The random variable $X∼Binomial(n= 2,θ)$ and $θ∼Uniform(0,1)$. Calculate the posterior probability $Π(θ <0.5|X= 2)$.


$θ|X∼Beta (1 +x,1 +n−x)$

$θ|X= 2∼Beta (1 + 2,1 + 2−2) = Beta(3,1) $ $i.e. \pi(\theta | X= 2) = 3 \theta^2$

$Π(θ <0.5|X= 2) =\int_{0}^{0.5}3θ^2dθ =1/8$.

Why is $θ|X∼Beta (1 +x,1 +n−x)$? I don't understand how they came up with that.

Also I don't understand how $i.e. \pi(\theta | X= 2) = 3 \theta^2$ for beta$(3, 1)$.

  • $\begingroup$ Seem the prior distribution (unmentioned in your question) may have been $\theta \sim UNIF(0,1) \equiv BETA(1,1).$ Then with data $n = x = 2,$ you'd have posterior dist'n $BETA(3,1).$ Result is from multiplying Prior $\times$ Likelihood to get Posterior according to Bayes' Thm. $\endgroup$ – BruceET Apr 12 at 0:33

Let $\text{Unif}(0,1)$ be a prior distribution on $\theta$, $p(\theta)=1$ on $(0,1)$. Thus, the prior is just a constant (equal 1).

The likelihood for a single data point is the binomial pmf $p(x|\theta)=$ $n \choose x$ $\theta^{x}(1-\theta)^{n-x}$.

Then, the posterior distribution, $p(\theta| \text{data}=x)$ is proportional to a multiple of prior and likelihood:

$$ p(\theta| x) \propto \theta^{x}(1-\theta)^{n-x}. $$

Note that we treat the RHS as a function of $\theta$ and drop the multiplying constants, such as $n \choose x$.

Now, you need to recognize the RHS as a form of a density (of $\theta$) and it is a Beta density with parameters $x+1$ and $n-x+1$ (see the definition to convince yourself). In particular, for $n=2, x=2$, $p(\theta|x)=3\theta^2$.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.