• Foundation
  • About 12 min
  • 5 practice questions

Theoretical and experimental probability

Tell reasoning-based probability from probability estimated by trials, and use expected counts to predict how often an event should occur.

If something looks wrong, please tell us.

So far we have worked out probabilities by reasoning about equally likely outcomes. In real life you often cannot do that, and the only way to learn a probability is to try the experiment many times and count. This lesson explains the difference between theoretical and experimental probability, and how to predict how often something should happen.

Two kinds of probability

KindHow it is foundExample
TheoreticalBy reasoning about equally likely outcomesA fair coin shows heads with probability 1/2
ExperimentalBy repeating the experiment and countingHeads appeared in 27 of 50 tosses: 27/50 = 0.54

The experimental probability, also called the relative frequency, is the number of times the event happened divided by the number of trials. It need not equal the theoretical value, especially for a small number of trials. As the number of trials grows, the relative frequency tends to settle near the theoretical value, though it is never guaranteed to equal it.

Two kinds of probability

  • Theoretical
    • Found by reasoning about equally likely outcomes
    • Fair coin: heads 1/2 = 0.5
  • Experimental
    • Times it happened ÷ number of trials
    • Heads in 27 of 50 tosses: 0.54
    • Tends to settle near the theoretical value as trials grow

Worked example 1: experimental probability

Imagine a coin is tossed 50 times and shows heads 27 times. The experimental probability of heads is 27 ÷ 50 = 0.54. The theoretical probability is 0.5. The difference is small, which is normal for 50 tosses. If you tossed the coin 5,000 times, you would expect the relative frequency to be much closer to 0.5.

Expected number of occurrences

If an event has probability P and the experiment is repeated n times, the expected number of occurrences is n × P. This is an average over many sets of trials, not a promise about any one set. A die rolled 120 times is expected to show a six 120 × 1/6 = 20 times. Seeing 17 or 23 sixes would be quite normal; getting exactly 20 is not guaranteed.

Expected number of occurrences = number of trials × probability. Experimental probability = times it happened ÷ number of trials.

— Rule to remember

Worked example 2: expected counts

A fair die is rolled 90 times. How many even numbers are expected? P(even) = 3/6 = 1/2, so the expected number is 90 × 1/2 = 45. How many numbers greater than 4 are expected? The favourable outcomes are 5 and 6, so P = 1/3, and the expected number is 90 × 1/3 = 30.

Worked example 3: estimating what is in a bag

A bag holds 50 balls, some red and the rest blue. A ball is drawn, its colour noted and the ball put back. This is done 100 times, and red appears 30 times. The experimental probability of red is 30/100 = 0.3. So an estimate of the number of red balls is 0.3 × 50 = 15. This is only an estimate. If the real number were 16 or 14, you would not be surprised to see 30 reds in 100 draws.

Estimating the red balls in a bag of 50

  1. Repeat the draw 100 draws, the ball put back each time.
  2. Experimental probability Red 30 times: 30/100 = 0.3.
  3. Estimate 0.3 × 50 = 15 red balls (only an estimate).

The gambler's mistake

Suppose a fair coin shows heads five times in a row. What is the probability that the next toss is heads? It is still 1/2. The coin has no memory, and the tosses are independent. The idea that tails is 'due' is a common error. Over a long run, the proportion of heads gets close to a half, not because the coin corrects itself, but because a few extra heads in the early tosses make little difference when thousands of tosses are added.

Key takeaways

  • Theoretical probability comes from reasoning; experimental probability comes from trials.
  • Expected occurrences = n × P.
  • More trials usually bring the relative frequency closer to the theoretical value.
  • Past results do not change the probability of the next independent trial.

Check your understanding

  1. 1. A coin is tossed 50 times and shows heads 27 times. What is the experimental probability of heads?

    1. 0.50
    2. 0.54
    3. 0.46
    4. 0.27
    Show answer

    Answer: B. 27 ÷ 50 = 0.54.

  2. 2. A fair die is rolled 120 times. About how many sixes are expected?

    1. 12
    2. 24
    3. 30
    4. 20
    Show answer

    Answer: D. The expected number is 120 × 1/6 = 20.

  3. 3. A fair die is rolled 90 times. How many even numbers are expected?

    1. 45
    2. 30
    3. 15
    4. 60
    Show answer

    Answer: A. P(even) = 1/2, so the expected number is 90 × 1/2 = 45.

  4. 4. A bag holds 50 balls. Balls are drawn 100 times with replacement and 30 are red. What is the best estimate of the number of red balls in the bag?

    1. 10
    2. 20
    3. 15
    4. 30
    Show answer

    Answer: C. The relative frequency is 0.3, so the estimate is 0.3 × 50 = 15.

  5. 5. A fair coin has shown heads five times in a row. What is the probability that the next toss is heads?

    1. 1/2
    2. 1/32
    3. 1/6
    4. Less than 1/2
    Show answer

    Answer: A. The tosses are independent, so the probability of heads is still 1/2.

Privacy · Terms · Refunds · Delivery