There are lots of misinterpretation with p-value, ref. to the blog of Prof. Gelman.
Especially, I doubt it makes sense in the Bayesian world, but I don’t want to open the “box of Pandora”.
With fancy plots I mean things like Rank histogram plots. Aki did and does a lot of good things.
In my small world, I mostly look if the parameters looks sound, doing boxplots and rethink
my problem, the diagnostics, my data and my priors. Changing my perspective, learn
about the data, the priors, and then improve my model(s). Eg. Having a logistic regression,
but also can use a beta logistics model. Thinking about what is the difference? What is
the multivariate variation (-> Dirichlet). Can it be justified? How that will change the looic diagnostic?
…