Law of the Unconscious Statistician (LOTUS)

The Law of the Unconscious Statistician (LOTUS) is the rule for finding the expected value of a function of a random variable, E[g(X)], without first finding the distribution of g(X). In Intro to Probability, it is a shortcut for continuous random variables.

What is the Law of the Unconscious Statistician (LOTUS)?

In Intro to Probability, the Law of the Unconscious Statistician, or LOTUS, is the shortcut you use when you want the expected value of a transformed random variable. Instead of finding the full distribution of g(X), you plug the function into an integral with the original random variable’s PDF: E[g(X)] = integral of g(x)f_X(x) dx over all x.

That means LOTUS is not a separate kind of expected value. It is the same expectation idea you already know, but written in a way that lets you work directly from X. If X is continuous and you want the average value of X squared, absolute value of X, or some other nonlinear expression, LOTUS lets you set up the calculation right away.

The main trick is that the function goes inside the integral before you do any probability transformation. You do not need to build a new PDF for g(X) unless the problem specifically asks for it. That is why LOTUS feels so efficient in continuous settings, where finding the distribution of a transformed variable can be messy.

A simple example is finding E[X^2]. If X has PDF f_X(x), LOTUS says to compute integral x^2 f_X(x) dx. You are averaging the squared values of X, weighted by how likely each x is. The same idea works for E[sin(X)] or E[(X - 3)^2].

This rule also ties directly to variance. Since variance uses E[X^2] and (E[X])^2, LOTUS often shows up when you need the second moment of a continuous random variable. A common mistake is to average the function values without weighting by the PDF, but in probability, the density is what tells you which x-values count more.

Why the Law of the Unconscious Statistician (LOTUS) matters in Intro to Probability

LOTUS matters because it gives you a practical way to compute expectations for transformed continuous random variables, which shows up all over Intro to Probability. If the problem asks for an average payoff, a mean squared error, or a risk measure, you are often really being asked for E[g(X)], not just E[X].

It also keeps you from doing unnecessary work. Finding the distribution of g(X) can take algebra, case splitting, and a separate PDF. LOTUS skips that step and uses the distribution you already know for X. That makes it a standard move in expected value and variance problems, especially when the transformation is nonlinear.

This is one of the first places where the course shifts from “what values can X take?” to “what happens after you process X through a formula?” That shift matters later when you see variance, expected utility, and other summary measures built from transformed random variables.

How the Law of the Unconscious Statistician (LOTUS) connects across the course

Expected Value

LOTUS is really an expected value rule, just written for a function of X instead of X itself. If you already know how to compute E[X] for a continuous random variable, LOTUS extends that same idea to E[g(X)]. The core habit is unchanged: average outcomes by weighting them with the PDF.

Continuous Random Variable

LOTUS is most often used with continuous random variables because the expectation is written as an integral. You are not summing over separate outcomes the way you would in a discrete problem. That means the shape of the PDF matters, and the function g(x) gets integrated against that density.

Probability Density Function (PDF)

The PDF is the weighting function inside LOTUS. Without f_X(x), you do not know how likely each x-value is, so you cannot compute the weighted average of g(X). A lot of LOTUS mistakes come from forgetting that the PDF belongs in the integral from start to finish.

Linearity of Expectation

Linearity of expectation is a separate property that lets you split sums inside expectation, while LOTUS tells you how to evaluate the expectation in the first place. Sometimes you use both, for example when expanding a squared term before integrating. They work together, but they are not the same rule.

Is the Law of the Unconscious Statistician (LOTUS) on the Intro to Probability exam?

A problem set or quiz question will usually give you a PDF and ask for E[g(X)] or a quantity built from it, like E[X^2], E[|X|], or Var(X). Your job is to write the integral with g(x) multiplied by f_X(x), then compute it carefully over the support of X. If the question is about variance, you often use LOTUS twice, once for E[X] and once for E[X^2].

A common trap is trying to find the distribution of g(X) first, when the faster path is already sitting there in the PDF. Another trap is forgetting the limits of integration or using the wrong density. If the class asks for interpretation, say what the expectation means in context, like an average cost, average loss, or long-run payoff.

Key things to remember about the Law of the Unconscious Statistician (LOTUS)

• LOTUS is the shortcut for finding the expected value of a function of a random variable, written as E[g(X)].

• In the continuous case, you compute it with an integral of g(x) times the PDF of X over the support of X.

• You do not need to find the distribution of g(X) first unless the problem specifically asks for it.

• LOTUS shows up a lot in variance problems because variance depends on E[X^2] and E[X].

• The big mistake is forgetting that the PDF does the weighting, not just averaging the function values by themselves.

Frequently asked questions about the Law of the Unconscious Statistician (LOTUS)

What is the Law of the Unconscious Statistician (LOTUS) in Intro to Probability?

LOTUS is the rule for computing E[g(X)] directly from the distribution of X. For a continuous random variable, you integrate g(x)f_X(x) over the support of X. It saves you from having to find the distribution of the transformed variable first.

How do you use LOTUS for a continuous random variable?

First identify the function g(X) you want to average. Then multiply g(x) by the PDF f_X(x) and integrate over the valid x-values. The result is the expected value of the transformation, like E[X^2] or E[|X|].

Is LOTUS the same as expected value?

Not exactly, but it is built from the same idea. Expected value usually refers to E[X], while LOTUS tells you how to compute E[g(X)] for any function g. So LOTUS is the general rule, and ordinary expected value is one special case.

Why do I use LOTUS instead of finding the distribution of g(X)?

Because finding the transformed distribution can be slower and more error-prone. LOTUS lets you work straight from the PDF you already have, which is especially useful for nonlinear functions. If the assignment asks for an average or variance, LOTUS is often the cleanest route.