Bayesian ANCOVA - Single Model Inference table - Specified Model Terms?
I am struggling to understand what the coefficients in the Single Model Inference table represent. My intuition was that once I see the Model Comparison table, I can select the model (terms) that shows the "best" performance, and then in the Single Model Inference, select these terms under the Specified Model Terms, and then the coefficient estimates and associated CIs are for this single "best" model.
But, when I do this, the other coefficients from terms I excluded still appear and have estimates. What do these mean in this scenario? Shouldn't they be 0-ed out if I didn't select them?
e.g.,
Table for all models run
Then I select Single Model Inference with only c_PAQ(ALEXI) and c_RME-TOTAL
This, i believe should just give me the "best" model inference from the Model comparison table. But below you see coefficient estimates for the terms I omitted.
What would this table then be telling me exactly? Thank you in advance.
-Mircea
Comments
That looks more like a bug, strange. I'll bring this to the attention of your expert.
EJ
Hi Mircea,
That indeed looks like a bug, thanks for bringing it to our attention!
I will keep you posted when it is resolved.
Don
This should be fixed in the upcoming release, a pull request was just merged that corrected this bug.
Excellent! Thank you for the fast resolution!*
*(there goes my break in writing the manuscript :P)