What is experimental probability?
Experimental probability is the probability you measure by actually doing something and counting what happens. Flip a coin 50 times, get 27 heads, and the experimental probability of heads is 27 out of 50, or 0.54. It is also called empirical probability or relative frequency, and it is how people estimate odds that cannot be worked out on paper, from the chance of rain tomorrow to how often a light bulb fails in its first year.
The experimental probability formula
Each repetition of the experiment is called a trial. The answer can be written as a fraction, a decimal or a percentage. For example, roll a die 60 times and count 13 sixes. The experimental probability of rolling a six is 13/60, which is about 0.217, or 21.7%. The theoretical probability is 1/6, or 16.7%, so this run came out a little high.
Theoretical vs experimental probability
Theoretical probability is what should happen, worked out before you do anything by counting equally likely outcomes. Experimental probability is what did happen. The two answer the same question from opposite ends.
| Theoretical | Experimental | |
|---|---|---|
| Based on | Reasoning about equally likely outcomes | Results of real trials |
| Formula | Favourable outcomes ÷ possible outcomes | Times the event happened ÷ number of trials |
| Coin landing heads | 1/2 = 50% | 27 heads in 50 flips = 54% |
| Changes if you repeat it? | No | Yes, every new trial can shift it |
Neither number is wrong. Theoretical probability assumes a perfect coin or die. Experimental probability makes no assumptions, which is why it can catch a die that is not fair. Casinos test their dice exactly this way.
Why the two numbers rarely match
Small samples swing a long way. Flip a coin 10 times and you will get exactly 5 heads only about one time in four. Seven or more heads, a 70% experimental probability, happens about one time in six. None of this means the coin is biased. It is what randomness looks like at small numbers.
As the number of trials grows, the experimental probability settles down. This is the law of large numbers. After 1,000 flips of a fair coin, the share of heads will usually sit within about three percentage points of 50%. Try it above: run 10 trials a few times and watch the Actual column jump about, then add 1,000 and watch it steady. The coin also has no memory. A run of tails does not make heads more likely next time, and believing it does is called the gambler's fallacy.
Spinner experiments
The spinner in the simulator is deliberately uneven. Red covers half the circle, blue a quarter, and green and yellow an eighth each. That means students have to work out the theoretical probabilities from the angles rather than by counting outcomes: red is 180° out of 360°, or 1/2, while green is 45° out of 360°, or 1/8. A good warm up question is how many greens to expect in 80 spins. The answer is 10, and the simulator will show how far a real run lands from it.
A 15 minute classroom activity
- Pick an experiment and ask students to write down the theoretical probability of each outcome.
- Run 10 trials on the board and record the results. Ask whether the results prove the coin or die is unfair.
- Add 100 trials, then 1,000, and compare the Actual and Theory columns each time.
- Discuss why the difference shrank, and why it never quite reached zero.
- Finish with a challenge: what result would convince you that a die was loaded?
The Print worksheet button produces a one page sheet with a prediction table, a results table and four questions, followed by an answer key for the teacher. There are no accounts, results reset when the page reloads, and nothing about your students is stored or sent anywhere. For exact odds of a given number of heads or a streak, use the coin flip probability calculator.