Joint distribution of order statistics

Q: What are the standard starting assumptions for calculating the joint distribution of order statistics?

We assume we have $n$ independent and identically distributed (i.i.d.) random variables $X_1, X_2, \dots, X_n$ that are then sorted as $X_{(1)} \le X_{(2)} \le \dots \le X_{(n)}$.

Q: Why is the domain constraint \(x_1?

Because the variables have been sorted, the joint density is only non-zero where the values strictly follow the assigned order.

Q: What is the joint PDF formula for all $n$ order statistics?

$f_{X_{(1)}, \dots, X_{(n)}}(x_1, \dots, x_n) = n! \cdot f(x_1) \cdot f(x_2) \cdot \dots \cdot f(x_n)$ for $x_1 < x_2 < \dots < x_n$.

Q: What does the factor $n!$ represent in the joint PDF of order statistics?

It accounts for all possible permutations (orderings) of the original $n$ variables that are condensed into a single sorted sequence.

Q: What is the general formula for the joint PDF of the $i$-th and $j$-th order statistics?

$f_{X_{(i)}, X_{(j)}}(u, v) = \frac{n!}{(i-1)!(j-i-1)!(n-j)!} [F(u)]^{i-1} f(u) [F(v) - F(u)]^{j-i-1} f(v) [1 - F(v)]^{n-j}$ for $u < v$.

Q: What is the simplified joint PDF for the minimum and maximum order statistics?

$f_{X_{(1)}, X_{(n)}}(u, v) = n(n-1) [F(v) - F(u)]^{n-2} f(u) f(v)$ for $u < v$.

Q: What is the 'sample range' and how is it used in Actuarial Science?

The sample range is $R = X_{(n)} - X_{(1)}$, representing the spread of risk or the difference between the largest and smallest observations.

Q: What is a common pitfall when setting integration limits for joint order statistics?

Forgetting that the limits of the inner integral must depend on the outer variable to satisfy the condition $u < v$.

Exam P – Probability · SOA (Society of Actuaries) · 4 min read

Welcome to the World of Sorted Data!

In our previous studies, we looked at individual Order Statistics—like finding just the minimum or just the maximum of a group of random variables. But what if we want to know how the minimum and the maximum behave together? That’s where the Joint Distribution of Order Statistics comes in.

Think of it like an insurance company looking at 100 claims. They don't just care about the smallest claim or the largest claim in isolation; they might need to know the joint probability that the smallest claim is over \$500 while the largest is under \$10,000. Understanding how these sorted values interact is a key skill for Exam P.

1. The Foundation: Quick Refresh

Before we dive into the joint formulas, let's remind ourselves of the setup:
1. We start with $n$ independent and identically distributed (i.i.d.) random variables: $X_1, X_2, \dots, X_n$.
2. We "sort" them from smallest to largest: $X_{(1)} \le X_{(2)} \le \dots \le X_{(n)}$.
3. $X_{(1)}$ is the minimum, and $X_{(n)}$ is the maximum.

Important Note: Because we are sorting them, the joint density is only defined where the order is maintained (e.g., where $x_1 < x_2 < \dots < x_n$). If you ever calculate a joint density for order statistics and ignore the bounds, you'll likely get the wrong answer!

2. The Joint PDF of ALL Order Statistics

If we want to find the joint probability density function (PDF) for the entire set of sorted variables, the formula is surprisingly simple.

For a set of i.i.d. variables with PDF $f(x)$, the joint PDF of all $n$ order statistics is:
$f_{X_{(1)}, \dots, X_{(n)}}(x_1, \dots, x_n) = n! \cdot f(x_1) \cdot f(x_2) \cdot \dots \cdot f(x_n)$
Provided that $x_1 < x_2 < \dots < x_n$ (and zero otherwise).

Why the $n!$?

Analogy: Imagine you have 3 distinct colored balls (Red, Blue, Green). There are $3! = 6$ different ways (permutations) you could have picked them up. However, once you "sort" them by size, there is only one specific way they can stay sorted. The $n!$ accounts for all the original possible orderings that now collapse into this single sorted sequence.

3. The Joint PDF of Two Order Statistics

On Exam P, you are most likely to be asked about the joint distribution of just two order statistics, usually the minimum $X_{(1)}$ and the maximum $X_{(n)}$.

The general formula for the joint PDF of the $i$-th and $j$-th order statistics (where $i < j$) is:
$f_{X_{(i)}, X_{(j)}}(u, v) = \frac{n!}{(i-1)!(j-i-1)!(n-j)!} [F(u)]^{i-1} [f(u)] [F(v) - F(u)]^{j-i-1} [f(v)] [1 - F(v)]^{n-j}$
For $u < v$.

Don't panic! This looks like a monster, but it's just a Multinomial logic in disguise. Let's break down the special case you'll see 90% of the time.

The "Famous Couple": Minimum and Maximum

For the joint PDF of $X_{(1)}$ and $X_{(n)}$, the formula simplifies significantly:
$f_{X_{(1)}, X_{(n)}}(u, v) = n(n-1) [F(v) - F(u)]^{n-2} f(u) f(v)$
For $u < v$.

