What is an indicator random variable?

An indicator random variable converts an event into a numerical object by assigning the value 1 to outcomes in the event and 0 to all other outcomes. It's a random variable that takes only values 0 or 1, representing membership in an event numerically. Denoted I_A for event A, it equals 1 if the outcome is in A and 0 otherwise.

What is the expected value of an indicator random variable?

The expected value of an indicator random variable equals the probability of its event: E[I_A] = P(A). This holds because I_A takes value 1 exactly on outcomes in A and 0 otherwise, so expectation counts how often the event occurs in probability terms.

How are indicator variables used for counting problems?

Counting problems can be expressed as sums of indicator random variables. Each object gets an indicator that equals 1 if a condition is met and 0 otherwise. The total count is the sum of indicators. Using linearity of expectation, this converts counting into expectation calculations without enumerating all outcomes.

What's the difference between an indicator variable and a probability?

An indicator is a random variable whose value depends on the outcome, not a probability value. It takes values 0 or 1 based on whether an event occurs. Its expected value equals the probability of the event (E[I_A] = P(A)), but the indicator itself is not a probability or probability distribution.

Indicators in Probability

Q: Do indicator random variables need to be independent?

No. Indicator random variables may be dependent, and this doesn't affect the validity of linearity of expectation. Expectations of sums of indicators can still be computed term by term without independence assumptions. However, dependence becomes important when computing higher-order quantities like variance.

Key Terms

From Events to Random Variables

Definition of an Indicator Random Variable

Basic Properties of Indicator Random Variables

Expectation of an Indicator Random Variable

Linearity of Expectation and Indicators

Counting with Indicator Random Variables

Indicators and Dependence

Indicators in Probability Models

What Indicators Are Not

Summary

Indicators at a Glance

Why Indicators Exist

Many probability problems ask whether a specific event occurs or does not occur.
Indicator random variables encode this yes–no information numerically.

By representing events as random variables that take values

0

1

, indicators allow probability questions to be handled using expectation and algebraic operations.
This makes them a structural tool for counting and for applying linearity of expectation.

Key Terms

Random Variable— an indicator is a binary-valued random variable

Bernoulli Distribution— an indicator follows a Bernoulli distribution

Expected Value—

E[\mathbf{1}_A] = P(A)

Event— the event

A

that the indicator tracks

See All Probability Definitions →

From Events to Random Variables

In probability theory, events are sets of outcomes.
Sets cannot be added, averaged, or combined algebraically.

Random variables assign numerical values to outcomes.
An indicator random variable assigns the value

1

to outcomes in a given event and

0

to all other outcomes.

This construction converts an event into a numerical object defined on the same sample space.

Definition of an Indicator Random Variable

Let

A

be an event in a probability experiment.
The indicator random variable of

A

, denoted by

I_A

, is defined by

I_A(\omega)=\begin{cases}1, & \text{if } \omega \in A \\ 0, & \text{if } \omega \notin A\end{cases}

The indicator

I_A

is a random variable defined on the same sample space as the experiment.
It represents membership in the event

A

numerically.

An indicator is not a probability value; it is a random variable whose value depends on the outcome.

Basic Properties of Indicator Random Variables

Indicator random variable have a simple structure.

• They take only two values:

0

1

• They are completely determined by the event they represent
• They are defined on the same sample space as the underlying experiment

Simple relationships follow directly from the definition:

• The indicator of the complement event equals

1 - I_A

• Indicators reflect set operations at the level of outcomes

These properties allow indicators to be combined and manipulated algebraically without introducing new probability assumptions.

Expectation of an Indicator Random Variable

The expected value of an indicator random variable equals the probability of its event.

For an event

A

with indicator

I_A

E[I_A]=P(A)

This holds because

I_A

takes the value

1

exactly on outcomes in

A

and

0

otherwise.
Expectation therefore counts how often the event occurs in probability terms.

This identity connects events directly to expectation without introducing a full probability distribution.

Linearity of Expectation and Indicators

Expectation is linear.

For indicator random variables, this means:

• The expectation of a sum equals the sum of expectations
• No independence assumptions are required

As a result, counting problems can be solved by expressing the quantity of interest as a sum of indicator variables and taking expectations term by term.

This technique avoids direct probability calculations and remains valid even when the indicators are dependent.

Counting with Indicator Random Variables

Many counting problems can be expressed as sums of indicator random variables.

Each object or outcome is associated with an indicator that takes the value

1

if a specified condition is met and

0

otherwise.
The total count is obtained by summing these indicators.

This approach converts counting questions into expectation calculations and allows results to be obtained without enumerating all outcomes or computing complex probabilities.

Indicators and Dependence

Indicator random variables may be dependent.

Dependence does not affect the validity of linearity of expectation, so expectations of sums of indicators can still be computed term by term.
However, dependence becomes important when higher-order quantities such as variance are considered.

This distinction explains why indicators are effective for expectation-based counting but require additional care in variance calculations.

Indicators in Probability Models

Indicator random variables are used across many probability models.

They commonly appear in:

• Bernoulli trials, where each trial contributes a

0

1

• Binomial models, where the total number of successes is a sum of indicators
• Occupancy and matching problems, where indicators track whether a condition is satisfied
• Random structures, where indicators isolate local events

Indicators function as a structural tool rather than as a probability model of their own.

What Indicators Are Not

Indicator random variables are often misunderstood.

• They are not probabilities
• They are not probability distributions
• They are not limited to Bernoulli experiments

An indicator is a random variable that represents an event numerically.
Its value depends on the outcome, not on the probability of the event itself.

Summary

Indicator random variables convert events into numerical random variables that take values

0

1

.

Their expected value equals the probability of the corresponding event, and sums of indicators allow counting problems to be handled through linearity of expectation without requiring independence or full probability distributions.

Indicators function as a structural tool in probability theory, linking events, expectation, and counting within finite and discrete models.

Indicators at a Glance

The table below collects the anatomy of indicator random variables — what they are, what they aren't, the key identity that ties them to events, the algebra they support, the linearity-of-expectation property that makes them powerful, the counting technique they enable, how dependence affects them, and where they show up across probability models — into a single reference card.

Aspect	Statement	Note / example
What it is	a random variable assigning 1 to outcomes in A and 0 to all others	I_A : Ω → {0, 1}; follows a Bernoulli distribution with parameter P(A)
What it is NOT	not a probability value, not a distribution, not exclusive to Bernoulli trials	its value depends on the outcome, not on P(A)
Key identity	E[I_A] = P(A)	the bridge between events and expectation
Complement	I_Aᶜ = 1 − I_A	indicator algebra mirrors set operations
Linearity	E[∑ I_Aᵢ] = ∑ P(A_i)	no independence assumption required
Counting technique	express the count N = ∑ I_Aᵢ, then take expectation termwise	converts counting into a sum of probabilities, avoiding enumeration
Dependence	allowed and harmless for expectation	becomes essential when computing variance
Typical uses	Bernoulli / binomial models, matching, occupancy, random structures	a structural tool across probability models, not a model of its own