Random Experiments, Outcomes and Sample Space

A random experiment is a procedure with an uncertain result, an outcome is one possible result and the sample space contains all possible outcomes under the model. In brief: rolling one ordinary six-sided die has sample space {1, 2, 3, 4, 5, 6}; rolling an even number is an event containing outcomes 2, 4 and 6.

Probability calculations begin with a clear description of what is observed and which possibilities are included. A poorly chosen sample space can make even simple counting misleading. This guide explains outcomes, events, ordered results and systematic listing, with particular attention to when outcomes can reasonably be treated as equally likely.

What is a random experiment?

A random experiment is a specified process whose exact outcome is not known in advance. Examples include tossing a coin, drawing an item from a mixed collection or observing whether a selected component meets a condition. The procedure and the recorded result must be defined clearly enough to identify the possible outcomes.

Random does not mean that every imaginable result is possible or that all possible results have equal probabilities. A die roll has a restricted set of face values, while an unevenly weighted die may favour some faces. Uncertainty and equal likelihood are separate properties of the model.

The same physical action can support different experiments depending on what is recorded. For a die roll, one experiment may record the exact face, while another records only odd or even. Both are valid descriptions, but they have different sample spaces and retain different amounts of information.

What is an outcome?

An outcome is one complete result at the level of detail chosen for the experiment. For a single die, an outcome may be 4. For two ordered coin tosses, an outcome may be HT, meaning heads on the first toss and tails on the second.

An outcome should contain enough information to answer the questions being studied. If you record only the total on two dice, you cannot later determine whether the first die showed 6. A more detailed outcome such as (6, 2) retains both individual results and can also be used to calculate their sum.

Outcome labels need not be numbers. They can be categories, ordered pairs, sequences or measured values. The important requirement is that each possible experiment result fits the sample-space description consistently, without ambiguity about whether two labels refer to the same result or distinct results.

What is a sample space?

The sample space, often denoted S or Ω, is the set of all possible outcomes under the stated model. For one coin toss recorded as heads or tails, S = {H, T}. For one ordinary die roll, S = {1, 2, 3, 4, 5, 6}.

A finite sample space can be listed explicitly when it is small. Larger finite spaces may be described by a rule, such as all ordered pairs (a, b) where a and b each belong to {1, 2, 3, 4, 5, 6}. This defines the 36 outcomes for two distinguishable dice without writing every pair in a paragraph.

Some sample spaces are countably infinite or continuous. The number of tosses until the first head can be 1, 2, 3 and so on, while a modelled waiting time may take any nonnegative real value. Simple equally-likely counting formulas apply to finite spaces under additional assumptions, not automatically to every type of sample space.

How is an event different from an outcome?

An event is a subset of the sample space: it collects all outcomes satisfying a specified condition. For one die, the event A of an even result is {2, 4, 6}. The experiment produces one outcome, and the event occurs if that outcome belongs to its subset.

A single-outcome event contains just one outcome, such as {4}. The certain event is the entire sample space, and the impossible event is the empty set. These cases connect set language with probability: the whole space has probability 1 and the empty event has probability 0.

Several events can occur on the same trial. If a die shows 6, the events even and greater than 4 both occur. Events therefore need not be mutually exclusive. Distinguishing event membership from individual outcomes prepares the way for intersections, unions and conditional probabilities.

How do you list a sample space systematically?

Use a tree, table or ordered sequence to avoid missing or repeating outcomes. For two coin tosses, list the first result H or T and then each possible second result. The complete ordered sample space is {HH, HT, TH, TT}, containing four outcomes.

Order matters because the first and second toss are distinct positions. HT and TH both contain one head, but they are different sequences. If the experiment records only the number of heads, the reduced sample space is {0, 1, 2}; its outcomes do not inherit equal probabilities merely because there are three labels.

When an experiment has m choices at one stage and n choices at the next for every first-stage result, the number of ordered outcomes is mn. Two six-sided dice therefore have 6 × 6 = 36 ordered pairs. If the number of choices depends on earlier results, count each branch according to its actual possibilities.

1. Specify what is recorded and whether order matters.
2. List first-stage possibilities.
3. Attach every permitted later result to each earlier branch.
4. Check that each complete outcome appears once.
5. Group outcomes into events only after the full space is clear.

When are sample-space outcomes equally likely?

Equal likelihood is a modelling assumption justified by symmetry or the experiment’s design, not by the act of listing outcomes. For two independent fair coin tosses, HH, HT, TH and TT each have probability 1/4. This permits event probabilities to be found by counting favourable sequences.

The reduced outcomes zero heads, one head and two heads have probabilities 1/4, 1/2 and 1/4. One head corresponds to two underlying sequences, while each other count corresponds to one. Treating the three head-count outcomes as equally likely would therefore give an incorrect model.

The same issue occurs with two-die sums. Possible sums run from 2 through 12, but sum 7 can occur in six ordered ways while sum 2 occurs only as (1, 1). Choose a detailed equally likely sample space when using a counting formula, or assign the correct unequal probabilities to the reduced outcomes.

How do replacement and order affect a sample space?

Drawing with replacement returns the selected item before the next draw, so the same item can appear again. Drawing without replacement removes it, changing the later possibilities. From three distinct items A, B and C, two ordered draws with replacement give 3 × 3 = 9 outcomes, including AA, BB and CC.

Without replacement, two ordered draws give 3 × 2 = 6 outcomes: AB, AC, BA, BC, CA and CB. If order is ignored, these reduce to the three pairs {A, B}, {A, C} and {B, C}. The experiment’s recording rule determines which representation is appropriate.

These distinctions matter before any probability is calculated. A replacement assumption affects both the available outcomes and often their probabilities. An order assumption affects whether sequences are distinct or grouped together. Writing those choices explicitly prevents a counting method for one experiment from being applied to another.

Why is sample-space modelling important, and what errors are common?

The background idea is to describe uncertainty with a complete set of possibilities and a rule assigning probabilities to them. The sample space supplies the possibilities, while the probability model supplies their weights. Both are needed for a numerical probability; a list alone does not determine the weights.

Common errors include omitting outcomes, counting the same outcome twice, ignoring order when it matters and assuming equally likely labels after grouping unequal numbers of detailed outcomes. A systematic table and a clear statement of the recorded result resolve many of these problems before arithmetic begins.

Finally, distinguish an impossible outcome under the model from an unobserved outcome in a small dataset. Something may not have occurred yet while still being possible. A sound probability calculation begins with the experiment’s structure and assumptions, then uses observed data or symmetry appropriately to assign chances within that structure.