The second rule makes use of the Beta and its usage as prior
conjugate when inferring of a binomial, as we have seen in
Sec. . The idea is to see the pdf
estimated by JAGS with flat prior as a `rough Beta'
whose parameters can be estimated from the mean and the standard deviation
using Eqs. ()-().
We can then imagine that the pdf of could have
been estimated by a `virtual' Poisson processes whose outcomes
update the parameters of the Beta according to
Eqs. ()-().
The trick consists then in modifying the Beta parameters
according to the simple rules:
where
and
are evaluated
from
and
making use of
Eqs. () and ().
Then the new mean and standard deviation are evaluated from and
(see Sec. ).
For example, in the case of Fig.
we have (with an exaggerated number of digits)
, which could derive from a Beta
having
and
.
If we have a prior somehow peaked around 0.3, e.g.
, it can be parameterized by a Beta with
and . Applying the above rule we get
which yield then
, very similar to
what was obtained by reshaping or re-running JAGS
(
at two decimal digits).