The problem: Suppose you have 10 voters and 2 candidates. Each voter has a probability $p$ of voting for a Candidate 1, and a probability $1-p$ of voting for Candidate 2. What is the probability that exactly 6 voters will vote for Candidate 1, if $p$ is described by a Beta(3,3) function?

My thought process: I think I need to look for the marginal probability $P(X=6|n=10)$. I know that the beta function can be integrated as $30\int_0^1 x^2 (1-x)^2 dx$, but I'm not sure how I should go about solving this process. Could someone please get me started on the right path? Thank you!

  • $\begingroup$ Here's a down and dirty method. Beta(3,3) has a mean of .50 A Binomial of 10C6 (.5)^10 = .205 $\endgroup$ – user104266 Feb 15 '17 at 5:31
  • $\begingroup$ Start by fixing $p$ and finding the chance that exactly six voters choose Candidate 1. Then average over $p$. $\endgroup$ – whuber Feb 15 '17 at 23:24

You can frame this as a hierarchical model. In this case, $p$ has a Beta distribution. Then conditional on a particular value of $p$, $v$ - the number of votes for Candidate 1 - has a Binomial distribution with parameters 10 and $p$. This can be written as

$p \sim Beta(3, 3)$

$v | p \sim Binomial(10, p)$

First, find the joint density using the distributions given above. From the definition of conditional probability, we can write the joint as

$f_{v,p}(v,p) = P(v|p)P(p)$

Now integrate out $p$ to get the marginal of $v$

$ f_v(v) = \int P(v|p)P(p) dp$

You can use this marginal distribution to calculate probabilities.

I really like hierarchical models because they let you express complex system in terms of more tractable components. For example, calculating the expected number of votes for candidate 1 is easy in this setting. Using the law of iterated expectation,

$ E[v] = E[E[v|p]] = E[10p] = 10E[p] = 10(.5) = 5$

| cite | improve this answer | |

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.