Not really sure where to post this, but here it goes:
On https://docs.pymc.io/notebooks/multilevel_modeling.html
After cell 24, and having obtained an estimate for a the floor
coefficient, it is stated that:
The estimate for the floor coefficient is approximately -0.66, which can be interpreted as houses without basements having about half (exp(−0.66)=0.52) the radon levels of those with basements, after accounting for county.
I think this interpretation is mistaken. The relationship between radon levels and floor
is assumed to be linear.
y_hat = a[county] + b * floor_measure
What’s expontential is the probability of the radon level conditioned on a certain value of floor
, but as floor
varies, the expectation of radon levels varies linearly.
Either that or I’m terribly confused, which is always a possibility