Step-by-Step Logic:
1. We need one item to be the minimum ($u$): That's $f(u)$.
2. We need one item to be the maximum ($v$): That's $f(v)$.
3. The other $n-2$ items must fall between them: That's $[F(v) - F(u)]^{n-2}$.
4. The constant $n(n-1)$ comes from picking which of the $n$ variables is the min, and which of the remaining $n-1$ is the max.

Key Takeaway:

When working with these formulas, always identify your $n$, find your $F(x)$ and $f(x)$, and plug them into the $n(n-1)$ template. Double-check your bounds: the PDF is only non-zero when $u < v$.

4. Common Exam Task: Calculating Probabilities

The most common question involves finding a probability like $P(X_{(n)} - X_{(1)} \le k)$. This is often easiest to solve using the joint PDF and double integration.

Example Trace:
Suppose you have 3 i.i.d. variables from a Uniform(0,1) distribution. What is the joint PDF of the min and max?
1. For Uniform(0,1): $f(x) = 1$ and $F(x) = x$.
2. $n = 3$.
3. Plug into the formula: $f_{X_{(1)}, X_{(3)}}(u, v) = 3(3-1) [v - u]^{3-2} (1)(1)$.
4. Result: $6(v - u)$ for $0 < u < v < 1$.

Did you know?
The quantity $X_{(n)} - X_{(1)}$ is called the Sample Range. Actuaries use the range to understand the spread of risk in a portfolio!

5. Avoiding Common Pitfalls

Mistake 1: Forgetting the Range of Integration.
When integrating the joint PDF, remember that $u$ (the minimum) can never be larger than $v$ (the maximum). Your integral limits will often look like $\int_{0}^{1} \int_{0}^{v} \dots du dv$.

Mistake 2: Mixing up PDF and CDF.
In the formula $n(n-1) [F(v) - F(u)]^{n-2} f(u) f(v)$, note that the "middle" part uses the CDF ($F$), while the specific points use the PDF ($f$).

Quick Review Box:
• All $n$ sorted: $n! \prod f(x_i)$
• Min and Max: $n(n-1)[F(max)-F(min)]^{n-2} f(min)f(max)$
• Always check the condition: $x_{(1)} < x_{(2)} < \dots < x_{(n)}$

Summary: Why This Matters

Joint distributions of order statistics allow us to look at the "shape" of our data's extremes. Whether you are calculating the probability that the range of claim sizes is within a certain limit or finding the expected difference between the second-highest and highest value, these formulas are your toolkit. They might look intimidating, but they are just logical extensions of the basic rules of probability and combinations. You’ve got this!

Quick check

Can you answer these now?

Open each question to check the key ideas from this chapter.

What are the standard starting assumptions for calculating the joint distribution of order statistics?

We assume we have $n$ independent and identically distributed (i.i.d.) random variables $X_1, X_2, \dots, X_n$ that are then sorted as $X_{(1)} \le X_{(2)} \le \dots \le X_{(n)}$.

Why is the domain constraint $x_1 < x_2 < \dots < x_n$ critical for joint order statistics?

Because the variables have been sorted, the joint density is only non-zero where the values strictly follow the assigned order.

What is the joint PDF formula for all $n$ order statistics?

$f_{X_{(1)}, \dots, X_{(n)}}(x_1, \dots, x_n) = n! \cdot f(x_1) \cdot f(x_2) \cdot \dots \cdot f(x_n)$ for $x_1 < x_2 < \dots < x_n$.

What does the factor $n!$ represent in the joint PDF of order statistics?

It accounts for all possible permutations (orderings) of the original $n$ variables that are condensed into a single sorted sequence.

What is the general formula for the joint PDF of the $i$-th and $j$-th order statistics?

$f_{X_{(i)}, X_{(j)}}(u, v) = \frac{n!}{(i-1)!(j-i-1)!(n-j)!} [F(u)]^{i-1} f(u) [F(v) - F(u)]^{j-i-1} f(v) [1 - F(v)]^{n-j}$ for $u < v$.

What is the simplified joint PDF for the minimum and maximum order statistics?

$f_{X_{(1)}, X_{(n)}}(u, v) = n(n-1) [F(v) - F(u)]^{n-2} f(u) f(v)$ for $u < v$.

What is the 'sample range' and how is it used in Actuarial Science?

The sample range is $R = X_{(n)} - X_{(1)}$, representing the spread of risk or the difference between the largest and smallest observations.

What is a common pitfall when setting integration limits for joint order statistics?

Forgetting that the limits of the inner integral must depend on the outer variable to satisfy the condition $u < v$.

Ready to test yourself?

Turn these notes into exam-style practice. Get unlimited AI questions on this topic with instant marking and explanations.

Practice This Topic

More Exam P – Probability chapters

* The content provided by thinka is generated by AI and may not always be accurate or up-to-date. Please use it as a supplementary resource and verify with official materials.

Put These Notes into Practice

Reading the notes is a great start. Now practise with unlimited AI-generated questions and get instant feedback. 100,000+ students are already improving their grades.

Start Practising Now View Pricing

Done reading? Test yourself with AI practice questions

Practice This Topic Now