Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Let $X_i$, $1\le i\le t$, be $t$ independent random variables with Binomial distribution $B(n,\frac1t)$.

I would like to find the distribution of $X_{Max}=\max_{i=1}^t(X_i)$

Note that this is the specific case where the probability of success is equal to the reciprocal of the number of variables I am taking the maximum over.

I expect that the answer will be too complex to use directly, so an approximation would be good to have. In fact just having an approximate formula for $\mathbb{E}(X_{Max})$ and $\mathbb{Var}(X_{Max})$ would be sufficient, and I am most interested in the case where $n\gg t\gg1$.

share|cite|improve this question

Despite considerable upvoted interest in the question itself, ... this question has remained unanswered for over 4 months.

We are given $X$ ~ Binomial$(n,p)$ with pmf $f(x)$:

Let $(X_1, ..., X_t)$ denote a random sample of size $t$ drawn on $X$. Then, the pmf of the $t$-th order statistic (i.e. the pmf of the sample maximum), denoted say $g(x)$, is given by:

where OrderStat[t, f, samplesize] is a mathStatica function that finds the pmf of the $t$-th order statistic, and where the Hypergeometric2F1 is described here . The domain of support is, of course:

domain[g] = {x, 0, n}

This solution, for general probability parameter $p$, nests the special case the original poster wishes to solve for, namely where $p=\frac{1}{t}$. So simply plug in: $p=\frac{1}{t}$.

Normally, of course, as the sample size (here $t$) increases, the distribution of the sample maximum shifts out to the right. However, in this particular set-up, because $p=\frac{1}{t}$, the opposite happens ... the distribution of the sample maximum shifts to the left as $t$ increases, as the following plot illustrates:

[I have also 'checked' the above solution using Monte Carlo methods: all seems fine. ]

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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