University College London · FACULTY OF STATISTICS

PHAS0083 Chap.4 Joint, Marginal and Conditional Distributions

- one subject, every graph, every model, every mark
6 Chapters4-page Bible
Our own words - no uploaded lecturer files
Updated for this semester
Chapter 4 of 9 · PHAS0083

Joint, Marginal and Conditional Distributions

Joint, Marginal and Conditional Distributions extends single-variable probability to random vectors. A joint pmf or pdf assigns probability to regions of the plane, and the chapter teaches the two operations that answer most questions: a marginal is a projection, found by summing or integrating out the other variable, and a conditional is a slice of the joint divided by the marginal of the variable held fixed.

Independence means the joint factorises into its marginals, so every slice has the same shape; a single point that breaks the product rule proves dependence. Joint distributions are built in three ways: from independent pieces, from a chain of conditionals as in a hierarchical model, and by transformation of variables.

For a discrete transformation the pmf is summed over preimages; for a continuous one-to-one map the density is multiplied by the absolute Jacobian of the inverse map. Adding an auxiliary variable then integrating it out gives the distribution of a single function such as a sum, and these steps produce results with physical meaning, including the thinning and additivity of Poisson counts.

In this chapter

What this chapter covers

  • 01

    Random vectors and joint pmf or pdf

  • 02

    Marginal distributions as projections

  • 03

    Conditional distributions as slices

  • 04

    Independence and the product rule

  • 05

    Joint distributions of hierarchical models

  • 06

    Change of variables with a Jacobian

Worked example · free

Test independence for a joint density on the unit square

Q [5 marks]. X and Y have joint pdf f(x, y) = x + y on the unit square. Find the marginal of X, the probability that both are below one half, and decide whether X and Y are independent. The marks shown are AskSia practice weighting, not a published UCL marking scheme.
For f(x, y) = 6x on 0 < x < y < 1, the marginal of X integrates along the vertical strip from y = x to 1, while the conditional of X given y = 0.7 lives on the For f(x, y) = 6x on 0 < x < y < 1, the marginal of X integrates along the vertical strip from y = x to 1, while the conditional of X given y = 0.7 lives on the horizontal slice.00.5100.50.71xysupport: x < yorange strip:y from x to 1gives the marginal of Xgreen slice at y = 0.7:x runs from 0 to 0.7
For f(x, y) = 6x on 0 < x < y < 1, the marginal of X integrates along the vertical strip from y = x to 1, while the conditional of X given y = 0.7 lives on the horizontal slice.
  • 2Marginal: integrate over y from 0 to 1 to get the marginal pdf of X, x + 1/2, and by symmetry the marginal pdf of Y, y + 1/2. This allocation is practice weighting, not a UCL marking scheme.
  • 2Probability: integrate x + y over the square of side one half, giving 1/16 + 1/16 = 1/8.
  • 1Independence: the product (x + 1/2)(y + 1/2) differs from x + y, for example at (0, 0), so the variables are dependent.
The marginal is x + 1/2, the probability is 0.125, and the variables are dependent even though the support is a rectangle. A rectangular support is necessary for independence but not sufficient. The practice marks are not an official UCL allocation.
Sia tip — To prove dependence, exhibit one point where the joint differs from the product of marginals; to prove independence, show the factorisation everywhere.
Glossary

Key terms

Joint density
A function whose integral over a region gives the probability that a random vector falls in it.
Conditional density
The joint density at a fixed value of one variable divided by that variable's marginal density.
Jacobian determinant
The determinant of partial derivatives of the inverse map, which rescales a density under a change of variables.
FAQ

Joint, Marginal and Conditional Distributions FAQ

What is the quickest way to find the distribution of a sum?

Introduce an auxiliary variable so the map is one-to-one, for instance U equal to X plus Y and V equal to Y. Transform the joint density with the Jacobian of the inverse map, then integrate out V. For discrete variables the same idea becomes a sum over all pairs that give each value of the total.

How do I find the limits when integrating a joint pdf?

Sketch the support first. For a marginal of X, fix x and read off the range of y inside the support at that x; those are the inner limits. Triangular supports such as x less than y are the usual source of wrong limits.

Does a rectangular support mean the variables are independent?

No. A rectangular support is required, because otherwise the range of one variable depends on the other, but the density must also factorise into a function of x times a function of y. The density x plus y on the unit square is a counterexample.

Why is the Jacobian taken of the inverse map?

The density of the new variables is the old density evaluated at the preimage point, multiplied by how much a small area shrinks or stretches when mapped back. That factor is the absolute determinant of the derivatives of the inverse functions with respect to the new variables.

Study strategy

Exam move

Treat every joint-distribution problem as a geometry problem first: draw the support, shade the region asked about, and only then write integrals with limits taken from the picture. Practise marginals and conditionals on one discrete table and two continuous densities, one on a rectangle and one on a triangle, and verify that each marginal integrates to one.

For independence, train both directions: a factorisation proof and a single-point disproof. Then work the three construction routes in turn: a product of independent pieces, a chain of conditionals for a latent-variable model, and a transformation where you must compute the inverse map and its Jacobian.

Finish with the auxiliary-variable trick for the distribution of a sum, and check the result by simulating a few thousand pairs and comparing the empirical cdf of the sum with your formula.

Working through Joint, Marginal and Conditional Distributions in PHAS0083? Sia is AskSia’s AI Statistics tutor — ask any PHAS0083 Joint, Marginal and Conditional Distributions question and get a clear, step-by-step explanation grounded in how PHAS0083 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.

A+Everything unlocked
Unlocks this Bible + all 32 of your subjects at University College London, plus 1,000+ more Bibles across the AskSia library.
Sia - your PHAS0083 tutor, unlimited, worked the way the exam marks it
The full 4-page Bible + practice bank with worked solutions
Chrome extension - sync your LMS so Sia knows your deadlines
Bilingual EN / Chinese on every Bible and every Sia answer
$0.99 Trial
30-day money-back · cancel in one tap · how it works