Joint Probability Distribution — Definition, Formula & Example
A joint probability distribution is a single table or function that gives the probability (or, for continuous variables, the probability density) of every combination of values that two or more random variables can take together. It shows how the variables behave jointly, not just one at a time.
For discrete random variables and , the joint probability mass function is , where and . For jointly continuous and , a joint density satisfies for every region , and its integral over the whole plane equals 1.
Key Formula
Where:
- = Joint probability that X = x and Y = y at the same time
- = Marginal probability that X = x, found by summing a row of the joint table
- = Marginal probability that Y = y, found by summing a column of the joint table
Worked Example
Problem: At a bakery, let if a customer buys a drink and if not, and let be the number of muffins bought (0, 1, or 2). The joint distribution is: , , , , , . Verify that the table is valid, find the marginal distribution of , and decide whether and are independent.
Step 1: Every entry is between 0 and 1. Check that the six entries sum to 1.
Step 2: Sum each column over both values of to get the marginal of . Summing rows the same way gives and .
Step 3: Independence requires for every cell. One failing cell is enough to rule it out.
Answer: The table is a valid joint distribution. The marginal of is , , . Because , and are not independent.
Why It Matters
In a college probability or statistics course, you compute , covariance, and correlation directly from a joint distribution. The same idea powers regression and the probabilistic models used in machine learning, where many variables are modeled together.
Common Mistakes
Mistake: Rebuilding the joint table by multiplying the two marginal distributions.
Correction: Many different joint distributions share the same marginals. The product rule gives the joint only when the variables are independent. In the bakery example it predicts for , but the true value is .
