Bayes factor favours reduced model without the predictor of interest: how do to explain?

Modeling
, ,
2

previous discussion on this topic here: Proper interpretation of Bayes factor (from rstanarm + bridgesampling). hope this helps