is considered a gold standard method but is computationally intensive and thus a comparison is made between posterior predictive assessments and cross validation. The analyses revealed that mixed prediction methods produced results close to cross validation whilst the full posterior predictive assessment gave predictions that were over-optimistic (closer to the observed disease rates) compared with cross validation. A mixed prediction method that simulated random effects from both higher levels was best at identifying the outlying level two (farm-year) units of interest. It is concluded that this mixed prediction method, simulating random effects from both higher levels, is straightforward and may be of value in model criticism of multilevel logistic regression, a technique commonly used for animal health data with a hierarchical structure. |
| first_indexed | 2025-11-14T18:14:47Z |
| format | Article |
| id | nottingham-1273 |
| institution | University of Nottingham Malaysia Campus |
| institution_category | Local University |
| last_indexed | 2025-11-14T18:14:47Z |
| publishDate | 2009 |
| publisher | EDP Sciences |
| recordtype | eprints |
| repository_type | Digital Repository |
| spelling | nottingham-12732020-05-04T20:26:49Z https://eprints.nottingham.ac.uk/1273/ Use of posterior predictive assessments to evaluate model fit in multilevel logistic regression Green, Martin J. Medley, Graham F. Browne, William J. Assessing the fit of a model is an important final step in any statistical analysis, but this is not straightforward when complex discrete response models are used. Cross validation and posterior predictions have been suggested as methods to aid model criticism. In this paper a comparison is made between four methods of model predictive assessment in the context of a three level logistic regression model for clinical mastitis in dairy cattle; cross validation, a prediction using the full posterior predictive distribution and two “mixed” predictive methods that incorporate higher level random effects simulated from the underlying model distribution. Cross validation is considered a gold standard method but is computationally intensive and thus a comparison is made between posterior predictive assessments and cross validation. The analyses revealed that mixed prediction methods produced results close to cross validation whilst the full posterior predictive assessment gave predictions that were over-optimistic (closer to the observed disease rates) compared with cross validation. A mixed prediction method that simulated random effects from both higher levels was best at identifying the outlying level two (farm-year) units of interest. It is concluded that this mixed prediction method, simulating random effects from both higher levels, is straightforward and may be of value in model criticism of multilevel logistic regression, a technique commonly used for animal health data with a hierarchical structure. EDP Sciences 2009 Article PeerReviewed Green, Martin J., Medley, Graham F. and Browne, William J. (2009) Use of posterior predictive assessments to evaluate model fit in multilevel logistic regression. Veterinary Research, 40 (4). Article 30. ISSN 0928-4249 model fit posterior predictive assessment mixed predictive assessment cross validation Bayesian multilevel model http://dx.doi.org/10.1051/vetres/2009013 doi:10.1051/vetres/2009013 doi:10.1051/vetres/2009013 |
| spellingShingle | model fit posterior predictive assessment mixed predictive assessment cross validation Bayesian multilevel model Green, Martin J. Medley, Graham F. Browne, William J. Use of posterior predictive assessments to evaluate model fit in multilevel logistic regression |
| title | Use of posterior predictive assessments to evaluate model fit in multilevel logistic regression |
| title_full | Use of posterior predictive assessments to evaluate model fit in multilevel logistic regression |
| title_fullStr | Use of posterior predictive assessments to evaluate model fit in multilevel logistic regression |
| title_full_unstemmed | Use of posterior predictive assessments to evaluate model fit in multilevel logistic regression |
| title_short | Use of posterior predictive assessments to evaluate model fit in multilevel logistic regression |
| title_sort | use of posterior predictive assessments to evaluate model fit in multilevel logistic regression |
| topic | model fit posterior predictive assessment mixed predictive assessment cross validation Bayesian multilevel model |
| url | https://eprints.nottingham.ac.uk/1273/ https://eprints.nottingham.ac.uk/1273/ https://eprints.nottingham.ac.uk/1273/ |