Reference Posterior Definition

[nattube]

Dear stan-dev team,

Thank you for your work on posteriordb!
So far I really enjoy working with it, but looking into the definition for the reference posterior, I got slightly confused.
Point 4 in the paper and in this repository states:

4. All Expected Fraction of Missing Information (E-FMI) is below 0.2

I never came across this measurement so I looked up the referenced source. It (https://arxiv.org/pdf/1701.02434) does talk about E-BFMI and that values below 0.3 have empirically been shown problematic. I further checked a couple of reference posteriors and all of their e-fmi info fields were strictly above that value. Is this a wrong formulation on your side, or is the measurement that this rule talks about calculated differently than the E-BFMI in the reference? I was unfortunately unable to find anything in relation to why it might be good to have an E-FMI below 0.2, just the opposite.

Thank you in advance and I hope you could clarify this for me,
Nathanael

[MansMeg]

Hi! Yes. This is actually an error. It's on my to-do list to update this/correct this. It's also incorrect in our paper. =)

[nattube]

Thank you very much for the incredibly quick reply 🙏