A topic in the Open Knowledge Graph — a free, open map of 15,290 topics and the order to learn them in.
The joint distribution of multiple random variables describes their probabilities together. For discrete: P(X=x, Y=y) sums to 1. For continuous: f(x,y) integrates to 1. Joint distributions reveal dependence structure between variables.
Create joint probability tables for simple two-variable scenarios. Integrate joint PDFs over regions. Recognize that knowing joint distributions allows computing all other probabilistic quantities.
Assuming variables are independent without checking. Confusing joint probability with conditional probability. Not recognizing that joint distributions contain complete information about the system.
When you studied probability mass functions (PMFs) and probability density functions (PDFs), you were describing a single random variable in isolation — P(X = x) or f_X(x). A joint distribution extends this to two or more variables simultaneously, capturing not just how each behaves alone, but how they relate to each other. The joint distribution is the complete description of a random system: everything you might want to know about the variables can be derived from it.
For two discrete random variables X and Y, the joint PMF is P(X = x, Y = y) — the probability that X takes value x *and* Y takes value y at the same time. The full collection of these probabilities is often arranged in a table where rows index values of X and columns index values of Y. The entries must sum to 1 over all pairs (x, y). From this table, you can read off everything: P(X = x, Y = 3) sums the column for Y = 3 at row x; P(X ≤ 2) sums all entries where x ≤ 2. For continuous variables, the joint PDF f(x, y) works the same way but sums become integrals: probabilities are volumes under the joint density surface over regions in the xy-plane.
The most important operation on joint distributions is marginalization: recovering the distribution of one variable by summing (or integrating) over all values of the other. The marginal PMF of X is P(X = x) = Σ_y P(X = x, Y = y) — you sum across each row of the joint table. This collapses the two-variable picture back to a one-variable picture. The marginals tell you about each variable separately, but they do not tell you about the relationship between them. Two very different joint distributions can have identical marginals.
Independence is the special case where the joint distribution factors: P(X = x, Y = y) = P(X = x) · P(Y = y) for all pairs, or equivalently f(x, y) = f_X(x) · f_Y(y). Independence means knowing X gives you no information about Y. You should never assume independence without justification — it is a strong claim. For example, a student's score on exam 1 and their score on exam 2 probably have positive dependence (students who do well on one tend to do well on the other); the joint distribution will not factor into the product of the marginals. Joint distributions are the foundation for understanding conditional distributions, covariance, and correlation — concepts that precisely quantify how much and in what direction variables influence each other.