Yes, I noticed that. I switched and put prior on std instead. I settled on Half Cauchy. The model does well now, though it still has a good deal of autocorrelation, even with QR. I really wonder where it might come from.
Tommaso11397
21
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| Sampler getting stuck? | 18 | 3918 | May 26, 2019 | |
| Different results using QR parameterization in Hierarchical Model | 2 | 535 | January 25, 2021 | |
| Stan_betareg intercept coefficient unaffected by prior | 15 | 2094 | August 26, 2017 | |
| My first Stan model - hierarchical logistic regression | 28 | 8628 | June 6, 2017 | |
| Standardizing predictors and outputs in a hierarchical model | 18 | 5696 | April 30, 2019 |