0% found this document useful (0 votes)
8 views9 pages

Some Part of Bayesian

Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
8 views9 pages

Some Part of Bayesian

Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 9

Example 4.

2 Binomial

Example 4.3 Poisson


Example 4.4 Normal
Different types of prior:

(i)Conjugate Prior:

(ii) Uninformative Prior Distributions:

An uninformative prior is one in which little new explanatory power about the
unknown parameter is provided by intention. Despite the unfortunate name,
uninformative priors are very useful from the perspective of traditional
Bayesianism.
(a)Uniform Prior:

An obvious candidate for the uninformative prior is the uniform distribution.


Uniform priors are particularly easy to specify in the case of a parameter with
bounded support. For instance, a uniform prior for the probability parameter in a
Bernoulli, binomial, or negative binomial model can be specified by:
𝑝(θ) = 1, 0 ≤ θ ≤ 1, or if there is some reason to specify a non-normalized
uniform: : 𝑝(θ) = 1, 0 ≤ θ ≤ 𝑘. The second is nonnormalized because it
does not integrate to one, but this provides no problem in Bayesian analysis. Both
of these forms are referred to as proper since they integrate to a finite quantity.

It is also possible to specify improper uniform priors. Improper priors are tbhose
that sum or integrate to infinity. For example, Sometimes we assume prior
distributions which are constant over the whole real line. For example,
π(θ) = 𝑘, 𝑘 > 0, − ∞ < θ < ∞.

This pdf violates the above condition. This would be called an improper prior
distribution. Such prior distributions can be assumed only if the resulting posterior

distribution is proper, i.e. ∫ π(𝑥)𝑑θ < ∞.
−∞
4.4.2 Jeffreys Prior
Jeffreys (1961, p.181) addresses the problems associated with uniform priors by
suggesting a prior that is invariant under transformation.
1 1
which is Beta ( 2 , 2
) distribution.

Example: Let 𝑋1, …, 𝑋𝑛 ∼Poisson(λ). Find Jeffreys prior for λ.


4.5 Informative Prior Distributions

Where do informative priors come from? Generally there is an abundance of


previous work in the social and behavioral sciences that can guide the researcher,
including her own. So generally, these priors are derived from:

_ previous studies, published work,


_ researcher intuition,
_ interviewing substantive experts,
_ convenience through conjugacy,
_ nonparametrics and other data derived sources.

You might also like