Interesting. I’ve long wondered why there seems to be preference among the experts for separate-models-and-loo-comparison over the mixture approach. How is it that the former skirts the closed-world assumptions of the latter?
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| Compare posterior distribution across different models | 1 | 162 | July 23, 2025 | |
| Identifying relationships between parameters estimated from different data sources | 6 | 683 | August 12, 2022 | |
| Multiple likelihoods with same outcome, shared parameter | 5 | 298 | May 28, 2024 | |
| Use posterior predictive distribution in another model | 5 | 1421 | March 25, 2023 | |
| How to properly compare priors | 3 | 1106 | November 12, 2018 |